The Role of Infrastructure in Reducing Chronic and Transient Poverty The Case of JBIC Supported Irrigation Project in Sri Lanka Yasuyuki Sawada, Masahiro.

The Role of Infrastructure in Reducing Chronic and Transient Poverty ：

The Case of JBIC Supported Irrigation Project

in Sri Lanka

Yasuyuki Sawada,

Masahiro Shoji, and Shinya Sugawara

University of Tokyo

2

Research Objective and Strategy

• Is infrastructure an effective and efficient device to reduce poverty?

• Does irrigation infrastructure play an important role in reducing chronic and transient poverty?

• Research strategy: evaluation of a large irrigation infrastructure project in Southern Sri Lanka funded by the Japanese government through JBIC

3

Presentation Outline

• Overview of the project• Data description• Econometric analysis• Results

4

JBIC’s WLB project in Sri Lanka to be examined

• The Walawe Left Bank Upgrading and Extension Project – Initiated in 1997– Funded by the Japanese gov’t through JBIC – Improvements of irrigation systems complete

d in 2001 (Phase I)– On-going extensions (Phase II-)

5

JBIC Institute’s Evaluation Project in Sri Lanka

• “Impact assessment of irrigation infrastructure development on poverty reduction” – Collaboration with IWMI– 858 households were randomly sampled from six

strata of the whole left bank area– Unique household panel data exclusively for the study

• Rare seasonal panel for two years:

Maha(Rainy)

Yala(Dry)

Maha(Rainy)

Yala(Dry)

Oct.2000

2ndJul.

3rdOct.

4thMay2002

5thSep.

1stMay2001

6

Descriptive statistics Fig 1. Average monthly expenditure per adult male

Rainy season: Oct~March (Maha); Dry season: April~Sep (Yala)

500

1000

1500

2000

2500

Oct Dec Feb Apr J un Aug Oct Dec Feb Apr J un Aug

Rs.

per

Adu

lt M

ale

Food Consumption- Irrigated Total Consumption- IrrigatedFood Consumption- Rainfed Total Consumption- Rainfed

7

Descriptive statistics Fig 2. Average monthly Ag. Income per adult male

-500

0

500

1000

1500

2000

2500

3000

Oct Nov Dec J an Feb Mar Apr May J un J ul Aug Sep

Rs.

per

Adu

lt M

ale

Irrigated Rainfed


Average income: 799Rs. in irrigated, 610Rs. in rainfed.

8

Empirical model A

• Estimation equation a la Paxson (1993)• We regress monthly household expenditure (E) on

– constant– log income (lnY)– monthly dummies (M)– interactions of M with irrigation dummies (z).

jijjjjji ZMMYE *)ln()ln( 10

9

Table 5. Estimation result of model A

Food Non Food

Log of average monthly income

(measure of permanent income)

0.139(28.55)

0.235(13.55)

N 9016 8813

t-values in parentheses.

10

Estimation Result of Empirical Model A: Fig 3. Month effects of expenditure by irrigation availability


2.000

3.000

4.000

5.000

6.000

7.000


Food-Rainfed Food- IrrigatedNon Food-Rainfed Non Food- Irrigated

11

Summary of empirical model A

• There are significant differences in the month effects b/w irrigated and rainfed areas:– Chronic poverty is more serious in the rainfed

area– Decreasing month effects of non-food consum

ption in planting season, suggesting transient poverty in both areas

12

Empirical model B

• Consider poverty dynamics, particularly the role of credit explicitly

• Estimation equations (Type 5 Tobit model):Cj

j

jCCZj

Cj

Cj v

X

XZYE

)ˆ(1

)ˆ(lnln ,

0

if Hj < 0, If credit constraint is binding

Nj

j

jNNZj

Nj

Nj v

X

XZYE

)ˆ(

)ˆ(lnln ,

0

if Hj 0. If credit constraint is not binding

jjjj XEHandsinCashH *)(

13

Estimation result of empirical model B:Probit estimation of credit constraint eq. (reduce

d form version of Table 7)Credit Constraint

Coef. Std. Err.

