The impacts of local procurement for a school

advertisement
Local Food for Local Schools:
The impacts of local procurement for a school feeding
program in Burkina Faso
Joanna B. Upton, Erin C. Lentz, Christopher B. Barrett, and Teevrat Garg
Cornell University
May 23, 2012 seminar at
University of California at Riverside
Background:
The Policy Issues
• “Food Assistance” rather than “Food Aid”
– Rapid expansion in cash, vouchers and, especially,
local and regional procurement (LRP) of food aid:
• 11% of global food aid flows in 1999, then jumps to
39% by 2008 and then 67% in 2010 following the
2008 Farm Bill LRP pilot and launch of USAID EFSP
– New (draft) int’l treaty: Food Aid Assistance Convention
• The promise:
– More timely and rapid delivery
– “Structured demand” can perhaps stimulate ag dev’t
– More “culturally appropriate” foods?
• The risks:
– Market disruption from demand interventions
– Food safety/quality
Background:
Burkina Faso and LRP
• Burkina Faso: an ideal candidate for LRP?
• Frequent food aid recipient
• Land-locked, so expensive/slow to reach
• Agrarian, with surplus zones
• Evaluation opportunity
– Natural experiment with partial replacement of
food aid from US under a Multi-Year Assistance
Program (MYAP) with USDA-funded LP pilot for
ongoing school feeding programs runs by the
same agency (Catholic Relief Services).
– Gnagna and Namentenga Provinces
• 364 schools serving 58,227 primary school students
• Surplus cowpea production; net food buyers
Background:
Burkina Faso and LRP
Methodology and Results: LEAP and MYAP
Background:
Burkina Faso and LRP
• Procurement:
– Vitamin-A Fortified Vegetable Oil
• Competitive Tender—Ouagadougou and Bobo Dioulasso
• 72.1 MTs purchased
– Millet
• Semi-competitive process through large grain cooperatives
• Four unions, providing between 54 and 319 MTs each
• 628.1 MTs purchased
– Cowpeas
• Semi-competitive process through small farmers’
associations
• 22 associations, providing between 2.8 and 21.4 MTs each
• 143.5 MTs purchased
Background:
Burkina Faso and LRP
Background:
Burkina Faso and LRP
Methodology and Results:
Overview
Key Logistical Concerns:
• Timeliness: does LRP really deliver faster?
• Cost: is LRP really cheaper?
• Quality: is LRP equally safe/reliable?
Macro-/Meso-level (Markets) Concerns:
• Market Prices: does LRP disrupt local markets?
Micro-level (Recipients and Communities):
• Recipient Preferences: do they prefer local foods?
• Producer Impacts: do suppliers benefit?
Methodology and Results:
Timeliness
• Direct
comparison
observations
each) versus
Evidence
from(three
Burkina:
Timeliness
multiple comparison
Average Time (in weeks from procurement startdate)
KEY DATES
Call Forward /
Identification of suppliers
MYAP: Shipment from U.S.
