institutional competition and governance quality: experimental evidence from afghanistan
DESCRIPTION
NES 20th Anniversary Conference, Dec 13-16, 2012 Institutional Competition and Governance Quality: Experimental Evidence from Afghanistan (based on the article presented by Ruben Enikolopov at the NES 20th Anniversary Conference). Authors: Andrew Beath, Fotini Christia, Ruben EnikolopovTRANSCRIPT
Institutional Competition and Governance Quality Experimental Evidence from Afghanistan
Andrew Beath Fotini Christia Ruben Enikolopov
Motivation } Longstanding academic inquiry on the relationship between
institutions and economic and political development.
} Presumed but largely unproven causal effect of specific political institutions.
} In particular, does the creation of representative institutions produce more equitable policy outcomes and reduces corruption?
} Use a randomized field experiment to assess whether democratically elected local institutions improve quality of governance
Intervention } Part of the bigger project based on impact evaluation of National
Solidarity Programme in Afghanistan } Creation of democratically elected village councils
} Food aid distribution as an example of a task performed by local leaders which is } Relevant } Measureable } Comparable across villages
} Outcomes of interest: } the quality of targeting } the extent of corruption
HypothesesTargeting
H1: Democratically Elected Institutions lead to improved Objective Targeting
H2: Democratically Elected Institutions lead to improved Subjective Targeting
Elite Capture H3: Democratically Elected Institutions lead to a decrease in
Embezzlement H4: Democratically Elected Institutions lead to a decrease in Nepotism
Participation H5: Democratically Elected Institutions lead to an increase in
Participation
Treatment
} The National Solidarity Program (NSP) is the largest development program in Afghanistan.
} Over 29,000 of Afghanistan’s 38,000 villages have received NSP. } Sponsored by international donors and run by the Ministry of
Rural Rehabilitation and Development. } Implemented by NGOs in two main stages:
} Election of Community Development Councils (CDCs) through secret-ballot election } Size of CDC proportional to size of community with equal
number of males and females. } Project Selection
} $200 per household; max community grant $60,000
Herat Ghor
Daykundi
Balkh
Baghlan
Nangarhar
500 villages across 10 districts in 6 provinces. Cluster Matched-Pair used to assign villages to control and treatment.
Ø Customary structure: Headman; Tribal Elders; and Clergy;
Ø Purdah precludes female participation;
Ø Democratic councils exist in parallel to customary structures
Ø Official responsibilities limited to managing development projects
Ø But in some cases they assume responsibilities for conflict resolution etc.
Ø Some overlap between customary leaders and elected members:
Ø 40% of elected members identified as village leaders ex ante
Ø elected members are on average younger and better educated.
Ø In partnership with WFP, 725 mt. of wheat distributed across sample in fall 2011
1st Visit: Local leaders informed of distribution and asked to prepare list of recipients;
2nd Visit: Food aid delivered by WFP; recipient list retrieved;
3rd Visit: Male and female surveys administered: (i) random sample of village households; (ii) listed recipients; (iii) unlisted recipients; Data collected on wheat distribution (process and amounts) as well as recipient and non-recipient characteristics
Mechanisms
} Democratically elected gender balanced councils can affect distribution outcomes through different channels: } 1) Direct Effect: Council assumes responsibility of distribution.
} 2) Indirect Effect: Council affects distribution by affecting the behavior of customary village leaders in those villages that now also have elected councils.
} 3)The results of the aid distribution can be affected by higher female involvement in the distribution.
Ø To get at underlying mechanisms, we vary the identity of people responsible for the food aid distribution.
Sample (500 Villages)
Democratic Council (250 Villages)
Council Distributes (125 Villages)
“Leaders” Distribute (125 Villages)
No Democratic Council (250 Villages)
“Leaders” Distribute (125 Villages)
“Leaders” & Women Distribute (125 Villages)
Effect of Creating Democratic Councils
Effect of Involving Women Effect of Creating & Designating Democratic Councils
Outcomes Council Exists
(But Not Delegated) Council Manages
Distribution Women Involved in
Distribution Obs.
