Pattern of rice variety adoption and potential impact of improved

variety in GSR target countries

Huaiyu Wang

Sushil Pandey

Orlee Velarde

June 22, 2012

GSR• What is GSR?• The improved lines are expected to be stress-tolerant

and high yielding. It is anticipated that the efficiency of chemical inputs will be increased when these stress tolerant varieties become widely adopted (Zhang 2007).

• What is GSR project?• “Green Super Rice for the Resource Poor of Asia and

Africa (GSR)” funded by Gates foundation

• IRRI: South Asia and Southeast Asia

• My study: Cambodia, Sri Lanka and Pakistan

Objectives of this study1. To analyze the patterns of adoption and diffusion of existing improved

varieties, and identify constraints to adoption;

2. To analyze the economics of rice production and farmer livelihood strategies and understand the gender roles in rice production and women’s participation in decision making;

3. To estimate the potential impact of improved varieties being developed under the project on rice production, farmer income and poverty reduction;

4. To draw implications for technology development, targeting and policy reforms.

5. To build capacity for socio-economic analysis of technology.

Outline

• Background

• Objective

• Methodology

• Key findings

• Study on specific issue

• Summary and implications

Methodology

• Data: • Secondary data• Focus group discussion (FGD) • Household baseline survey

• Analytical framework: • Secondary data and household level analysis• Descriptive statistics and econometric analysis

KEY FINDINGS

Rough rice area, yield and production, GSR countries, Asia (2008-2010)

Data source: FAOSTAT and national statistics.

 

Area(million ha)

Production (million ton)

Yield(t/ha)

SE Asia

Cambodia 2.7 7.7 2.9

S Asia

Sri Lanka 0.9 3.9 4.4

Pakistan 2.7 6.2 2.3

Main features of rice production ecosystem

Countries Production environment

Market orientation Yield level

SE Asia

Cambodia Mostly rainfed Increasing export potential

Low

S Asia

Sri Lanka Mostly irrigated Self-sufficient High

Pakistan Mostly irrigated An important rice export country; High quality rice (Basmati etc.)

Low

Figure 1. Rice yield trend in 1980-2009 (t/ha)

Data source: FAOSTAT.

Poverty in GSR countries

Data source: World Bank database

   

National poverty

Rural poverty

Number of rural poor people

    (%) (%) (million)

SE Asia 35.1

Cambodia 30.1 34.5 4.0

S Asia 83.2

Sri Lanka 15.2 15.7 2.7

  Pakistan 22.3 27.0 29.1

Farm level analysis

Sample design

Sample size

Key stresses

Surveyed districts

Institutions Collaborator

SE Asia         

 

Cambodia 607 D/Sub/Sal

Battambang, Pursat,

Kampong Thom, Kampot,

Prey Veng, Takeo

SMESam Bona; Piset Mease

S Asia

 

Sri Lanka 404 D/Sub/SalKurunegala, Kalutara, Puttalam

RRDI, SEPDC

Nimal Dissanayake

 Pakistan 210 D/Sub/Sal Sindh, Panjab UAF

Abedullah Anjum

Characteristics of farm households

  CountryFarm size

(ha/hh)

Rice yield (t/ha)

% of rice income in total

household income

Irrigation of paddy

(%)

Crop intensity

(%)

SE Asia

Cambodia 1.8 2.8 44 23 113

S Asia

Sri Lanka 1.2 2.6 11 40 150

  Pakistan 5.5 3.6 16 90 187

Major varieties grown in the countries

Data source: Household survey in GSR project 2010.

Variety name % area Yield (t/ha)Released

year 

Cambodia 504 (IR 50404-57-2-2-3) 21 4.1 1990  

IR 66 13 2.7 1990

TV varieties 59 2.3 -

Pakistan IRRI-6 56 3.5 1971

IRRI-9 12 3.7 1999

Pukhraj (hybrid) 11 4.9 -

Super Basmati 10 1.9 -

Sri Lanka BG300 62 2.6 1987

BG352 13 3.1 1992

BG358 7 2.4 1999

• Adoption of improved varieties generally high but adoption in stressed environments characterized by “patchiness”.

• One or two major varieties accounting for a large area (or mega varieties)

• Adopted varieties are generally older, with limited adoption of newly-released varieties in the main wet season.

• Average yields are low despite high incidence of adoption of improved varieties.

Key results for variety adoption

Economics of rice production

Structure and sources of household income

Cambodia Pakistan Sri Lanka

% rice 44 16 11

% non rice 1 38 10

% animal sale 13 1 17

% off-farm income 2 9 1

% nonfarm income 40 36 61

Total income (in USD) 1,688 3,075 3,475

Per capita income (in USD) 0.94 1.35 2.44

Gender analysis

Objective• Gender roles play an important role on rice farming and

household’s decision-making process.

