th4_how accessibility to seeds affects the potential adoption of an improved rice based technology:
DESCRIPTION
3rd Africa Rice Congress Theme 4: Rice policy for food security through smallholder and agribusiness development Mini symposium 4: Evidence of impact and adoption of rice technologies Author: DibbaTRANSCRIPT
How accessibility to seeds affects the potential adoption of an improved rice based technology: The case of New Rice Varieties for Africa (NERICA) in
The Gambia Varieties for Africa (NERICA) in The Gambia
Lamin Dibba*, Manfred Zeller, Aliou Diagne and Thea Nielsen
1
Outline of the presentation
1. Introduction
2. Objective of the study
3. Methodology
4. Results and discussion
5. Conclusions
6. Acknowledgments
7. Reference
2
Introduction• The per capita consumption of rice is estimated at
177kg per annum (PSU, 2011)• Of the 195,811 metric tons of rice consumed in
2011, only 51,137 metric tons was produced locally (Agricultural census, 2012)
• Out of the 51,137 metric tons of rice produced locally in 2011, 23,302 metric tons were entirely attributed to NERICA cultivation (Agric census, 2012)
• Past studies that assess NERICA adoption in The Gambia control only exposure or awareness (Dibba et. al., 2012; Diagne et. al., 2012)
3
Objectives• Assess NERICA adoption by controlling for both
exposure and seed access• Provide estimates of actual and potential adoption
rates and their determinants of the NERICA varieties
• Determine the adoption gap that arises due to lack of access to adequate supply of NERICA seeds
4
Methodology: Sampling procedure and data
• Multi-stage stratified random sampling procedure to select villages and farmers across the six agricultral regions
• 5 NERICA seed dessimination and non NERICA seed dessination villages randomly selected from each region
• 10 rice farmers randomly selected in each village• Data collected included both agronomic and
socio-economic information
5
Methodology: Conceptual framework
• This study relies on the potential outcome framework to assess the effect of exposure and access to seeds on NERICA adoption
• Every farmer has two potential or counterfactual outcomes (Y1 and Y0 ) for each treatment (Rosenbaum and Rubin, 1983)
• The causal effect of each treatment (Y1 - Y0 ) • In the adoption context Y0 = 0 for any observational
unit whether treated or untreated• The adoption impact for farmer i is given by Yi1 and
the average impact is given by ATE = E(Y1 )
6
Methodology: Estimation of adoption rates
• The Conditional Independence (CI) assumption (Rosenbaum and Rubin,1983): • w(s) is independent of Y1 and Y0 conditional of X
• Potential adoption is independent from Zi conditional on Xi (Diagne and Demont, 2007)
• Exposure or access to seed is independent of Xi conditional on Zi
• Overlap for all covariates. Then ATE is semi-parametrically identified by equation 1
7
)1......(....................)(ˆ)(ˆ1ˆ
1
,
n
i i
ise
zp
xm
nETA
Results and DiscussionTable 1: Comparison of 2006 and 2010 survey results
8
Variable 2006 (N=600)
2010 (N=515)
Difference (T-test)
Exposure to NERICA 0.47 (0.02) 0.88 (0.01) 0.41 (0.02)***
Adoption within NERICA exposed sub-population
0.85 (0.03) 0.77 (0.03) -0.08 (0.03)***
NERICA sample adoption 0.40 (0.02) 0.66 (0.02) 0.26 (0.03)***
Practice of upland rice production
0.53 (0.02) 0.78 (0.02) 0.25 (0.03)***
Practice of lowland rice production
0.80 (0.02) 0.43 (0.02) - 0.36 (0.03)***
Farmer contact with NARI 0.5 (0.01) 0.21 (0.02) 0.16 (0.02)***
Farmer contact with DAS 0.31(0.02) 0.32 (0.02) 0.01(0.03)
Results and DiscussionTable 2: Actual adoption of NERICA varieties in 2010