Objective Targeting (H1) Asset Ownership of Food Aid Recipients -0.043 [0.096] -0.023 [0.097] -0.025 [0.094] 8,159 Whether Food Aid Recipients Belong to Vulnerable Group -0.003 [0.020] 0.054** [0.024] 0.006 [0.021] 8,767 Omnibus Measure of Need 0.0299 [0.143] -0.1611 [0.121] -0.0303 [0.116] 7,170 Mean Effect Index for H1 0.001 0.062* 0.044
Subjective Targeting (H2) Share of respondents mentioning recipient as vulnerable ex-ante 0.002 [0.003] -0.001 [0.002] 0.002 [0.003] 3,262 Proportion of Recipients Reported as Vulnerable Ex-Post -0.008 [0.013] -0.013 [0.008] -0.025** [0.012] 5,966 Distribution Primarily Benefited Vulnerable Households -0.083 [0.105] 0.188* [0.100] 0.001 [0.143] 6,345 Mean Effect Index for H2 -0.002 -0.003 -0.013
Embezzlement (H3) Wheat Retained by Village Leaders 0.038* [0.020] -0.011 [0.020] 0.053** [0.024] 6,112 Some Wheat was Sold by Village Leaders 0.006 [0.012] 0.013 [0.008] 0.026** [0.011] 6,968 Wheat Revoked by Village Leader safter Distribution -0.009 [0.012] -0.016 [0.010] -0.022 [0.014] 5,620 Difference btw. Amount Allocated and Amount Received 5.653** [2.529] 0.155 [2.259] 5.312* [2.838] 3,803 Mean Effect Index for H3 -0.099* 0.008 -0.105*
Nepotism (H4) Proportion of Recipients Connected to Village Leaders, Self-Identified 0.004 [0.032] -0.030 [0.027] -0.013 [0.028] 0.004 Proportion of Recipients Connected to Village Leaders, Reported -0.020 [0.033] 0.045 [0.035] 0.030 [0.046] 5,993 Distribution Primarily Benefited Connected Households 0.024 [0.016] -0.013 [0.014] 0.004 [0.019] 7,049 Non-Decision-Making Local Leader(s) Received Wheat -0.008 [0.013] -0.008 [0.013] -0.015 [0.015] 6,531 Mean Effect Index for H4 (Listed Recipients) -0.022 0.005 -0.010
Process (H5) Decision Made by Multiple People -0.012 [0.028] 0.031 [0.019] -0.007 [0.027] 13,011 Ordinary Villagers Participated in Decision Making Process -0.113** [0.044] 0.008 [0.036] -0.090* [0.046] 10,440 Number of People (ln) Involved in Decision-Making Process -0.139** [0.060] 0.120** [0.051] -0.039 [0.072] 12,317 Respondent Involved in Decision-Making Process -0.032** [0.012] 0.018 [0.015] -0.016 [0.014] 15,980 Women Involved in Decision-Making Process -0.004 [0.010] 0.005 [0.010] 0.001 [0.012] 14,549 No Conflicts Related to Wheat Distribution 0.000 [0.009] -0.019*** [0.006] -0.006 [0.008] 13,415 Decision Publicly Announced 0.028 [0.031] 0.019 [0.024] 0.014 [0.039] 13,570 Mean Effect Index for H5 -0.066* 0.046 -0.015
Outcomes Council Exists
(But Not Delegated) Council Manages
Distribution Women Involved in
Distribution Obs.
Objective Targeting (H1) Asset Ownership of Food Aid Recipients -0.043 [0.096] -0.023 [0.097] -0.025 [0.094] 8,159 Whether Food Aid Recipients Belong to Vulnerable Group -0.003 [0.020] 0.054** [0.024] 0.006 [0.021] 8,767 Omnibus Measure of Need 0.0299 [0.143] -0.1611 [0.121] -0.0303 [0.116] 7,170 Mean Effect Index for H1 0.001 0.062* 0.044
Subjective Targeting (H2) Share of respondents mentioning recipient as vulnerable ex-ante 0.002 [0.003] -0.001 [0.002] 0.002 [0.003] 3,262 Proportion of Recipients Reported as Vulnerable Ex-Post -0.008 [0.013] -0.013 [0.008] -0.025** [0.012] 5,966 Distribution Primarily Benefited Vulnerable Households -0.083 [0.105] 0.188* [0.100] 0.001 [0.143] 6,345 Mean Effect Index for H2 -0.002 -0.003 -0.013
Embezzlement (H3) Wheat Retained by Village Leaders 0.038* [0.020] -0.011 [0.020] 0.053** [0.024] 6,112 Some Wheat was Sold by Village Leaders 0.006 [0.012] 0.013 [0.008] 0.026** [0.011] 6,968 Wheat Revoked by Village Leaders after Distribution -0.009 [0.012] -0.016 [0.010] -0.022 [0.014] 5,620 Difference btw. Amount Allocated and Amount Received 5.653** [2.529] 0.155 [2.259] 5.312* [2.838] 3,803 Mean Effect Index for H3 -0.099* 0.008 -0.105*
Nepotism (H4) Proportion of Recipients Connected to Village Leaders, Self-Identified 0.004 [0.032] -0.030 [0.027] -0.013 [0.028] 0.004 Proportion of Recipients Connected to Village Leaders, Reported -0.020 [0.033] 0.045 [0.035] 0.030 [0.046] 5,993 Distribution Primarily Benefited Connected Households 0.024 [0.016] -0.013 [0.014] 0.004 [0.019] 7,049 Non-Decision-Making Local Leader(s) Received Wheat -0.008 [0.013] -0.008 [0.013] -0.015 [0.015] 6,531 Mean Effect Index for H4 (Listed Recipients) -0.022 0.005 -0.010
Process (H5) Decision Made by Multiple People -0.012 [0.028] 0.031 [0.019] -0.007 [0.027] 13,011 Ordinary Villagers Participated in Decision Making Process -0.113** [0.044] 0.008 [0.036] -0.090* [0.046] 10,440 Number of People (ln) Involved in Decision-Making Process -0.139** [0.060] 0.120** [0.051] -0.039 [0.072] 12,317 Respondent Involved in Decision-Making Process -0.032** [0.012] 0.018 [0.015] -0.016 [0.014] 15,980 Women Involved in Decision-Making Process -0.004 [0.010] 0.005 [0.010] 0.001 [0.012] 14,549 No Conflicts Related to Wheat Distribution 0.000 [0.009] -0.019*** [0.006] -0.006 [0.008] 13,415 Decision Publicly Announced 0.028 [0.031] 0.019 [0.024] 0.014 [0.039] 13,570 Mean Effect Index for H5 -0.066* 0.046 -0.015
Outcomes Council Exists
(But Not Delegated) Council Manages
Distribution Women Involved in
Distribution Obs.
