World Bank Annual Conference on Land and Poverty, April 19-20, 2011 Securing Rural Land Rights: Experimental Evidence from the Plans Fonciers Ruraux in Benin Harris Selod (team leader) Klaus Deininger Markus Goldstein Kenneth Houngbedji Déo-Gracias Houndolo Florence Kondylis Michael O Sullivan This work has been undertaken in collaboration with MCC & MCA Benin. 1 Funding from AADAPT/DIME, BPRP/LPRP, BNPP and GAP is also gratefully acknowledged.
The PFR approach Policy to consolidate land rights in rural areas Introduces written documentation of rights (Certificat Foncier Rural - CFR) Stated objectives: improve tenure security of landholdings and stimulate agricultural investment Originality: recognizes existing customary land rights Benin currently has two PFR programs Large one under MCC/MCA (implemented by GTZ-IS) Smaller one in the North (ProcGRN) 2
The Benin context Low investment in land Tenure insecurity Customary law typically holds sway Thin rural credit markets Emerging land markets Conflicts over land Farmers vs. pastoralists, farmers vs. farmers Inheritance 3
The Benin context (cont.) Unequal access to land Women often cannot inherit land and rely on husbands for access to land Evidence from other African countries suggests that insecure tenure limits women s investment (Udry 1996), productivity (Goldstein & Udry 2008) and land market participation (Holden et al 2011) Marginalized groups (migrants and pastoralists) also face challenges 4
The PFR process in Benin Participatory process to establish written documentation of land rights with legal recognition Main steps in each village: 1. Information campaign 2. Preparation of village profiles 3. Socio-legal inquiry 4. Parcel surveying and mapping of land use plan 5. Temporary recording of rights and rights holders 6. Public review of village land use plan (60 days) 7. CFR delivery and facilitation of formal, written records of secondary land rights 5
Example of a PFR 6
Impact evaluation research topics Effect on tenure security (reduction or reactivation of conflits dormants?) Changes in land market participation & prices? Effect on investments in land, production & yields? Labor effects? (incl. off farm activities) Possible differential effects on men and women? 7
Measuring program impact 300 village PFRs in 40 of Benin s 77 communes Random selection of villages that submit a proposal and meet eligibility criteria agricultural production, poverty but economic opportunities, presence of land conflict, willingness to promote women s access to land, rural Villages are selected through commune-level lotteries, establishing clearly defined PFR treatment villages and non-pfr control villages Phased-in implementation (ongoing) 8
PFR locations 9
The data: EMICoV Nationally representative panel household and community survey (EMICoV 2006 and 2010) with a large intersection with PFR villages EMICoV 2010 extended to more treatment and control villages EMICoV panel wave Treatment villages Control villages TOTAL 2006 & 2010 98 71 169 2010 only 103 35 138 Total 201 106 307 Total with withincommune matched pairs 194 99 293 10
The data: WB survey A 3,500 HH survey designed by the World Bank and linked to EMICoV (data collection just finished) + community survey Justification Pre-program data Longitudinal data (effects take time) Baseline for some villages (due to phased-in implementation) Very detailed plot-level info on land and agriculture 11
The data: WB survey (cont.) Preliminary plot-level data from the World Bank 2011 household survey. Plots will be linked with administrative data to compare PFR landholdings with agricultural parcels. 12
Impact evaluation challenges Several agro-climatic zones Complex and heterogeneous tenure situations (Lavigne-Delville, 2010) Distinction between agricultural plots and land parcels (and plot definitions across surveys) GPS measurements of plots (tracks and waypoints) and linking to program data Identification of households (EMICoV sampled on enumeration areas but program implemented at village level) 13