• Gender roles and responses vary across and within cultures.

• Taking Cambodia and Sri Lanka as examples, the objective of this study is to compare the women farmers’ empowerment and gender roles in rice farming systems between subsistence- and market-oriented rice farmers.

Sri Lanka Cambodia

Women empowerment index (WEI)Sri Lanka Cambodia

Rice farming decisions1.   What rice variety(ies) to grow 2.1 3.02.   Adoption of technology in rice production 2.2 2.83.   What farm implements to purchase 2.2 2.84.   Who and number of farm labor to hire 2.2 3.25.   Whether to sell  or consume the harvested crop 2.4 3.46.   Quantity of output to sell and consume 2.5 3.47.   When and where to sell the harvested crop 2.3 3.48.   What price to sell the output 2.3 3.4Income and expenditure9.    Allocation of farm income 2.5 3.410.  Allocation of household income 2.7 3.511.  What types of food to consume in times of crisis 2.8 3.612.   Where to borrow 2.7 3.2Childcare13.   Children’s education 3.0 3.214.   Number of children to raise -- 3.2Others15.  Participation in voting/politics 2.5 3.116.  Whether to sell or slaughter the animal -- 2.9Average WEI 2.5 3.2

OLS regression model of the factors contributing to women empowerment

Dependent variable is WEI Sri Lanka Cambodia

Distance to market (km) 0.00178 0.0435***

Years of education of wife 0.0373** 0.0143

Age of wife 0.0112*** 0.00746***

Dummy for wife with non-farm primary occupation -0.123 0.0309

Percentage of females in the households 0.00619** 0.00273*

Farm size (ha) -0.00755 -0.0542***

Percentage contribution to non-farm income of female 0.00318* 0.0000825

Percentage contribution to non-farm income of male 0.00105 -0.00197***

Dummy for husband who attended a training -0.196* -0.195***

Dummy for wife who attended a training 0.435*** 0.154***

Constant 1.088*** 2.713***

N 378 593

Key results for gender analysis

• Women in Cambodia have higher women empowerment index (WEI) compared to Sri Lanka.

• Women are involved in more diverse rice farming activities in Cambodia than in Sri Lanka.

• In both countries, women exposure to training has positive significant effect on women empowerment and the effect of training on husband is negatively significant.

• In Sri Lanka, education levels of wives and their contribution to nonfarm income increase the women empowerment index significantly.

Ex-ante impact assessment

Periods: • Short term (3-5 years) and Long term (6-10 years)

Assumptions: • the size of the potential yield gain: 10%• the adoption rate: 10% and 20%

A simple pragmatic approach• Rice farmers being lifted out of poverty:

(Rice farmers*rural poverty ratio) * % poor lifted out of poverty in the survey

• Additional people meeting food requirement:

Target area * Yield gain * Price/Consumption

Ex-ante impact assessment (1)

Ex-ante impact assessment (2)

Summary and implications

• The three countries analysed represent a diversity of rice production environments, technology levels and the institutional set up.

• Yield levels in all countries are low, especially in areas that are stress-prone.

• Efforts to develop improved rice varieties that are tolerant to such stresses are thus very important.

Implications

Germplasm development strategy

• Grain quality characteristics: The new lines and /or varieties to be developed should go through the proper grain quality test and evaluation or whatever is needed to make sure these are the traits that farmers desired, especially the quality issues for hybrid rice.

• Mega varieties and Breeding strategy: The dominance of mega varieties basically indicates that breeding strategy may build on the existing materials and include some additional desirable traits to facilitate rapid dissemination. Grain quality could be such an additional consideration as farmers did rank grain quality as second most important trait after the yield.

Implications

Targeting

• Poverty reduction: In terms of the potential impact on rural poverty, it would be desirable to consider environments with abiotic stress as the primary target of GSR varieties given the high incidence of poverty in such environments and the low current average yield.

• Rice farmers: Farm-level impact of adoption of GSR varieties in terms of the incremental income is higher for those farmer categories (or locations) for whom rice accounts for a larger share of total household income. Hence, it is desirable to have a dissemination strategy (at least at the initial stages) that is targeted to such farmers/locations.

• Training: To provide training on women farmers would be helpful to improve women empowerment in the family decision.

Implications

Dissemination

• Availability and access to quality seeds: In poor rainfed areas, limited access to quality seeds of improved varieties remains a problem due to a number of institutional constraints. Increased investments in extension and participation of local agencies and NGOs will be needed for accelerating the process of technology diffusion.

