Using CDD for Post-conflict reintegration: Lessons from the impact evaluation of the BRA-KDP Program in Aceh Adrian Morel, Conflict & Development Program, WB Indonesia Presentation to the Development Impact Evaluation Initiative (DIME) workshop Dubai, June 1st, 2010 Summary o Assistance to conflict victims through community-based approach increasingly used in post-conflict reintegration programs o CDD approach viewed as an effective mechanism for channeling such support: o Ensure the right people benefit (targeting) o Funds are effectively used (transparency, accountability) o Help repair the social fabric (participation, inclusion) o Based on assumption that CDD may not only achieve welfare outcomes but also lead to improvements in social and state-society relations o The BRA-KDP program (2006-2007) delivered US$ 20.4 million in assistance to conflict victims in 1,724 villages (one third of Aceh) o This presentation is based on a large-scale mixed methods evaluation of the program. o Main findings: o BRA-KDP successful in reaching a large number of conflict-affected people (over 230,000) o Substantial welfare impacts o Less successful in improving social relations and building trust in state Reintegration in Aceh o Six months after the Dec 26th 2004 tsunami, the Helsinki MoU peace accord (August 2005) ended 30 years of separatist conflict between the Free Aceh Movement (GAM) and the Government of Indonesia (GoI): o 30,000 deaths; over 400,000 displaced o US$ 10.7 billion in damage and loss; 50% of rural infrastructure damaged or destroyed o 1.5 million Acehnese or 39% of the population consider themselves conflict victims o Government Reintegration Fund (US$ 150 million) set up to help combatants, prisoners and conflict-affected groups o The Aceh Reintegration Board (BRA) established to manage funds and programs – Supervision by National Planning Agency (Bappenas) Community-Based Reintegration Assistance for Conflict Victims (BRA-KDP) o Genesis of the BRA-KDP program: o Initial attempts to individually target victims problematic – BRA lacked capacity and proper mechanisms to verify claims o Change to use of Kecamatan Development Program (KDP) which already operated in every village in Aceh o Targeting devolved to community-level, with safeguards to ensure funds are used effectively in a transparent and accountable manner o BRA-KDP mechanisms: o Village meetings, facilitated by KDP staff, decided on local criteria for who was a victim and how funds should be spent o Open menu: communities can decide whether to finance public infrastructure or economically beneficial private goods o Ex-combatants not eligible (they receive targeted support from other BRA program) Selecting Sub-districts to Receive BRA-KDP • Program aimed to target the most conflict-affected sub-districts. Combined several variables into an index of conflict affectedness • Only sub-districts that surpasses a spending capacity threshold (60 percent disbursement of 2005 KDP funds) were eligible. Within Sub-district Allocations • Within selected sub-districts, all villages get a block grant allocation • Grant sizes depend on conflict intensity (sub-district level) and village population Community-Based Reintegration Assistance for Conflict Victims (BRA-KDP) o From 2006-2007, US$ 20.4 million channeled to 1724 villages in 67 sub-districts (one-third of Aceh) o BRA-KDP objectives: providing assistance to civilians most affected by the conflict in ways that improve their well-being, enhance relations between groups, build trust in Gvt Questions o How successful was BRA-KDP in improving the welfare of conflict victims, improving social relations and building trust in the state? o What does the BRA-KDP experience tell us about how CDD can be used for reintegration in other post-conflict areas? Learning from BRA-KDP: Data Sources o Primary data source: the Aceh Reintegration and Livelihoods Survey (ARLS) o Administered in all 67 treatment sub-districts and 68 matched control subdistricts, one year after program completion (2008) o Interviews with 2,150 randomly selected households (5 per villlage), 460 village heads and around 1,000 ex-combatants o Statistical strategies for ex-post matching of treatment and control subdistricts, and account for systematic differences between treatment and control areas (sampling weights to account for variation in village and household size; clustering at level of treatment) --- this allows for accurate estimation and attribution of program impacts o Design and data analysis by Macartan Humphreys (Columbia), Jeremy Weinstein (Stanford), Laura Paler (Columbia), Yuhki