Access to Irrigation Dummy -0.188*** 0.041

Active Organization Membership -0.165*** 0.042

Land Holding 0.120* 0.070

(Land Holding)2 -0.038* 0.022

Monthly Income 1.94E-06 4.99E-06

(Monthly Income)2 -5.03E-11 1.11E-10

Age of Head 0.004** 0.002

Female Head 0.115** 0.055

Head Count of Adult Male -0.051*** 0.019

Head Count of Adult Female 0.030 0.019

Head Count of Children -0.0003 0.014

Constant -1.250*** 0.093

N 9060

14

Table 8 and 9. Estimation result of model B:

Constrained Unconstrained

Food Non Food Food Non Food

Log of average monthly income

(measure of permanent income)

0.162(9.65)

0.202(3.53)

0.135(25.90

)

0.223(11.95)


15

Estimation result of empirical model B: unconstrained group (Table 5)


2.000

3.000

4.000

5.000

6.000

7.000


Food-Rainfed Food- IrrigatedNon Food-Rainfed Non Food- Irrigated

16

Estimation result of empirical model B: Credit constrained group (Table 4)


2.000

3.000

4.000

5.000

6.000

7.000


Food- Rainfed Food- IrrigatedNon Food- Rainfed Non Food- Irrigated

17

Summary of empirical model B • Access to irrigation mitigates :

– probability of binding credit constraint– negative welfare effects of credit constraints

• The monthly effects for the irrigated group seems consistently larger than those for the rainfed group. – Statistically, the difference is significant for un-constrained group– As to the constrained group, the gap of non-food consumption d

uring Yala season is significant – However, credit constraints cannot fully explain the remaining dif

ferences in the month effects between the irrigated and rainfed groups, suggesting that irrigation accessibility reduce poverty through multiple paths other than improvements in credit accessibilities.

18

A test of the model structure

• We test the validity of Paxson (1993) structure by using the following framework:– Examine whether household expenditure is sensitive to income change

s, treating the change as an endogenous variable– Investigate the coefficients on income variability π where the lack of cre

dit accessibility will enhance vulnerability even with irrigation.

Cj

j

jCj

CCj

Cj X

XAYE

)ˆ(1

)ˆ(lnln 0

Nj

j

jNj

NNj

Nj X

XAYE

)ˆ(

)ˆ(lnln 0

if Hj < 0, If credit constraint is binding

if Hj 0. If credit constraint is not binding

jjjj XEHandsinCashH *)(

19

Table 11 and 12 A test of the model structure


Constrained Group

Unconstrained Group

FoodNon Food

FoodNon Food

The smoother parameter π

Ratio of income earned in month to average monthly income

-0.02(-0.75)

0.21(2.12)

0.002(0.41)

0.003(0.21)

Permanent incomeAverage monthly income 0.14

(3.43)0.62(3.69)

0.17(10.82

)

0.25(7.91)

20

Concluding remarks• Irrigation reduces chronic poverty by enhancing permane

nt income possibly through improving productivity of agriculture.

• Access to Irrigation enhances credit availability of households– Theoretically, this will mitigate the negative welfare effects arisin

g from seasonality, i. e., transient poverty

• However, credit constraints cannot fully explain the remaining differences in the month effects between the irrigated and rainfed groups.– This suggest that irrigation accessibility reduce poverty through

multiple paths other than improvements in credit accessibilities.

21

Thank you very much!

The Role of Infrastructure in Reducing Chronic and Transient Poverty The Case of JBIC Supported Irrigation Project in Sri Lanka Yasuyuki Sawada, Masahiro.

Documents

yala slide

binding slide

jbic slide

yala season

n9060 slide

maha dry season

irrigation project

estimation result of