the impact of cash and food transfers: evidence from a randomized intervention in niger

14
The impact of cash and food transfers: Evidence from a randomized intervention in Niger John Hoddinott – International Food Policy Research Institute Susanna Sandström – Abo Akademi University Joanna Upton – Cornell University (Presenter) AAEA and CAES Joint Annual Meeting Washington, D.C. August 6, 2013

Upload: international-food-policy-research-institute-ifpri

Post on 14-Jul-2015

964 views

Category:

Education


3 download

TRANSCRIPT

Page 1: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Theimpactofcashandfoodtransfers:EvidencefromarandomizedinterventioninNigerJohn Hoddinott – International Food Policy Research InstituteSusanna Sandström – Abo Akademi UniversityJoanna Upton – Cornell University (Presenter)

AAEA and CAES Joint Annual MeetingWashington, D.C.August 6, 2013

Page 2: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Literature&ResearchQuestions

• Food versus Cash transfers• Timeliness advantages, and in some cases cost (Gentilini 2007; Lentz et al. forthcoming). 

• Key advantage of cash is that it provides recipients with choice• Several contributions• Randomized design• Important setting for understanding impacts of cash transfers

• Geography (land‐locked), poverty, food insecurity• Functioning food markets• Government and agency movement toward cash transfers

• Food security analysis

Page 3: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Context• Niger, and the intervention region• Fifth poorest in per capita GNI; 186/187 on the HDI; 93% suffering from deprivation 

• Food insecure (availability, access, and use); severe food crises in parts of country in 2005‐2006, 2010, and again in 2012

• Zinder; surplus production region, yet often hardest hit by food crises

• The Project (2011)• Large‐scale cash/food pilot implemented by the World Food Programme

• Cash/food for work (April‐June), followed by unconditional transfers (July‐Sept)

• 126 villages in 12 departments of Mirriah, Zinder• Food basket, and cash transfer (of equivalent value)

Page 4: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Context

Estimated food security conditions, 2nd Quarter (April‐June) 2011 Source: FEWS.net

Page 5: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

ResearchDesign• Randomization at worksite level (52 worksites)

• Two survey rounds:• Following public works, all households in all evaluated villages (5,668 households)

• Follow‐up with unconditional transfer recipients (2,268 households)

Mirriah District           (52 worksites)

Agricultural Zone (29 sites)

RANDOMIZE

Cash for work (15 sites, 1747 HHs)

Cash transfers (15 sites, 686 HHs)

Food for work (14 sites, 1658 HHs)

Food transfers (14 sites, 635 HHs)

Agro‐Pastoral Zone (23 sites)

RANDOMIZE

Cash for work(12 sites, 1202 HHs)

Cash transfers (12 sites, 493 HHs)

Food for work (11 sites, 1061 HHs)

Food Transfers (11 sites, 395 HHs)

Page 6: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Data&Balance

• Survey coverage• ALL households: Composition, ethnicity/status, credits prior to transfer period, lodging, assets, livestock, production, public works participation

• Additionally, for subset and follow‐up: Credits and transfers during the intervention, food consumption and diversity, coping strategies

• Inter‐seasonal component • Balance and Quality• At worksite level, balances on all observables• Followed up with 2,209 of 2,268 chosen for unconditional transfers (attrition of 2.6%)

you had them booked for a tentative PS by Shahid

Page 7: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

OutcomeVariableDefinitions

• Foods and food groups consumed• How many occasions in the past 7 days

• Household Dietary Diversity Index (HDDI)• Sum of different foods consumed (1 to 25)

• Household Dietary Diversity Score (HDDS)• Sum of different food groups consumed (1 to 11)

• Food Consumption Score (FCS)• Weighted sum of food groups consumed, based on dietary quality• Categories: Poor (≤ 21), Borderline (> 21, ≤ 35), Acceptable (> 35)

• Coping Strategies Index (CSI)• Index based on reliance on a diverse set of coping strategies