Description Regions Total
WCR LRR CRS NBR CRN URR
Total number of farmers 89 85 89 92 78 82 515
Proportion of farmers exposed to
NERICAs in 2010 (%)
99 95 62 100 86 89 88
Proportion of exposed farmers who
had access to NERICA seeds in
2010 (%)
84 93 38 80 71 68 72
Proportion of farmers who adopted
at least one NERICA (%)
2008 54 69 20 67 31 56 50
2009 65 79 29 67 59 72 61
2010 76 88 35 72 62 65 66
9
Results and DiscussionATE exposure
model
ATE access to
seeds model
Adoption
gap due to
lack of
seeds
NERICA population adoption rate (ATE) 0.76 (0.29)*** 0.92(0.09)*** 16%
Adoption rate within the NERICA-exposed and
seed accessed subpopulation (ATE1)
0.76 (0.34)** 0.92(0.11)*** 16%
Adoption within the NERICA non exposed
and seed accessed subpopulation (ATE0)
0.73 (0.11)*** 0.89(0.05)*** 16%
Joint exposure and adoption (JEA) 0.66(0.28)*** 0.66(0.08)***
Adoption gap of NERICA (GAP) -0.10 (0.02)*** -0.26(0.01)***
Expected population selection bias when
using the within NERICA – exposed and seed
accessed sub-sample estimate (PSB)
0.01 (0.05) -0.01 (0.03)
10
Table 3: ATE semi-parametric estimation of potential adoption rates
Results and DiscussionTable 4: Factors affecting exposure, access to seeds and adoption
Coefficients of Exposure
Coefficient of Access to seeds
Coefficient of Adoption
Age -0.01 (0.007) -0.01* (0.005) -0.01 (0.006)
Years of experience in upland farming 0.01 (0.009) 0.02***(0.006) 0.017** (0.008)
Formal education 0.49 (0.507) 0.02 (0.242) 0.16 (0.312)
Household size 0.069** (0.030) 0.02 (0.016) -0.02 (0.019)
Off-farm labor -1.705*** (0.491) -0.67* (0.364) 0.449 (0.738)
Woman -1.017***(0.476) 0.03 (0.523) 0.31 (0.289)
Member of association -0.059 (0.261) -0.28 (0.182) -0.28 (0.212)
Log of rice area in 2006 -0.233** (0.118) -0.09 (0.077) 0.00 (0.089)
Farmer contact with extension 0.54** (0.246) 0.59*** (0.154) 0.494***(0.175)
Access to credit -0.017 (0.250) 0.30* (0.169) 0.45**(0.217)
Farmer contact with NARI 0.43*** (0.169)
Practice of upland farming 1.60***(0.226) 0.73*** (0.161)
Practice of lowland farming -0.07 (0.222) 0.06 (0.148)
West coast region 1.22*** (0.46) 0.32 (0.208)
NERICA introduction village 0.22 (0.209) 0.28**(0.141)
11
Conclusion• If every rice farmer is aware of the NERICA
varieties 16% will not be able to adopt due to insufficient supply of seeds
• For successful dissemination and adoption of NERICA concerted efforts should be made to increase farmer contact with extension
• Future studies should focus on measuring the intensity of NERICA adoption
12
Acknowledgment• Global Rice Science Program
• Africa Rice Center
• University of Hohenheim
13
14
Thank you!
14
References• Agricultural census. 2012: Technical report of the agricultural census of The
Gambia• Diagne A, Glover S, Groom B and Phillips J. 2012. “Africa’s Green
Revolution? The determinants of the adoption of NERICAs in West Africa” SOAS Department of Economics Working Paper Series, No. 174, SOAS, University of London
• Diagne A and Demont M. 2007. Taking a New look at Empirical Models of Adoption: Average Treatment Effect estimation of Adoption rate and its Determinants. Agricultural Economics, Vol 37 (2007). 30p.
• Dibba L, Diagne A, Fialor SC and Nimoh F (2012). Diffusion and Adoption of New Rice Varieties for Africa (NERICA) in the Gambia. African Crop Science Journal, Vol. 20, No. 1, pp. 141 – 153
• Planning Service Unit. 2011. Rice fact book. Unpublished technical report• Rosenbaum PR and Rubin DR.1983. “The Central Role of the Propensity
Score in Observational Studies for Causal Effects,” Bometrika 70, 41-55.
15