Determinants of Changing Behaviors of NERICA Adoption: An Analysis of Panel Data from Uganda Yoko Kijima (University of Tsukuba) Keijiro Otsuka (FASID)

Determinants of Changing Behaviors of NERICA Adoption: An Analysis of Panel Data from

Uganda

Yoko Kijima (University of Tsukuba) Keijiro Otsuka (FASID)

Dick Sserunkuuma (Makerere University)

Workshop on African Green Revolution Tokyo

December 7, 2008 1

2

Motivation and Background

• There is need for enhancing crop yield and food production in SSA for alleviating widespread poverty and food insecurity in SSA

• Rice in Uganda:– Rice is one of the high value crops and consumption has

been increasing in Uganda – Except very limited areas, most of the farmers had

never grown rice before NERICA– Seed distribution program: Seed (+ Training)

3

Motivation and Background

NERICA 1 Survey (2004) Data was collected soon after NERCIA seed dissemination program

initiated by Vice President of Uganda Sample: 10 NERICA growing areas in Central and Western regions

(29 LC1s), 250 NERICA and 150 non-NERICA households in 2004 (stratified random sampling)

• Poverty alleviation in Uganda using NERICA– Kijima, Sserunkuuma, and Otsuka (2006): NERICA yield

is more than twice as high as upland rice yield in SSA. – Kijima, Otsuka and Sserunkuuma (2008): Simulation

analysis suggests that introduction of NERICA increases in per capita income by 12% and decreases in poverty head count ratio by 5 percentage points, without deteriorating income distribution

4

Motivation and Background

NERICA 2 Survey Total sample size becomes 347 households

Attrition due to moving-out from sample areas, dissolution of households, no contact during the data collection period

Four types of farmers: (1) non-adopters, (2) dropouts (adoption only in 2004), (3) continuous adopters, and (4) late adopters (adoption only in 2006).

• Adoption pattern: Panel data in 2004 and 2006:– 50% of NERICA adopter in 2004 abandoned it in 2006– 20% of non-adopter in 2004 adopted NERICA in 2006

• Is NERICA really profitable?• Is the profitability of NERICA sustainable?

5

Other Observations• Changes in markets between 2004 and 2006

– Market of rice milling service have developed (Fig. 1)– The use of self-produced seeds by farmers became

common– NERICA yield among continuous adopter is higher than

dropout and late adopter

0

100

200

300

400

500

600

700

1999 2000 2001 2002 2003 2004 2005 2006 2007

Source: Alphonse et al. (2008)

Table 1a. Sample Household Characteristics in 2004Non-

adopterDropou

tContinuous adopter

Late adopter

Number of households 94 129 99 25

Rice experience (years) 0.07 1.49 1.88 0.50

Household head’s age 48.7 48.2 43.9 49.4

HH head’s years of schooling

4.9 7.07.7

4.5

Number of male age 15-59

1.43 1.801.81

0.99

Number of female age 15-59

1.50 1.961.68

1.34

Female headed household 0.33 0.10 0.08 0.30

Bakiga tribe 0.00 0.08 0.09 0.20

Per capita land owned (ha)

0.24 0.380.47

0.38

Land (owned + mailo) (ha)

2.56 4.524.23

2.53

Land cultivated (ha) 0.92 1.19 1.30 0.78

Household asset (USD) 160.2 149.4 171.9 53.7

Value of livestock (USD) 306.9 370.5 389.7 79.5

6

Table 1b. Sample Community Characteristics

Non-adopter

Dropout

Continuous adopter

Late adopter

Availability of seed program 04

0.19 0.370.34

0.17

Availability of seed program 06

0.11 0.200.29

0.24

Distance to rice miller 04 (km)

19.0 15.426.9

28.9

Distance to rice miller 06 (km)

14.1 11.16.2

5.5

Traveling time to town (hour)

