Is Mozambique on the eve of rice Green Revolution? A case study of

the Chokwe irrigation scheme

Kei Kajisa (IRRI & FASID)Ellen Payongyong (Michigan State Univ.)

Introduction (1)• The importance of rice in Moz. has been

increasing rapidly.

0

100

200

300

400

500

600

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005

Year

1000

to

n

Production

Consumption

•Rapid consumption growth

•Stagnated production growth

•Rapid increase in rice imports

•Increasing foreign exchange expenditure

Introduction (2)• Why has production stagnated since

2000?

0

20

40

60

80

100

120

140

160

180

200

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005

Year

1000

ha

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

ton

/ha Area harvested

Paddy yield

•Stagnated area expansion after the completion of re-settlement.

•Stagnated rice yield at 1 t/ha.

Introduction (3)• Moz. needs an appropriate rice development

strategy.• Dominant ecosystem in Moz.

– 61% lowland wet condition, upland dry condition is minor (Balasubramaninan et al., 2007)

• Seeking an Asian-style GR with an appropriate localization effort could be a key strategy.– Controlled irrigation + modern varieties + modern

inputs (chemical fertilizer) Skepticism (Africa’s condition is different from

Asia!) • ↑But many are regional comparison. To investigate the

potential of rice, it is better to compare rice growing areas.• ↑Some anecdotal success stories are emerging

Introduction (4)

• More solid micro-level evidence is needed to design an appropriate strategy beyond mere skepticism.

• The aim of this study– Explore the potential of and constraints to a

an Asian-style lowland rice GR in Moz., using household-level data collected in the Chokwe irrigation scheme in 2007 (w/ financial support from JBIC, JSPS, IFAD).

Maputo

Chokwe

Why Chokwe ? (1)• Chokwe Irrigation Scheme

– Gravity system, managed by the state

• The most favorable area for rice cultivation.– Its potential is as high as the banks of Nile (10 t/ha)!

• Asia’s experience– GR started in the most favorable areas and then

extended to less favorable areas (David and Otsuka, 1994)

– Thus, it makes sense to explore if Chokwe has a chance to be a starting point of Moz’s rice GR.

We compare the current condition of Chokwe with those of Asia’s favorable areas in 60-70’s to conjecture about possible constraints in Chokwe. (1st step of our analysis)

Why Chokwe ? (2)• Besides, within Chokwe, we can observe

wide variation in yield, cultivation practice, and household socio-economic conditions.

• Yield: 3.8 – 0.6t/ha (top25% vs bottom 25%)• NPK use: 67 – 0 kg/ha (top25% vs bottom 25%) • Different access to irrigation water• HH size and schooling years

Statistically identify household level constraints by estimating a yield function and factor demand functions (2nd step of our analysis)

no water, a lot of weeds, low yield

Sufficient water, proper management, high yield

Two plots in Chokwe, different performance

Comparison w/ Asia: YieldCho-kwe

Laguna Philippines

Central Luzon, Philippines

Tamil Nadu*

Moz/ Asia’s 1960s

Moz/ Asia’s1970s

Year 2007 1966 1976 1967 1971 1971

Major rice varieties

MV1 TV MV1 TV MV1 MV1

Paddy Yield (t/ha)

Average 2.0 1.9 2.8 1.9 2.4 2.7 1.1 0.8

Average of Top 25%

3.8 2.7 4.2 2.7 4.5 3.7 1.4 0.9

Comparison w/ Asia: Prices

Cho-kwe

Laguna Filipinas

Época chuvosa

Luzon Central Filipinas

Época Seca

Tamil Nadu*

Moz/ Asia’s1960s

Moz/ Asia’s1970s

Year 2007 1966 1976 1967 1971 1971

Nitrogen(kg of rice/ kg of N)

7.9 3.2 3.0 2.6 2.4 2.7 2.9

Labor wage (kg of rice/day)

12.0 9.6 8.5 7.9 5.8 6.6 1.4 1.7

Comparison w/ Asia: Input Intensity

Cho-kwe

Laguna Filipinas

Época chuvosa

Luzon Central Filipinas

Época Seca

Tamil Nadu*

Moz/ Asia

Moz/ Asia

2007 1966 1976 1966 1976 1971 1960s 1970s

Nitrogen (kg/ha)

19 13 58 17 59 52 1.2 0.3

Labor (days/ha)