• Food and Non‐Food Expenditures

Page 8: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Method

• A single‐difference estimator of the form:

,

, 1

• Lack of baseline => cannot estimate a dif‐in‐dif; however, dif‐in‐dif only preferable when autocorrelation of outcomes is high (which it is not in this case)

Page 9: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Results:FoodDiversity

Food Security Outcomes (Table 4) July October

Dietary Diversity Index (DDI) 0.356* 0.544**

(0.207) (0.229)

Food Consumption Score (FCS) 3.923*** 4.647***

(1.424) (1.139)

“Acceptable” FCS (WFP cut‐off) 0.109** 0.121***

(0.043) (0.041)

Number of Households 2256 2187Standard errors (clustered at worksite level) in parentheses*, **, and *** indicate significance at the 10%, 5%, and 1% level, respectivelyControls included (but not shown):  age, sex, education, and ethnicity of household head; household size; asset score; whether in pastoral zone; village access to market, health clinic, and mobile phone coverage; distance to the main road; livestock prices; change in millet price during period; millet price at end of period; commune‐level fixed effects.

Page 10: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Results:CopingStrategies

CSI and Selected Coping Strategies (Table 7) July October

Coping Strategies Index ‐3.708* ‐3.168***

(1.916) (0.411)

Relied on less‐preferred foods ‐0.039* 0.024

(0.022) (0.020)

Borrowed food from relatives or neighbors ‐0.082*** ‐0.022

(0.024) (0.021)

Had to cancel debt repayments ‐0.038** 0.057***

(0.017) (0.009)

Reduced number of meals per day ‐0.025 ‐0.036**

(0.024) (0.015)

Number of households 2256 2187Notes: see Table 4 (previous slide)

Page 11: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Results:FoodConsumptionandExpendituresFood Groups (Table 5) Were items consumed Number of days consumed

July October July October

Cereals ‐‐ ‐‐ 0.093* 0.109***

(0.051) (0.035)

Pulses 0.064** 0.021 0.638** 0.820***

(0.032) (0.013) (0.314) (0.168)

Oils 0.106*** 0.042** 0.959*** 1.010***

(0.033) (0.017) (0.258) (0.186)

Food Purchases (Table 6) Made purchase Expenditure (CFA)

July October July October

Bulk Grain purchases ‐0.273*** ‐0.400*** ‐14289*** ‐25015***

(0.020) (0.034) (1570) (432)Consumption/purchase estimated using a Probit model; number of days using a Poisson; expenditures using a Tobit. Results reported are marginal effects.Additional notes in Table 4 (above).

Page 12: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Results:Non‐FoodExpenditures

Non‐Food Purchases (Table 6) Made purchase Expenditure (CFA monthly)

July October July October

Total non‐food expenditure ‐‐ ‐‐ 1874.7*** ‐592.0

(502) (1010.7)

Construction, repair, housing ‐0.034* 0.002 ‐2870.8* 495.2

(0.021) (0.016) (1686.3) (403.9)

Wages, animal care, seeds ‐0.105*** ‐0.090*** ‐1779** ‐5819**

(0.035) (0.029) (816.2) (2604)Purchase estimated using a Probit model; expenditures using a Tobit. Results reported are marginal effects.Additional notes in Table 4 (above).

Page 13: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Summary

• Households who received the food basket experienced larger positive impacts on measures of food security, including dietary diversity and coping strategies

• Cash recipients bought bulk grains• Also spent more on household repairs in July, and more on farm inputs/livestock care in both periods

• Pertinence of context• Response to (predictable) future price variation• Extreme poverty contributing to the lack of typical ‘diversity’ in diet

• Much remaining to explore…

Page 14: The impact of cash and food transfers: Evidence from a randomized intervention in Niger

Joanna [email protected]

Corresponding Author:John Hoddinott2033 K Street N.W., Washington, D.C. [email protected]