Objective Targeting (H1) Asset Ownership of Food Aid Recipients -0.043 [0.096] -0.023 [0.097] -0.025 [0.094] 8,159 Whether Food Aid Recipients Belong to Vulnerable Group -0.003 [0.020] 0.054** [0.024] 0.006 [0.021] 8,767 Omnibus Measure of Need 0.0299 [0.143] -0.1611 [0.121] -0.0303 [0.116] 7,170 Mean Effect Index for H1 0.001 0.062* 0.044
Subjective Targeting (H2) Share of respondents mentioning recipient as vulnerable ex-ante 0.002 [0.003] -0.001 [0.002] 0.002 [0.003] 3,262 Proportion of Recipients Reported as Vulnerable Ex-Post -0.008 [0.013] -0.013 [0.008] -0.025** [0.012] 5,966 Distribution Primarily Benefited Vulnerable Households -0.083 [0.105] 0.188* [0.100] 0.001 [0.143] 6,345 Mean Effect Index for H2 -0.002 -0.003 -0.013
Embezzlement (H3) Wheat Retained by Village Leaders 0.038* [0.020] -0.011 [0.020] 0.053** [0.024] 6,112 Some Wheat was Sold by Village Leaders 0.006 [0.012] 0.013 [0.008] 0.026** [0.011] 6,968 Wheat Revoked by Village Leaders after Distribution -0.009 [0.012] -0.016 [0.010] -0.022 [0.014] 5,620 Difference btw. Amount Allocated and Amount Received 5.653** [2.529] 0.155 [2.259] 5.312* [2.838] 3,803 Mean Effect Index for H3 -0.099* 0.008 -0.105*
Nepotism (H4) Proportion of Recipients Connected to Village Leaders, Self-Identified 0.004 [0.032] -0.030 [0.027] -0.013 [0.028] 0.004 Proportion of Recipients Connected to Village Leaders, Reported -0.020 [0.033] 0.045 [0.035] 0.030 [0.046] 5,993 Distribution Primarily Benefited Connected Households 0.024 [0.016] -0.013 [0.014] 0.004 [0.019] 7,049 Non-Decision-Making Local Leader(s) Received Wheat -0.008 [0.013] -0.008 [0.013] -0.015 [0.015] 6,531 Mean Effect Index for H4 (Listed Recipients) -0.022 0.005 -0.010
Process (H5) Decision Made by Multiple People -0.012 [0.028] 0.031 [0.019] -0.007 [0.027] 13,011 Ordinary Villagers Participated in Decision Making Process -0.113** [0.044] 0.008 [0.036] -0.090* [0.046] 10,440 Number of People (ln) Involved in Decision-Making Process -0.139** [0.060] 0.120** [0.051] -0.039 [0.072] 12,317 Respondent Involved in Decision-Making Process -0.032** [0.012] 0.018 [0.015] -0.016 [0.014] 15,980 Women Involved in Decision-Making Process -0.004 [0.010] 0.005 [0.010] 0.001 [0.012] 14,549 No Conflicts Related to Wheat Distribution 0.000 [0.009] -0.019*** [0.006] -0.006 [0.008] 13,415 Decision Publicly Announced 0.028 [0.031] 0.019 [0.024] 0.014 [0.039] 13,570 Mean Effect Index for H5 -0.066* 0.046 -0.015
Summary of Findings
} Differential effect of democratically elected institutions } When in the lead - better targeting of aid } When in competition with traditional elite structures - higher levels
of embezzlement, and lower levels of participation.
} Institutional competition is also at play when we mandate female participation.
} Creation of multiple institutional structures with no clear division of responsibilities can lead to institutional competition and underperformance rather than to additional checks and balances that enhance efficiency.