31 May 2013
Marie-Charlotte Buisson, Arijit Das, Aditi Mukherji
What determines contribution to a common fund for upkeep of water infrastructures? Evidence from
experimental game in Coastal Bangladesh
INTRODUCTION
• Embankments constructed by the Government of Bangladesh (GoB) across the entire coastal zone in the 1960s and 1970s.
• 1st objective: Protection for tidal surge, flood, natural calamity• 2nd objective: Increasing agricultural productivity
INTRODUCTION
• Operation and maintenance of the infrastructures is the key challenge to ensure the sustainability of the system.
• GoB requires local communities to organize themselves into Water Management Organizations (WMOs) and contribute towards minor maintenance of water infrastructure.
• National Water Policy of 1999 (MoWR, 1999) • Guidelines for Participatory Water Management, (MoWR, 2001).
RESEARCH PURPOSETo understand the factors
that help or impede collection of voluntary
maintenance funds from members of WMOs.
POLICY PURPOSEImprove water governance and the maintenance of the infrastructure for enhancing the productive uses of land
and water resources.
Some communities have been able to come together and collect fund for maintenance, while majority have not been able to do so. Why?
CONTENT
1. Motivation and background
2. Methodology
3. Descriptive statistics
4. Regression analysis and results
5. Concluding remarks, recommendations
MOTIVATION AND BACKGROUNDWater policy in Bangladesh
Before 60s
60s – 80s
• Protection by temporary and seasonal earthen.• Maintenance by the landlords (zamindars).• Voluntary labour from their tenants.
• Coastal Embankment Project (CEP)• No mention of participatory water management. • BWDB ‘khalashis’ responsible for managing and maintaining coastal embankments.
80s
• Involvement of communities in design and implementation of projects introduced. • Financial contribution towards maintenance not required.• Late 1980s, entry of LGED in the water sector.• Community contribution towards maintenance tested for the first time.
• Realization that regular upkeep of infrastructure is the Achilles heel of entire infrastructure investments.
• GoB enunciated community participation as its core principle of water management through its NWP (MoWR, 1999) and GPWM (MoWR, 2001).
• Requirement of financial contribution by the community for maintenance.
90s
MOTIVATION AND BACKGROUNDMaintenance situation
Perception of the infrastructures condition• 20% of the households consider the gates as being in good condition.• 15% of the households consider the canals as being in good condition.
Contributions• Both for LGED and BWDB data shows that maintenance funds always fall
to answer to the requirements.• 91% of the household did not contribute to maintain the gates in 2012.• 95 % of the household did not contribute to maintain the canals in 2012.
Polder 3
Polder 30
Polder 43-2F
Jabusha
Jainka
ti
Latab
uniaTo
tal0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
Very bad canal conditionVery bad gate condition
MOTIVATION AND BACKGROUNDInstitutional differences
• Theoretical and empirical research shows the importance of institutions in forging cooperative outcomes (Bardhan, 2005; Agrawal, 2001; North, 1990).
• Importance of institution in sustainable management of common property resources (Wade, 1988; Ostrom, 1990; Baland and Platteau, 1996).
• Entered the water sector in 1980s, culture of community participation was already well entrenched.
• Small Scale Water Resources Sector Development Project (SSWRDSP), phase I in 1994, now phases III and IV, funding support until 2017.
• WMCAs registered with the cooperative department.
• Communities contribute 4% of the capital cost of physical infrastructure.
• Maintenance funds , yearly audit statements.
• NGOs and extension agencies for implementing community participation
• Declining field presence.• WMGs or WMAs registered as rural
cooperatives since 2008.• No contribution of the WMO required at
initial stage.• Encouragement of the WMO for starting
maintenance fund and collecting subscriptions.
BWDB LGED
METHODOLOGYPublic good game
Purpose of the game• Designing a fictive situation to reproduce “real life”• Understanding the determinants of contribution to maintain a public good.• Understanding the behaviours: from cooperation to free-riding
Sample• Game played 18 times:
- Polder 3, polder 30, polder 31- Latabunia, Jabusha, Bagachra-Badurgachra
• 5 players per game 90 players• 30 rounds per game 2700 decisions
• Each player has to decide the allocation of a cash amount (20, 35/10) among a common fund and private fund.
• The incentive for contributing in the common fund is that if the fund reaches a certain threshold (50 or 95), a payment is added (25 or 75).
• The common pool is then distributed between the players.
The rules vary from one session to another to reflect real life conditions.