76 89 105 72 79 157 0.9 0.6

Prop of hired labor

36 49 71 60 43 73 0.6 0.5

Use of credit (%)

15 27

Comparison w/ Asia: Socio Economic CharacteristicsCho-kwe

Laguna Filipinas

Época chuvosa

Luzon Central Filipinas

Época Seca

Tamil Nadu*

Moz/ Asia

Moz/ Asia

2007 1966 1976 1967 1971 1971 1960s 1970s

Farm size (ha)

1.7 1.5 2.1 1.4 1.1 1.0

Household size

8.0 5.9 5.9 7.6 6.2 5.9 1.3 1.2

Schooling Years of head

3.6 4.6 4.4

Average Schooling Years

4.3 3.3 5.0

Comparison w/ Asia: Findings

• Yield is higher than Asia before GR but slightly lower than Asia during the early phase of GR. Potential exist, but slightly lagged behind

• Extremely expensive N, moderately expensive labor. Less use of these inputs

• Inaccessible credit and insufficient water could be other reasons for less intensive use of N.

• Synchronization of peak labor season and few landless may make intensive use of labor further difficult.

• On the other hand, no discernible differences in – the extent of mechanization, farm size, schooling

years, family size, age.

Determinants of rice yield (1): Approach

Yield function y=f(xc, xg, h)

Factor demand functionsxc=g(xg, z, h)

xc: inputs (choice of a farmer, endogenous, e.g. NPK)xg: inputs (given to a farmer, exogenous, e.g. irrig’n)h: HH management ability (educ and age)z: factor endowments/wealth (HH size, non-ag assets, etc)

• To the make the IV method valid, we would like to have less xc and more z.

linear yield function (no squared terms of xc) and include h only in factor dem. functions as a part of z.

Determinants of rice yield (2): Yield function

(5) IV NPK a 0.022 (2.02)** Labor a 0.008 (2.06)** Prop of hired labor a -0.386 (0.49) Use of tractor a 0.207 (0.29) Use of thresher a 1.483 (1.36) Transplanting a -0.369 (0.47) Size of cultivated area -0.126 (2.06)** Insufficient irrigation (relative freq.) -0.694 (1.81)* Downstream parcels -0.459 (1.34) Constant 1.536 (1.99)**

Determinants of rice yield (3): Factor Demand Functions (selected vars)

NPK Labor % of hired lab

Tractor Thresher

Transplanting

Cult area – + + + –

Insuf irrig’n – +

HH size – +

Sch yr +

Sch yr sq –

non-ag assets + + +

% of salary earners

+ – –

% of salary earners sq

– + +

Determinants of rice yield (3): Decomposition of yield increasing factors

Bottom

20%

Top 20%

Difference

Contribution

to yield

increase (kg)

Share of

contribution

(%)

Yield (t/ha) 0.97 3.55 2.58 NPK (kg/ha) 4.94 50.71 45.77 1.02 40 Labor (days/ha) 39.25 129.23 89.98 0.74 29 Prop of hired labor 0.26 0.39 0.13 -0.05 -2 Use of tractor (dummy) 0.38 0.71 0.34 0.07 3 Use of thresher (dummy) 0 0.23 0.23 0.34 13 Transplanting (dummy) 0.81 0.77 -0.04 0.01 1 Cultivated area (ha) 1.99 1.31 -0.68 0.08 3 Insufficient irrigation 0.36 0.03 -0.34 0.23 9 Downstream parcels (dummy) 0.38 0.09 -0.29 0.13 5 Sum of the contributions 2.58 100

14% incl. indirect impacts

Conclusion (1)• The conditions of Chokwe for embarking

on the Asian-style lowland rice GR are not as disadvantageous as those of Asia in the past, except for excessively expensive chemical fertilizer and moderately costly labor.

• These inputs are used less intensively in Chokwe, resulting in low yield.

• Inaccessibility to credit and insufficient water lower input intensity further, resulting in further low yield.

Conclusion (2)• Implications

– Improvement of irrigation management is important.

– Worth thinking strategies to improve• farmers’ credit access and • liquidity constraint (↑cash on hand)

– Introduction of improved modern varieties• Shorter maturity duration variety would relax the timing

of transplanting and thus spread out the peak season labor demand.

• Remaining issues– How to make chemical fertilizer cheaper?

Limpopo

Thank you very much