0.42 0.620.77

0.66

Area size per HH (mile2) 0.02 0.02 0.02 0.02

Relative price of maize to rice

0.39 0.510.47

0.45

Rainfall 04 (mm, crop. season)

368.5 424.0429.6

393.0

Rainfall 06 (mm, crop. season)

446.2 435.8432.0

450.2

Average rainfall (01-06) 409.4 435.5 441.3 421.6

C.V. of rainfall (01-06) 0.20 0.17 0.15 0.17

7

Summary of Descriptive Table 1

• Early adopters are more educated, have more adult members, and more assets than the other groups

• There are no differences in household characteristics between dropouts and continuous adopters

• In 2004, seed distribution program seems enhance NERICA adoption, while availability of the program declined in 2006, except late adopters

• Access to rice miller improved in 2006, especially in areas with continuous and late adopters

• Rainfall variation is higher in areas with non-adopter

Table 2a. Sample NERICA Plot Characteristics

2004 2006

Dropout

Continuous

adopter

Continuous

adopter

Late adopter

Yield (ton per ha) 2.01 2.97 2.54 1.49

Seed source (%)

Self-produced seed 5.2 7.7 41.5 5.8

Purchased seed from neighbor 3.8 11.7 12.0 46.5

Program seeds (NGO, VP) 53.8 42.9 15.0 10.2

Other (purchased, contract farming)

37.2 37.7 31.5 37.5

Yield (ton per ha) by seed source

Self-produced seed 0.99 3.41 2.06 1.15

Purchased seed from neighbor 1.35 2.72 2.97 1.17

Program seeds (NGO, VP) 1.76 2.96 2.95 3.46

Other (purchased, contract farming)

2.49 2.99 2.75 1.39

Plot size (ha) 0.42 0.38 0.47 0.24

Table 2b. Sample NERICA Plot Characteristics

2004 2006

Dropout

Continuous

adopter

Continuous

adopter

Late adopter

% plots in low lying location 11.8 14.1 21.7 21.8

Yield in low lying location 2.04 2.83 2.09 1.22

Yield not in low lying location 2.01 3.00 2.67 1.56

% late planting 8.5 2.8 11.1 6.1

Yield in late planting plot 2.20 3.05 2.56 1.43

Yield not in late planting plot 0.68 0.63 2.44 2.31

% plot with zero yields 4.4 0.4 6.9 24.9

Summary of Descriptive Table 2

• The average yield for continuous adopters are higher than that of dropouts and of late adopters

• 25% of the late adopters failed crop totally (zero yields)• Type of seeds used changed from program seeds to

self-produced and seeds from neighbors • Yield of self-produced seeds tends to be lower • Among continuous adopters, yield of self-produced

seeds is lower than that of the other type of seeds• Other management issues between continuous and late

adopters: low lying areas and timing of planting – no significant difference

Purposes of Estimation Model

• Why did so many early adopters become dropouts?• What are the underlying factors for the superior

performance of the continuous adopters?• How can we increase NERICA yields?

* Focus: effects of rainfall, availability of seed program, access to rice millers, and availability of seeds on adoption and yields

Estimation Model

Changing behavior of adoption decision• Multinomial logit model of 4 adopter types:

Non-adopter, dropout, continuous adopter, late adopter

Yield function• Cross section data (regress 2004 and 2006 separately) • Controlling for selection bias of adopting rice• Possibility of endogeneity of seed type used: 2

specifications – (a) availability of seed program, (b) set of seed types dummies

Table 3. Determinants of NERICA Adoption (Multinomial Logit Model, Marginal Effects)

Community characteristicsDropou

t

Continuous

adopter

Late adopter

Non-adopter

Average rainfall (01-06) -0.000 -0.000 0.001 -0.000

C.V. of rainfall (01-06) 0.421+ -0.131 -0.655** 0.365

Availability of seed program 04

0.096+0.082*

-0.122+ -0.056

Availability of seed program 06

0.017-0.021

0.077 -0.073

Relative price of maize to rice 04

0.042-0.042

0.143+ -0.142

Community area size per HH 04

-3.286*-2.112+

-1.561 6.959**

Distance to rice miller 04 (km)

