Impacts of land registration: Evidence from a pilot in Rwanda Daniel Ali; Klaus Deininger; Markus Goldstein Motivation: Do land rights matter? Insecure rights can lower productivity Goldstein and Udry, Ghana (2008) Registering land rights does not increase productivity Bardhan and Mookherjee, West Bengal (2009) Quality of land matters Registering land rights does increase investments Deininger and Ali; Holden, et al., Ethiopia (2011) Registering land rights might increase productivity, but it is not cost effective Jacoby and Minten, Madagascar (2007) 1
Impact of registration Channel for impacts to materialize Tenure security & investment Transferability (if there are gains to trade) Reduction of conflict (one-time or longer term) Formalization & credit supply (foreclosure, coverage) Land grabbing/elite capture possible if not transparent This paper aims to evaluate impact of the land tenure regularization program in the pilot cells Outline Land tenure in Rwanda Program characteristics Sample & estimation strategy Data & descriptive statistics Econometric results Conclusions 2
Land in Rwanda Land scarcity, dependence on agriculture Highest pop. density in Africa Average parcel size =0.35 ha, significant variation around this Environmental degradation; need for investment Continued fragmentation; active land market Contributing factor to 94 Genocide Land in Rwanda New legislation 1999 inheritance legislation: Equal rights to females 2004 land policy based on broad consultation: General principles 2005 OLL Recognizes existing (customary) rights, formalizes these Equality for spouses; registration compulsory Establishes institutional infrastructure (NLC, DLBs, LCs at cell, sector, dist. Level) Regulates expropriation & registration 3
Land in Rwanda Towards a national program Development of participatory, low-cost methodology based on photomaps Fine-tuning of procedures in 4 trial cells reflecting diversity of tenure situations 2006/7 Launch of nation-wide program in 2009/10 Currently 4 out of 11 mn. parcels registered Baseline IE survey completed, data entry ongoing Outline Land tenure in Rwanda Program characteristics Sample & estimation strategy Data & descriptive statistics Econometric results Conclusions 4
Explaining process and map Field adjudication with neighbors 5
Locating parcels on the index map Processing claims receipts 6
One of the cells in Kigali after digitisation Corrections and Objections phase: digitised maps and lists of parcels and their owners are displayed at a public place (cell offices) and the public is invited to inspect and ask for corrections or register objections to claims made 7
During Corrections and Objections, the adjudication committe records any changes admitted in red ink in the claim register and this is used to update the database Checking Room crosschecking that what was entered in the database is what is in the claim registers 8
Pilot registration: program basics Cells (4 of 2,146) were selected based on differing tenure and land use situations Gatsata (Kigali, urban) High expropriation risk Interspersed with small agricultural plots Kabushenge (North, rural) Polygamy a major issue Biguhu (West, rural) Severely affected by genocide Mowga (East, rural) Past practice of land sharing leads to insecurity Boundaries well marked Pilot registration: program basics 14,908 parcels registered, total area of 3,448 ha. Low tech, low cost cost per parcel less than $5 Comprehensive and compulsory title registration Claimants required to pay nominal fees 9
Pilot locations Outline Land tenure in Rwanda Program characteristics Sample & estimation strategy Data & descriptive statistics Econometric results Conclusions 10
Estimation strategy We examine the effects of land registration: Y ph α β β β Z + ε = + 1T ph + 2X ph + 3 Where Y ph is the outcome of interest for parcel p in household h, T is a title registration indicator, X is a set of parcel characteristics, Z household characteristics, and ε an error term. h ph Endogenity and Identification If registration were voluntary, the investment decisions, etc. would be endogenous But here we have administrative units (cells) in which registration is complete and compulsory We will use the discontinuity provided by the cell boundaries to identify effects 11
Sample Design So our sample looks like this 12
Estimation strategy Identification assumption: no other major policy or market conditions happen at the cell level to affect outcomes of interest Policies of interest happen at higher levels (soil = district, inheritance=national) In addition, we use spatial fixed effects (Magruder 2010, G&U 2008, Conley & Udry 2008) to control for unobservable conditions (market and soil) Outline Land tenure in Rwanda Program characteristics Sample & estimation strategy Data & descriptive statistics Econometric results Implications 13