Preliminary conclusions EMICoV 2006 data suggest a pre-program balance across treatments and controls EMICoV 2010 data reveal: some observable differences between treatments and controls in simple mean comparisons but 2010 differences tend to disappear when commune and EMICoV 2006 controls are included Justifies follow-up survey waves 14
Balance test: Community, EMICoV 2010 Variables Sample size T-test Treated Primary school 271 0.8644 0.013-0.003 Local market 271 0.2655 0.042 0.030 microfinance 271 0.1017 0.038 0.040 Power supply 267 0.1314-0.075-0.044 Water network 271 0.0904-0.005 0.001 Water pump 271 0.7627 0.039 0.026 Paved road 269 0.1143-0.056-0.033 Laterite road 269 0.7600 0.037 0.008 Land line 271 0.0226-0.009-0.004 Cell phone 271 0.9322 0.007-0.014 OLS Treated (1)-(0) Significance Significance * p < 0:10, ** p < 0:05, *** p < 0.001 15
Balance test: Individuals, EMICoV 2010 Variables Sample size T-test Treated Age 23,853 20.7882 0.117 0.106 Ethnicity: Adja 23,855 0.1745 0.002-0.030 Ethnicity: Bariba 23,855 0.1804 0.069*** 0.042 Ethnicity: Fon 23,855 0.3876-0.067*** -0.012 Ethnicity: Peulh 23,855 0.0891-0.025*** -0.040 Ethnicity: Yoruba 23,855 0.1351 0.021*** 0.034 Can write French 23,855 0.2763-0.025*** -0.009 Illiterate 23,855 0.4740 0.024*** 0.014 Relig. indig 23,855 0.1823 0.012** 0.011 Christian 23,855 0.1682-0.025*** -0.007 Muslim 23,855 0.1845 0.003-0.013 Migrant 23,855 0.0822-0.000-0.000 Primary education 23,855 0.2653-0.009 0.006 Poverty 23,855 0.4834-0.069*** -0.108** OLS Treated (1)-(0) Significance Significance * p < 0:10, ** p < 0:05, *** p < 0.001 16
Balance test: Plots, EMICoV 2010 T-test OLS Variables Sample size Treated Treated (1)-(0) Significance Significance Land-use agreement 5,485 0.3977-0.043 ** -0.059 Land title 5,485 0.0046 0.001 0.001 Land lease 5,485 0.0107-0.003 0.005 Land permit 5,485 0.0008-0.004 ** -0.005 Land-sales agreement 5,485 0.1108 0.024 ** 0.020 Can rent out 5,485 0.5181 0.027 * 0.036 Can sell 5,485 0.4426 0.016 0.032 Can mortgage 5,485 0.4354 0.034 ** 0.047 Can bequeath 5,485 0.7815 0.021 * 0.011 Land bought 5,485 0.1209 0.028 ** 0.021 Land inherited 5,485 0.6723-0.040 ** -0.047 Land under conflict 5,485 0.0563-0.001-0.006 * p < 0:10, ** p < 0:05, *** p < 0.001 17
Going forward Forthcoming data analysis (plots, HH and communities) Analysis can provide policy lessons for further scale-up (dialogue with authorities) Need for further wave(s) of data collection to track the PFR effects over time (2012 or 2013 envisioned) Possibility for experimentation of program variants? 18
Annex 1: Land right types 19
Annex 2: Example of a socio-legal inquiry card 20
21
Annex 3: Example of a CFR (full document) 22
23
24
25
26
27
Annex 4: Rental contract template 28
Annex 5: Balance test: Individuals, EMICoV 2006 Variables Sample size T-test Treated female 12745 0.5129 0.001 0.008 age 12742 20.5177 0.554* 0.485 eth_adja 12252 0.1841 0.001-0.026 eth_bariba 12745 0.1035 0.019*** 0.035 eth_fon 12745 0.4725-0.059*** -0.027 eth_peulh 12745 0.0729-0.018*** -0.024 eth_yoruba 12745 0.1338 0.040*** 0.031 relig_indig 12745 0.2603-0.003 0.011 relig_christ 12745 0.2719 0.013 0.014 relig_musli 12745 0.1724 0.003-0.007 Migrant 12745 0.1585-0.010-0.014 OLS Treated (1)-(0) Significance Significance edu_primar 12745 0.2571-0.002-0.001 edu_middle 12745 0.0505 0.003 0.007 edu_high 12745 0.0086 0.001 0.002 edu_unive 12745 0.0004-0.001* -0.001* 29
Balance test: Plots, EMICoV 2006 Variables Sample size T-test Treated lndtitle 3900 0.0103-0.001-0.003 duraccess 3860 13.5358-0.636-1.046 lndbought 3900 0.1414 0.014 0.020 lndinheri 3900 0.6152 0.062*** 0.048 lndshcropin 3900 0.0198-0.024*** -0.017 lndrentin 3900 0.0831-0.019** -0.029 lndrentout 3900 0.0018 0.000-0.001 lndfallow 3900 0.0256-0.014** -0.009 lndshcropout 3900 0.0022 0.002* 0.003 lndconfl 3900 0.0162 0.002 0.003 conflsetl 3900 0.0157 0.005 0.005 customsetl 3900 0.0036-0.002-0.002 OLS Treated (1)-(0) Significance Significance 30