ETHIOPIAN DEVELOPMENT RESEARCH INSTITUTE

The impact of scaling up row planting on farmers’ teff yield

Joachim Vandercasteelen, LICOS, University of LeuvenMekdim Dereje, Bart Minten, Alemayehu Seyoum Taffesse, IFPRI ESSP-II / EDRI

Conference on “Improved evidence towards better policies for the teff value chain”10 October 2013Addis Ababa

1

2

Why?

• Teff is important

– Produced by 6 million farmers

– 2 out of 3 Ethiopians consume teff daily

• Low agricultural productivity

– 1.4 ton/ha

• Sowing technology

– Row planting

– Promising on-station results

– Widespread promotion campaigns

3

Program evaluation

• Row planting package was promoted to farmers

– Row planting technique

– Reduced seed rate

– Quncho variety

– Fertilizer

• Effect of promotion campaign on teff production

• Measured direct effect of row planting on farmers yield

Row planting has a positive effect on teff yield

ETHIOPIA AND TEFF

4

5

Teff in Ethiopia

• Agricultural sector

– 46% of GDP (World Bank, 2011)

– Growing 6%

• Teff: major staple food

– 22 % of grain area

– 11-15 % of all calories consumed

• Low agricultural productivity

– Limited research

– Low input use

– Lodging, reduced fertility, post-harvesting losses

6

Sowing technology

• Low yield related to sowing practice

– Broadcasting at high seed rate

– Uneven distribution

• New technologies increase yield

– Planting seeds in rows

– Lower seed rate

– Reduced competition

Reduced seed rate technologies

7

• Reduced seed rate technologies

– Promising on-station results

– High expectations by farmers

– Nation wide roll-out at large scale

• Program evaluation

– Provide objective data at farm level

– Provide evidence on the impact of campaign on teff yield

RANDOM EXPERIMENT

8

Design

9

• Roll-out in line with government program

– Pre-scale-up phase in 2012

– Both broadcasters and row planters are included

– Selection and extension done by DA (Development Agent)

– Free improved seed and fertilizer to all farmers• Experimental plot of 300 m²

• Quncho: 150 gr (row planters) vs. 900 gr (controls)

• 3kg of urea and DAP

• Randomization

• Control group

– Same package but traditional broadcasting

10

Randomization

• Randomization

– 2 stage randomization approach• 4 Farmer Training Centers (FTC) in 10 Woreda’s of Oromia

• 10 farmers row planting/ traditional broadcasting

• Model farmers

– Farmers are balanced in characteristics• Similar in terms of

– Education

– Age

– Distance to FTC

– Asset value

– Non-farm income

11

Overview of Woreda’s

12

Balancedness of farmersVariable Controls treatment

Mean se Coeff. t value

Household head characteristics

Age (years) 43.6 (0.94) -0.44 -0.37

Gender (male=1) 99.4 (1.23) -2.90** -1.85

Literacy (yes=1) 69.9 (3.56) 4.93 -1.09

Primary education (yes=1) 66.7 (3.68) 4.99 1.07

Household characteristics

Distance to FTC (minutes) 33.7 (2.04) 0.13 -0.05

Total household assets value (ln of Birr) 7.2 (0.16) 0.15 0.73

Total agricultural assets value (ln of Birr) 6.8 (0.08) -0.07 -0.72

Income from other activities (yes=1) 79.5 (6.98) -11.0 -1.24

Area (m²) 572.5 (28.5) -199.8*** - 3.70

Experimental plot

Red colored soil (yes=1) 31.4 (3.69) -1.88 -0.40

Brown colored soil (yes=1) 9.6 (2.24) -1.74 -0.61

Black colored soil (yes=1) 57.7 (3.96) 0.18 0.04

Tan colored soil (yes=1) 1.3 (1.45) 3.44** 1.87

Sloped plot (yes=1) 16.7 (2.81) -3.67 -1.03

Improved Quncho seed used (yes=1) 99.4 (0.68) -0.15 -0.17

Distance to plot from house (minutes) 10.8 (0.80) -0.25 -0.24

Number of tilling (number) 4.9 (0.13) 0.09 0.60

Organic input used (yes=1) 12.8 (2.56) -2.19 -0.68

Inorganic fertilizer used (yes=1) 99.4 (0.40) 0.64 1.28

Number of weedings (number) 2.0 (8.14) 0.12 1.19

Amount of UREA used (g/m²) 9.1 (0.34) 1.22*** 2.43

Amount of DAP used (g/m²) 11.4 (0.51) 0.4 0.57

Value of herbicide used (birr/ha) 196 (17.7) 17.3 0.51

Surveys

13

• Quantitative: 3 surveys

– Baseline before harvest

– Crop-cut at harvest

– Impact after harvest

• Qualitative

– Focus groups

– Community questionnaires

– Farmers’ opinion, perceptions, planning

Data collected

14

• Area measure

– By compass and rope (crop-cut)

• Output measures

– At harvest

– After harvest

• 3 yield measures

– Assessed yield before harvest

– Measured yield at harvest (crop-cut)

– Reported yield after harvest

Methodology

15

• Farm level

– Randomized control trial

– 410 farmers

– Mean difference in yield is due to sowing only

– Direct row planting effect

• Village level

– 10 trials at each FTC (331 plots)

Effect of row planting

16

Farm: 2-22% FTC: 26%

22% before harvest2% measured at harvest16% after harvest

CONCLUSION

17

18

Conclusions

• Low teff yield asks for adoption of new technologies

• Promising on-station results of row planting

• Large scale promotion campaign

• On-farm effect of the promotion campaign

• Look at direct effect of row planting

– Teff yield increased by 2-22% at farm level

– Teff yield increased by 26% at village level

• Yield increase of more than 20% is already strong

– Value of several million $

19

Implications

• Effect is likely to be even higher

– On- farm constraints

– First year of adoption

– Farmers need to learn

– Implementation issues of promotion program

• Row planting is offered as a package

– Effect of whole package is stronger

• Assess on-farm constraints to adoption

• Research on different components of technology package and their interactions

THANK YOU

20