Tajima (Harvard) with help from C&D team o Field Implementation by AC Nielsen Learning from BRA-KDP: Data Sources o Management Information System (MIS): Designed to collect info on program progress, participation rates, proposal recapitulation, conflict victims maps, complaints, financial flows o Supervision missions and facilitator surveys PROGRAM IMPLEMENTATION Reaching Conflict Victims o A large number of people benefited from BRA-KDP: 233,282 direct beneficiaries (or 22% of the overall population of target sub-districts). o Challenge: widespread perception that everyone in Aceh is a conflict victim to some extent – tension between a tendency to spread assistance broadly to satisfy everyone and avoid creating disputes, and the need to prioritize the most-affected Reaching Conflict Victims o Multiple levels for trying to reach conflict victims: o Selecting most conflict affected sub-districts (but conditional on spending capacity and plans for a second round of programming) o Villages in more conflict affected areas receive more o Prioritizing conflict victims within villages, although others can also benefit (community decision-making) o Throughout analysis, we look at whether treatment effects are different for conflict-affected and most conflict-affected. These were based primarily on self-reported answers to the survey Targeting within Villages Communities used wide variety of criteria and approaches for deciding who was a victim and who should benefit (1) Degree of conflict-affectedness Asong Tongpeuding village, Titeu Keumala sub-district, Pidie • Four categories of conflict victims were decided upon, each receiving a different amount of assistance • High: people whose house was burned (Rp. 3.5 million/HH); Medium: people who had a death in the family (Rp. 2 million/HH); Low: people who were tortured (Rp. 1 million/HH); common community members (Rp. 500,000 /HH) (2) Degree of conflict-affectedness and economic need Meriah Jaya village, Timang Gajah sub-district, Bener Meriah • Out of 322 HHs, 167 were selected as beneficiaries. Beneficiaries were divided into three groups: • HHs who suffered a death in the family, or had houses and fields destroyed; HHs who suffered trauma or who were unable to tend fields; Poor HHs that did not possess land or did not have money to purchase food • The amount of assistance was scaled with the first group receiving the largest amount of assistance (3) Degree of conflict-affectedness and prior receipt of economic assistance Timanang Gading village, Kabayakan sub-district, Aceh Tengah • Twelve out of 49 IDPs were prioritized as BRA-KDP beneficiaries • Prioritized because they had not received assistance and were severely affected • Each beneficiary received Rp. 2.5 million. The remaining block grant funds were used for community projects. Prioritizing Conflict Victims • Conflict victims were more likely to be benefit from BRA-KDP than non-victims. – 54 percent of conflict victims were in treated areas, and 44% of these got benefits. The probability of benefitting for victims was 24% – 41 percent of non-victims were in treated areas and 40 percent of those got benefits. The probability of non-victim benefitting was 16% • Conflict victims got more Figure3.3: Average amounts of assistance perceived per victimhood category (.000 Rp) 700 600 500 400 300 569 629 668 200 100 0 Non-victims Victims Most-Affected Victims Perceptions of the Program The program was very popular: • 96% of villagers & 98% of most-affected thought BRA-KDP helpful • A useful injection of capital; provided victims with a sense that their suffering was acknowledged and compensated • Only 12% of conflict victims thought they did not benefit enough PROGRAM IMPACTS Welfare Impacts: Poverty BRA-KDP resulted in an 11 percent decrease in the share of households in the village classified as poor Welfare Impacts: Assets Index of 2008 asset holdings TABLE 1: ASSET INDEX Individuals in Individuals in control treatment communities communities (N) (N) Difference OLS (se) Difference IV (se) All 0.22 (1225) 0.07 (1090) -0.16** (0.07) 0.04 (0.12) Conflict victims 0.04 (455) -0.01 (528) -0.05 (0.09) 0.34** (0.17) Most conflict-affected 0.13 0.04 -0.09 0.43 (282) (269) (0.12) (0.26) *** Significant at 99%; ** Significant at 95%; * Significant at 90%. The table reports population average responses (with sample N’s below, where total sample size is 2315), and the difference for populations in treatment and control communities using least squares and instrumental variable regressions. All regressions control for conflict and spending capacity, their quadratic and cubed