Nonfarm income and technical efficiency in Sri Lanka

• One-stage stochastic frontier analysis (SFA) regression (Battese and Coelli,1995)

• Cobb-Douglas production function

• Technical efficiency model

jj

n

iijjj UVXY

1

lnln

n

iijjj ZU

1

Production function Inefficiency model

Seeds

Organic fertilizer

Chemical fertilizer

Pesticides

Herbicides

Power

(Animal, tractor, thresher, harvester)

Labor

Age

Education

Household size

Farm size

Nonfarm income share

Square of nonfarm income

share

Variables

Variables Mean Std. Dev. Min Max

Age of respondent (years) 53 12 24 88

Education of respondent (years) 8.5 2.8 0 13

Household size (persons/hh) 3.8 1.1 1 8

Farm size (ha/hh) 1.3 0.9 0.23 4

Household income (US$/hh) 3267 3328 123 17100

Nonfarm income share (%) 52 38 0 100

Share of rice income (%) 11 26 0 100

Characteristics of farmers

01

02

03

0N

o. o

f fa

rm h

ou

seh

old

0 .2 .4 .6 .8 1Share of nonfarm income

05

10

15

20

No.

of h

ous

eho

ld

0 .2 .4 .6 .8 1Technical efficiency

Figure 2. Distribution of technical efficiency

TE mean value = 0.63

Coefficient Standard error

Age of respondent (years) 0.007 (0.009)

Education of respondent (years) 0.029 (0.035)

Household size (persons/hh) -0.047 (0.076)

Farm size (ha/hh) 0.224* (0.127)

Share of nonfarm income (%) -3.567** (1.556)

Square of share of nonfarm income 4.187** (1.701)

Rice intensity 0.521* (0.294)

Constant -1.148 (1.145)

N 120

Determinants of household technical inefficiency

Standard errors in parentheses* p < 0.10, ** p < 0.05, *** p < 0.01

Conclusion

• The livelihood strategy of rice farmers in Sri Lanka is oriented more towards nonfarm income.

• There is substantial potential to improve farmers’ practices (TE=63%).

• The effect of nonfarm income is kind of U-shape effect.

• Adoption is measured using two indicators: incidence of adoption and intensity of adoption.

• For analyzing the incidence of adoption, a farmer is considered to be either an adopter or a non-adopter – Probit model

• The extent/ intensity of adoption is measured as the proportion of area under improved varieties – Tobit Model

Modelling

• The decision problem for a farmer involves the choice of two possible varietal categories, namely, modern varieties (MV) and traditional varieties (TV and iTV).

• Variations could be influenced by demographic characteristics, landholdings, access to market and variety, cropping pattern and location

Description of variablesExpected effect on adoption

Dependent variableAdopt Farmer grew modern varieties. 0= no, 1=yes

PMVarea Share of modern variety area in total rice area (%)

Explanatory variablesAge Age of respondent (years) +

DfemaleDummy variable of the gender of respondent. 0=male, 1=female

?

Hhsize Household size (persons) +Farm size Farm size (ha/hh) +Plowarea Percentage of lower field in the farm size (%) +

Pmidarea Percentage of middle area in the farm size (%) +

Pirrigarea Percentage of area irrigated (%) +

Tborder Dummy of location. 1= border with ; 0= border with -

Inland Dummy of location. 1= inland; 0= otherwise -

Market Distance from nearest market (km) -

Description of covariates

Non-adopter

(TV & iTV)MV adopter All

Hhsize (persons) 4.84 4.98 4.90

Labor (persons) 2.78 2.91 2.83

Respondent dummy (1= female) 0.67 0.50 0.60

Age (years) 45.81 44.55 45.28

Average education (years) 5.56 5.36 5.47

Market distance (km) 4.04 3.78 3.93

Farm size (ha/hh) 1.45 2.19 1.77

Lower field area (ha/hh) 0.76 1.16 0.93

Middle field area (ha/hh) 1.29 1.07 1.15

Area irrigated (ha/hh) 0.56 1.80 1.09

Rice area (ha/hh) 1.39 2.69 1.94

MV rice area (ha/hh) 0.00 1.88 0.80

Rice production (ton/hh) 2.84 8.75 5.37

Rice yield (t/ha) 2.09 3.36 2.63

% of rice production sold 24.71 55.45 37.83

Comparison on characteristics of non-adopters and adopters of different varieties

Factors affecting the incidence and intensity of modern variety adoption

Incidence of adoption

Intensity of adoption

Probit Tobit

Household size 0.0132 1.108

Age of respondent -0.0061 -0.255

Respondent gender (1=female, otherwise 0) -0.1900 -3.655

Farm size 0.2120*** 5.280***

Share of lower field 0.0096*** 0.435***

Share of middle field 0.0045* 0.254***

Share of irrigated area 0.0108*** 0.595***

Region dummy (Thailand border) -2.1900*** -83.76***

Region dummy (Inland province) -3.3900*** -150.9***

market -0.0470* -1.270

N 607 607