LEAP
MYAP
0
0
(Dec 15 - Jan 28)
(June 8, 2010)
(63 obs)
0
27
32
MYAP: Arrival in Lomé
Availability for Quality Testing
9
LEAP: Contract for oil
14
LEAP: Contracts for millet
11
Delivery
Diverse U.S. Programs
33 to 45
15.3
47
(March 25-April 7)
(April 14 - June 3)
34.96
– Average difference MYAP-LEAP: 31.7 weeks
– Average difference, U.S. Average – LEAP: 20 weeks***
Methodology and Results:
Cost
• Detail of Costs, and Costs per child
LEAP
MYAP
Evidence
from
Burkina:
Cost
Commodity Cost (& Vouchers)
Millet / bulgur wheat
$ 339.27
Cowpeas / lentils
$ 546.45
Vegetable oil*
$ 2,065.00
Ocean Freight and Transport to Ouagadougou
Bulgur wheat
N/A
Lentils
N/A
Vegetable oil
N/A
Transport to Schools**
Millet / bulgur wheat
$
80.78
Cowpeas / lentils
$
20.02
Vegetable oil
$
47.04
Quality Testing
Millet / bulgur wheat
$
18.55
Cowpeas / lentils
$
18.55
Vegetable oil*
N/A
TOTALS
Millet / bulgur wheat
$ 438.60
Cowpeas / lentils
$ 585.02
Vegetable oil
$ 2,112.04
Cost per Child (for three months)***
Combined Ration
$
9.48
$ 386.73
$ 585.85
$ 1,322.12
$
$
$
442.87
442.87
477.41
$
$
$
50.00
50.00
50.00
$
$
$
15.60
16.37
8.10
$ 895.20
$ 1,095.09
$ 1,857.64
$
15.41
* For LEAP, the Veg Oil costs
includes quality certification
** Estimate for MYAP
*** Based on the ration of 180
grams grains, 45 grams legumes, 25
grams oil, per child for 20 rations
per month for three months
LEAP cost 38% less
per child than MYAP
Methodology and Results:
Cost
MYAP Cost
% Savings
(per MT)
with LEAP
Commodity Cost
Millet / bulgur wheat
$
386.73
12%
Cowpeas / lentils
$
585.85
13%
Vegetable oil
$ 1,322.12
-56%
Millet / bulgur wheat
$
492.87
84%
Cowpeas / lentils
$
492.87
96%
Vegetable oil
$
527.41
91%
Transportation
Other Comparisons, for the rations delivered
(180 grams of millet/bulgur, 45 grams of cowpeas/lentils,
25 grams of vegetable oil):
MYAP
Per Child for 3-months' rations (60 days)
Per kilocalorie
Per gram of protein
Per gram of fat
LEAP
$
15.41
$
9.48
$
0.04
$
0.02
$
1.61
$
0.95
$
0.62
$
0.36
Methodology and Results:
Quality
Quality Criteria
Moisture Content
Damaged Grains
Broken Grains
Foreign Matter, organic
Foreign Matter, inert
Live Insects (per 100g)
Aflatoxins (ppb)
Millet
Cowpeas
USDA Contract Result (Avg) USDA Contract
GoB
5.8%
≤ 11%
≤ 13%
≤ 12%
----≤ 1%
--0%
≤ 1%
≤ 1%
--≤ 1%
≤ 2%
≤ .75%
0%
--≤ 1%
≤ .25%
--0
0
0
0.25
0
----≤ 10
• Questions and Challenges:
Result (Avg)
8.7%
17.7%
0.0%
0.4%
0.2%
0.095
0.0396
– Are quality products available? Is adequate testing available? Are
standards (a) locally present, (b) enforceable?
– Monitoring of stocks, especially with small-scale suppliers
– Replacing spoiled foods
Methodology and Results:
Market Impacts
Food aid and market disruption:
• Longstanding concerns about supply-side effects
of food aid distribution
• Are there demand side effects to worry about?
Use secondary data on market prices and a
range of control variables to study the
relationship between LRP and
- market prices
- market price volatility
Methodology and Results:
Market Impacts
Estimation Strategy
Price Level Impacts:
Price Volatility Impacts:
X = Full Set of Controls (incl. WFP LRP, rainfall, precip
shocks, seasonal dummies, world market price, domestic
CPI, transport costs, etc.)
c: commodity
i: region
s: price type (transmission channel)
t: time period
15
Methodology and Results:
Market Impacts
Millet Estimation Results
Price level impacts
(Expected % change)
Pooled across markets
Point Est.
(SE)
1.386
(3.814)
Procurement markets
0.760
(7.088)
Non-Procurement markets
1.525
(4.620)
Price volatility impacts
(Expected std dev % change)
Pooled across markets
-1.185
(2.670)
Procurement markets
-1.990
(4.452)
Non-Procurement markets
-0.626
(3.337)
No discernible impact of millet procurement, in general
or in procurement (or non-procurement) markets.