-0.002*-0.000

0.006** -0.004

Distance to rice miller 06 (km)

-0.001-0.003*

-0.006** 0.007**

Traveling time to town 04 (hour)

0.165*0.145**

-0.201* -0.109

14

Table 3. Determinants of NERICA Adoption, cont (Multinomial Logit Model, Marginal

Effects)Household characteristics

Dropout

Continuous adopter

Late adopter

Non-adopter

Rice cultivation experience

0.063*0.039*

0.006+ -0.163**

Household head’s age 0.021 0.005 -0.069** 0.043

HH head’s years of schooling

0.0230.002

0.007 -0.032

Number of male age 15-59

-0.075+

-0.0240.022 0.078

Number of female age 15-59

0.1160.034

0.367 -0.517*

Female headed household

0.001-0.001

0.000 -0.000

Bakiga tribe 0.005 0.008* 0.007 -0.020**

Per capita land (ha) -0.001 0.002 0.034 -0.037

Household asset (USD) 0.009 0.005 -0.034 0.021

Value of livestock (USD) 0.010 0.016 -0.155* 0.130

Constant -0.256* -0.102 -0.292+ 0.650**

15

16

Summary of Regression Results

NERICA adoption• Rainfall variation:

• Dropout: higher variation (mis-targeting of the program)• Later adopter: smaller variation (suitable)

• Availability of seed program:• Increase adoption in 2004• No effect on adoption in 2006 - probably due to the self-

production of seeds and trade among farmers. • Improved access to rice millers increases adoption

Table 4. Determinants of NERICA Yield (with selection correction term)

2004 2006 2004 2006

(1) (2) (3) (4)

Availability of seed program -0.347 0.378**

Seed source dummies

Seed from neighbor -0.294 0.051

Program seed -0.614 1.140**

Purchased from seed company

-0.670 -0.219

Rainfall (cropping season) -0.000 0.003** 0.003** 0.002**

Distance to rice miller (km) 0.010 -0.023* 0.008* -0.024*

Traveling time to town 0.160 0.172 0.292 0.450*

17

Table 4. Determinants of NERICA Yield, cont.

2004 2006 2004 2006

(1) (2) (3) (4)

Rice cultivation experience

0.073* 0.124** 0.031 0.128**

Number of male age 15-59

0.217+ 0.064 -0.176** 0.137+

Number of female age 15-59

0.162 -0.114 -0.024 -0.165*

Female headed household -0.106 -0.216 -0.341 -0.131

Bakiga tibe -1.529** -0.328+ 0.159 -0.586**

Household head age -0.032** -0.021** -0.026** -0.012**

HH head’s education 0.045 -0.034+ 0.024 -0.026

Per capita land (ha) -0.120 0.243 0.253* 0.253+

Household asset # 0.082 0.495 -0.242 0.577*

Value of livestock # -0.265 0.365* -0.291* 0.280

Constant 2.843** 1.745** 2.386** 1.095*18# Unit is thousand USD

19

Summary of Regression Results

NERICA yield• Quality of self-produced seeds seems to be deteriorated

over time:– Availability of seed program and use of program seed

(treated seeds) increase yield only in 2006• Rainfall increases yield• Rice cultivation experience increases yields• Shortened distance to rice millers increases yield in 2006

20

Conclusions

• Adoption process is dynamic, involving dropout

• We observed strong indications of the failure of extension (provision of inappropriate or insufficient information on seed production and crop management, mis-targeting of seed distribution program areas, etc.)– Strong need for capacity building of extension workers

• Markets have developed responding to demand for market services. Such development promoted adoption and yield growth.– Support for further development of markets will

contribute to the efficiency of rice production.