terms, and their interaction. Question: How many of the following things do you or a member of your household possess? • • • Gains for conflict victims in overall asset holdings Largest difference was for motorbikes Little evidence that the project directly funded motorbikes. Program participants used money they generated from extra economic opportunities to buy bikes. Welfare Impacts: Land Use 2 m of land that is farmed by household TABLE 1: LAND USE Individuals in Individuals in control treatment communities communities (N) (N) Difference OLS (se) Difference IV (se) All 7740.13 (644) 9437.52 (617) 1697.4 (2855.43) 12200.81 (7940.23) Conflict victims 6905.98 (245) 7043.51 (297) 137.53 (1114.56) 7590.92*** (2174.80) 8215.34 7607.05 (152) (200) *** Significant at 99%; ** Significant at 95%; * Significant at 90%. 2 Question: How many m of land is being farmed by this household? -608.29 (1614.86) 7382.25*** (2774.70) Most Conflict-affected Conflict victims in treatment communities have more than 7500 sq-meters of additional land cleared as a result of the program (a doubling of land use) Welfare Impacts: Subjective Perceptions Conflict victims 18% more likely to report improvements in welfare in project areas than control areas Welfare Impacts: Other Measures o Weaker or no improvements found with respect to: o Employment o Health and education outcomes o Availability of amenities o A possible result of: o Choice of villagers for private goods o One-round program – needs more sustained investments Social Impacts • Acceptance of returning groups, reported social tensions, measures of collective action capacity / involvement in associational life: similar between treatment and control areas • No consistent evidence of impacts – positive or negative – of BRA-KDP on attitudes towards government / state • These findings contrast with other research on KDP in Indonesia: positive impacts on various measures of social cohesion • Might result from BRA-KDP being one-off and spent on private goods – less opportunities for cohesion-building collective action Social Impacts: Acceptance of Former Combatants • Communities in project areas more accepting of ex-combatants BUT … conflict victims significantly less likely to accept former combatants after program implementation • Reasons: Ex-GAM extorted funds from BRA-KDP, leading to higher levels of non-acceptance; BRA-KDP empowered non-GAM groups in society, enabling them to voice their non-acceptance of GAM CONCLUSIONS Conclusion 1: Strong evidence of welfare impacts o Strong evidence on welfare gains o o o o 11% reduction in share of poor households Doubling of land use for victims Reported improvements in welfare Asset increases o Shows CDD projects can be an effective mechanism for : o rapid compensation in ways hat have significant welfare impacts in post-conflict environments o delivering private goods o Scope for post-conflict ‘compensation’ projects that have developmental impacts in other conflict-affected areas Conclusion 2: Social cohesion impacts o Some negative impacts on acceptance of former combatants: o Failure in reintegration programs targeted at ex-combatants had a negative spillover effect on BRA-KDP o Problems of separating victims and combatants: Ex-GAM more likely to cause problems in BRA-KDP villages where they were excluded from receiving benefits. Lower levels of conflict in BRA-KDP locations where GAM were among the beneficiaries. o Need to better link CDD with other post-conflict programs / possibly best to have same programs for ex-combatants and civilians rather than separate ones o CDD not a “silver bullet” peace-building solution. It does not inevitably results in enhanced social cohesion, in particular when programs have a limited duration. Recommendations • Continue to experiment with CDD as a mechanism for reintegration, gather information on what works and what does not, where and why in order to inform the design of future post-conflict interventions • In post-conflict areas where relations between former combatants and civilians are relatively good, deliver assistance to both groups altogether using the same mechanism • Deliver assistance over multiple rounds to maximize social impacts Full Reports o Adrian Morel, Makiko Watanabe, Rob Wrobel (2009), “Delivering Assistance to Conflict-Affected Communities: The BRA-KDP Program in Aceh”. ISDP No. 13. Jakarta, WB o Patrick Barron, Macartan Humphreys, Laura Paler and Jeremy Weinstein (2009), “Community-Based Reintegration in Aceh: Assessing the Impacts of BRAKDP”. ISDP No. 12. Jakarta, WB Available online at: www.conflictanddevelopment.org Thanks!