16
Methodology and Results:
Recipient Preferences
Exploit natural experiment to test recipient
preferences
based
on food
aid sourcing:
Evidence from
Burkina:
Recipients
• Fielded among 120 schools each in MYAP/LEAP
• Survey administered to the head school cook.
– Ask Likert scale questions (1=low/very unsatisfied to
5=high/very satisfied) about a range of consumption
and preparation attributes of the food.
– Ask in general/absolute terms and relative to last year,
when ALL schools were MYAP-supplied (diff-in-diff)
• Complemented by a school director survey
covering school characteristics, such as quality of
infrastructure, distance to markets, enrollment,
attendance, and composition of students.
Methodology and Results:
Recipient Preferences
Evidence from Burkina: Recipients
Methodology and Results:
Recipient Preferences
General Preferences (no comparison)
Criteria
Taste
Ration Size
Texture
Appearance
Cleanliness
Storability
Nutrition
General Satisfaction
Commodity Preferred
GRAIN (millet / bulgur wheat)
LEGUME (cowpeas / lentils)
Bivariate Test
Multivariate Test
Bivariate Test
Multivariate Test
(Mann-Whitney)
(Ordered Logit)
(Mann-Whitney) (Ordered Logit)
LEAP**
LEAP
LEAP
LEAP***
LEAP**
LEAP
LEAP***
LEAP**
LEAP***
LEAP*
LEAP***
LEAP***
LEAP***
LEAP**
LEAP***
LEAP***
MYAP***
MYAP**
LEAP
LEAP
LEAP***
LEAP***
LEAP**
LEAP
LEAP
MYAP***
LEAP**
MYAP***
LEAP
LEAP
LEAP
LEAP***
*, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively
Methodology and Results:
Recipient Preferences
Preparation Requirements (compared to prior year)
Criteria
Time
Effort
Cost
Fuel Use
Water Use
Oil Use
Commodity Preferred : Preparation
GRAIN (millet / bulgur wheat)
LEGUME (cowpeas / lentils)
Bivariate Test
Multivariate Test
Bivariate Test
Multivariate Test
(Mann-Whitney)
(Ordered Logit)
(Mann-Whitney) (Ordered Logit)
MYAP***
MYAP***
MYAP***
MYAP***
MYAP***
MYAP***
MYAP***
MYAP***
MYAP***
MYAP*
MYAP***
MYAP
MYAP***
MYAP***
MYAP***
MYAP***
MYAP***
MYAP***
MYAP***
MYAP
LEAP*
LEAP*
MYAP
MYAP
*, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively
Methodology and Results:
Producer Outcomes
Exploit natural experiment to test LRP
impacts
on from
smallholder
suppliers:
Evidence
Burkina:cowpea
Recipients
• Fielded among 160 members of LEAP-supplying
farmer associations, with 150 control group
members of MYAP zone farmer associations
• Standard farm survey questions were asked for
the project year (2010-2011) and prior year
• Use difference in differences estimation for
– farmers in LEAP associations (ITT) or
– direct participants who sold to LEAP (ATET)
Methodology and Results:
Producer Outcomes
Methodology and Results:
Producer Outcomes
• Transaction Costs / Profitability
Member of LEAP
Association
Direct Participant
Evidence from Burkina:
Producers
(ITT)
(ATET)
Random Effects, OLS or Ordered Logit
Travel and Time, project year relative to prior year
Number of trips taken to sell cowpeas
Reported difference from the prior year^
Total distance travelled to sell cowpeas (km)
Time travelling to sell cowpeas (min)
-0.4
(-24%)
FEWER***
-0.4
(-24%)
MORE
-11*
(-33%)
-32**
(-32%)
-14**
(-42%)
-29**
(-29%)
Results from the Randome Effects estimators are reported in the cases in which they were found to be unbiased, relative to the Fixed Effects
models. If found biased, the Fixed Effects estimates are reported. These latter include: the ATET of Price received for cowpeas in project
year, and both estimates for Individual transactions to sell cowpeas.