METHODOLOGYProcedure of the game
TREATMENTS C T1 T2 T3 T4Information No Yes Yes Yes YesInitial cash = = = ≠ ≠Gains distribution = = ≠ = ≠Threshold 50 50 50 50 50
DESCRIPTIVE STATISTICSAverage contribution per round
Control Treatment 1 Treatment 2 Treatment 3 Treatment 4810
1214
Ave
rage
indi
vidu
al c
ontri
butio
n
0 5 10 15 20 25Rounds
DESCRIPTIVE STATISTICSInformation effect
Control
No information
Treatment 1
Information
T-test of differenc
es(p-value)
Individual variables Individual contributions 11.448 9.442 (0.000)Individual gains 26.117 22.702 (0.000)Group variable Proportion of rounds with success 0.744 0.533 (0.003)Contribution standard deviation, within group
5.009 4.174 (0.020)
Control Round 5
No information
Treatment 1Round 6
InformationIndividual variables Individual contributions 12.277 10.233Individual gains 27.666 22.488
DESCRIPTIVE STATISTICSUnequal gain distribution effect
Treatment 1Equal gains distribution
Treatment 2Proportional
gains distribution
T-test of difference
s(p-value)
Individual variables Individual contributions 9.442 13.224 (0.000)Individual gains 22.702 27.842 (0.000)Group variable Proportion of rounds with success
0. 533 0.777 (0.000)
Contribution standard deviation, within group
4.174 4.005 (0.640)
Equal Endowment
DESCRIPTIVE STATISTICSInequalities in endowments effect
Equal Endowment
Treatment 1
Equal endowment
s
Treatment 3
Unequal endowmen
ts
T-test of differences(p-value)
Individual variables Individual contributions 9.442 10.208 (0.156)Individual gains 22.702 24.602 (0.010)Group variable Proportion of rounds with success 0.533 0.611 (0.294)Amount collected in the collective fund
47.211 51.044 (0.274)
Contribution standard deviation, within group
4.174 8.394 (0.000)
REGRESSION ANALYSISModel
Equal Endowment
x ijt=α1 Rt+α 2 I i+α 3G j+εijt
Individual contributionIndividual earning
Round characteristics• Game-rules variables• Past events from the game
(success, contributions, earning)
Group characteristics• Number of relatives and
friends• Heterogeneity of the group
(sex, religion, wealth)• Institutional context
Individual characteristics• Age, sex, religion • Level of education • Main source of income, land size • Participation and contribution
• Model estimated by OLS• Clustering at individual level for taking
care of unobserved characteristics of the individual.
REGRESSION ANALYSISEstimation strategy
Equal Endowment
RobustnessPanel analysis, with individual fixed effects
Consistency of the results
x ijt=α1 Rt+FEi j+εt
Learning effectEach player learns from the game, from the group he plays with and this learning also depend from his own background. • Individual, group level
• age, sex, education• clustering at individual level
• Game level: • continuous variable, number of round already played by the member• Game events from the 2 previous rounds: contribution, earning, failure
REGRESSION ANALYSISGame variables - Results
Equal Endowment
VARIABLES
(1)OLS
(3)OLS
Individual contribution Individual earning
Initial endowment 0.566*** 0.963*** (0.0348) (0.0270)Information -1.851*** -3.754*** (0.449) (0.871)Inequalities in endowments -2.497*** -3.400** (0.753) (1.425)Unequal sharing of the pot 1.204** 1.270 (0.508) (0.901)Previous round unsuccessful 0.249*** 0.423*** (0.0747) (0.144)Round, learning effect -6.277*** -7.717*** (0.559) (0.668)Observations 2,250 2,250R-squared 0.523 0.473
• Information has a negative and significant effect on the individual contribution as well as on the earning
• Endowment heterogeneity in the game design has a significant negative influence on the individual earning and contribution.
• Proportional distribution of the common fund has a significant and positive effect on the contributions.
REGRESSION ANALYSISIndividual variables - Results
Equal Endowment
VARIABLES(1)OLS
(3)OLS
Individual contribution Individual earningMain income from agriculture 2.364** -2.223** (0.955) (0.895)Main income from aquaculture 2.753*** -1.500* (0.917) (0.846)Sex, men -0.516 0.244 (0.988) (0.750)Religion, Muslim 3.768*** -1.711** (0.898) (0.833)Age 0.0539* 0.0312 (0.0321) (0.0336)Education level 0.319*** -0.0843 (0.101) (0.0812)WMCA, WMO member 0.683 -1.541** (0.678) (0.662)Contribution in maintenance fund -0.497 -1.400 (0.881) (0.976)Land size -0.00208** 0.00162 (0.000942) (0.00131)Observations 2,250 2,250R-squared 0.523 0.473
Players are drawing most of their income from agriculture or
aquaculture they are contributing more to
the common fund.
Age as well as the highest level of
education achieved determine positively and significantly the
individual contribution.
REGRESSION ANALYSISGroup variables - Results
Equal Endowment
VARIABLES(1)
OLS(3)OLS
Individual contribution Individual earningLGED sub-project 2.376*** 2.637*** (0.876) (0.831)Number of relative in the group -1.073** 0.641 (0.418) (0.475)Number of close friend in the group -0.312 0.718 (0.405) (0.450)Same religion within the group 0.767 -0.304 (0.902) (0.841)Standard deviation of land size 0.00177 0.00189 (0.00141) (0.00132)Group of men 0.765 1.452* (0.781) (0.731)Constant -7.898*** 7.467*** (2.250) (2.264)Observations 2,250 2,250R-squared 0.523 0.473
• Group composition in terms of gender, religion or wealth doesn’t have any significant effect on the individual contribution.
• The more a player is surrounded by relatives in his group, the less he contributes.
• Players from LGED villages are contributing higher amounts in the common fund than other players whatever are the individual, group and game characteristics.
CONCLUSIONMain results and recommendations
Equal Endowment
1. Principal users and beneficiaries of the infrastructures should be targeted first for contributing.
2. Homogeneous groups would contribute more and maintain their infrastructure better.
• But: How to create homogeneous groups in heterogeneous villages? • Solution: Membership conditions• Ex: In some WMOs, only landowners can be members.
3. Contributions are higher when there are related benefits.• But: In reality, benefits are not related to the contributions.• Solution: Introducing benefits for members, even if not related to water• Ex: In some WMOs, access to micro-credit for members, fishing rights…
4. Strong institutions support individual contributions for maintenance.
• Institutions created by the community itself• Involvement at the early stage of the project, create an ownership, a
willingness (and ability) to cooperate in the future.
Thank youYour questions, comments and suggestions
are welcome!
Top Related