Data We undertook a quick, light survey during April-May 2010 (there was no usable baseline) 3560 households split across pilot cells and bordering cells Questions on hh included demographics, housing, assets, credit, registration participation and knowledge Questions on parcel included land characteristics, investment, inheritance, sales Took GPS readings of hh location The empirical results are based on the rural sample not enough power in the urban sample Outline Land tenure in Rwanda Program characteristics Sample & estimation strategy Data & descriptive statistics Econometric results Conclusions 14
Comparison of means Total Control Treatment mean sd mean sd mean sd Perceived risk of expropriation (1 if 0.72 0.45 0.74 0.44 0.68 0.47 yes) *** Change in proportion of parcels receiving soil conservation measures Construction of new 0.07 0.43 0.04 0.42 0.10 0.44 conservation structures *** Maintenance of existing 0.09 0.37 0.07 0.35 0.11 0.38 structures *** New/maintenance of existing 0.14 0.51 0.10 0.50 0.19 0.52 Structures *** Changed seed type from local to 0.53 0.50 0.50 0.50 0.57 0.50 improved variety since 2007 *** Female jointly or alone owns parcel 0.87 0.34 0.88 0.33 0.85 0.35 *** Share of parcel owned by female (%) 42.55 27.40 42.14 27.12 43.11 27.77 Know who will inherit the parcel 0.64 0.48 0.60 0.49 0.69 0.46 *** Sons will inherit parcel 0.74 0.44 0.72 0.45 0.77 0.42 *** Daughters will inherit parcel 0.69 0.46 0.66 0.47 0.72 0.45 *** Spouse will inherit parcel 0.32 0.47 0.34 0.47 0.29 0.45 *** Children will inherit parcel 0.76 0.43 0.72 0.45 0.82 0.38 *** Number of parcels 6312 3619 2693 Regression results: Expr. risk Perceived risk of expropriation Treatment indicator -0.047 Treatment X Female head -0.042 Number of years possessed -0.000 Parcel was purchased -0.053* Parcel was inherited -0.073* Acquired through other means -0.004 Parcel size in hectares -0.010 Head's age 0.000 Female headed household -0.066 Number of observations 5345 includes controls for hh demographics, spatial FE 1000m, spatial SE 15
Rural investment dependent variable soil conservation (new( const + maint) imp seed Treatment indicator 0.099** 0.064 Treatment X Female headship 0.094** 0.003 Number of years possessed -0.002** 0.001 Parcel was purchased -0.030-0.074** Parcel was inherited -0.065-0.083 Acquired through other means -0.209** 0.087 Parcel size in hectares -0.002 0.032* Head's age -0.001-0.001 Female headed household -0.044-0.052* Number of observations 6325 6325 includes controls for hh demographics, spatial FE 1000m, spatial SE Results: Women access to land Female spouse/head owns or co-owns plot Treatment indicator -0.074** Treatment X Marriage certificate 0.171*** Treatment X Female head with no spouse 0.143** Has marriage certificate 0.075*** Female head with no spouse 0.094*** Male head with no spouse -0.837*** Number of observations 6209 includes controls for plot characteristics, hh demographics, spatial FE 1000m, spatial SE 16
Results: Inheritance dependent variable know inherit son inherit daughter inherit children inherit Treatment indicator 0.094** 0.102** 0.096** 0.133** Treatment X Female head -0.044-0.052-0.158** -0.046 Number of years possessed 0.001 0.001 0.001 0.000 Parcel was purchased -0.008 0.043 0.008-0.021 Parcel was inherited 0.026 0.072** 0.038 0.004 Acquired through other means -0.051 0.167* -0.030 0.025 Parcel size in hectares -0.023** 0.006 0.004-0.012 Head's age 0.001 0.002 0.004** 0.003** Female headed household 0.071* 0.210*** -0.003 0.094*** Number of observations 6325 4053 4053 6325 includes controls for hh demographics, spatial FE 1000m, spatial SE Results: Participation in land market (sold/purchased) dependent variable (participation)( (area ( of land transacted) Treatment indicator -0.052** -0.048** -0.045*** -0.054*** Treatment X Female headship -0.020 0.044 Head's age 0.001 0.001 0.001 0.001 Female headed household -0.009-0.002 0.013-0.003 Head has at least primary education 0.021 0.021-0.001-0.002 Number of observations 6325 4053 4053 6325 includes controls for hh demographics, spatial FE 1000m, spatial SE 17
Conclusions This low cost, participatory method produces a significant change in investment (maintenance and construction of soil conservation structures). These investments are being made for this generation, but also the next there is a boost in projected land inheritance for children (and it is being written down). Improves access to land to women with certified marriage certificate, but those without marriage certificate tend to be negatively affected. There is thus a need to understand the recordation of women s rights on land. Lowers participation in land sales market. Implications on plot subdivision and 1 ha required minimum size (much higher than the average plot size). 18