^ These outcomes were ordinal based on Likert-scale questions, and estimated using the ordered logit random effects model.
*, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively
– Approximately 30% less time and 40% less distance
traveled, on average
Methodology and Results:
Producer Outcomes
• Transaction Costs / Profitability
Evidence from Burkina:
Producers Direct Participant
Member of LEAP Association
(ITT)
(ATET)
Random Effects, OLS or Ordered Logit
Price Received
Price received for cowpeas in project year (CFA)
Reported difference from the prior year^
Profitability
Reported difference from the prior year^
Revenue
Revenue from cowpeas in project year (CFA)
Change project year relative to the prior year (CFA)
-41**
(+49%)
HIGHER
+18.9
(+53%)
HIGHER**
HIGHER
HIGHER
+16,668*
(+25%)
+1,076**
(+2%)
+31,091**
(+47%)
+8,541
(+13%)
Results from the Randome Effects estimators are reported in the cases in which they were found to be unbiased, relative to the Fixed Effects
models. If found biased, the Fixed Effects estimates are reported. These latter include: the ATET of Price received for cowpeas in project
year, and both estimates for Individual transactions to sell cowpeas.
^ These outcomes were ordinal based on Likert-scale questions, and estimated using the ordered logit random effects model.
*, **, and *** represent statistical significance at the 10%, 5%, and 1% levels, respectively
Methodology and Results:
Producer Outcomes
• Transaction Costs / Profitability
Burkina:
Producers
– 19%Evidence
higher pricefrom
received
on average
(relative to control
group) and 47% higher revenue
– Association members who sold to LEAP also self-report higher
price, as well as improved profitability project year relative to
prior year (on a Likert scale)
How, if agencies pay competitive prices (no market impacts)?
Cowpea Price Trends
Answer: LEAP allowed
400
350
300
250
200
150
100
50
0
Bogande (Producer)
Source: SO.NA.GE.S.S
Jun-11
Jul-11
Apr-11
May-11
.
Feb-11
Mar-11
Jan-11
Nov-10
Dec-10
Sep-10
Oct-10
Ouagadougou
(Consumer)
Aug-10
Weekly Prices, CFA per KG
Aug 2010-July 2011
farmers to delay sales
and reap the benefits
of seasonal market
price increases
Methodology and Results:
Producer Outcomes
• Behavioral Impacts (summary)
– Ordered Logit estimator, with controls
Member of LEAP Association
(ITT)
+++
--+++
Direct Participant
(ATET)
+
+++
Quality Knowledge
Storage
Small Investments
Traction Investments
++
The "+" or "-" symbols indicate a positive or negative coefficient; the number of signs
indicates the significance, +/-, ++/--, +++/--- for 10%, 5%, and 1%, respectively.
Summary: Trade-offs or Synergies?
Benefits/Synergies:
• Cost Savings all around
– 38% lower purchase cost for buyers (CRS)
– 40% less distance traveled, and 20% higher price for sellers
• Agencies save on transport
• Recipients eat what they know and like
• Much faster delivery
• Easier to enforce product quality standards
• Local benefits “on both ends”
– Benefits from both the purchase and the distribution
“It is a smart child who, when at market,
buys his ‘beignets’ from his own mom.”
- Mooré proverb
Trade-offs/risks:
• Buying small works, but…
– Can this be scaled up significantly?
– Response analysis capacity
– Maintaining quality
• Commodities people prefer, but…
– Preparation advantages?
– Nutritional advantages?
Overall: local food procurement
clearly a success by multiple
metrics in Burkina Faso. Merits
being made a permanent part of
the food assistance toolkit.
Thanks for your interest and comments
Download