Land Registration and Gender Concerns in Rural Benin

Transcription

1 Land Registration and Gender Concerns in Rural Benin KENNETH HOUNGBEDJI and VIBHUTI MENDIRATTA Paris School of Economics, CNRS & EHESS CRED, University of Namur INED February 2015 Preliminary results. Please, do not cite. Abstract Between 2008 and 2011, the Beninese government implemented a land registration programme in 294 villages which was expected to be scaled-up at a later stage. The programme sought to provide land tenure security to rural households through the issuance of formal documentary evidence of land rights. This exercise, however, is not straightforward since the same landholding can be claimed by several users at once. This paper examines the change in the allocation of landownership rights to women as a result of the land registration programme. To isolate the causal role of land registration, we follow a panel of households before and after the land registration programme and exploit differences in exposure to the programme across villages. Our estimates suggest that in anticipation of the land registration program, some women are denied control rights over land while others experience a reduction in their status from that of an independent cultivator to that of a tenant who is farming land belonging to another person. The male household members seem to have claimed landownership and increased their participation in agricultural activities. However, we find no negative impact on household welfare as proxied by total per capita consumption expenditure. Keywords: Land tenure, Formalization, Labor, Gender, Benin JEL Classification: J16, K11, O13, Q12, Q15. This work was supported by Région Île de France. We are grateful to the National Institute of Statistics of Benin for giving access to the data set used; to Denis Cogneau and Sylvie Lambert for their invaluable support, suggestions and comments throughout the preparation of this paper; to Gani Aldashev, Jean-Marie Baland, Guilhem Cassan, Andrew Clark, Catherine Guirkinger, Jean-Philippe Platteau and the participants of the CRED Workshop, the CFDS and the WIP seminars for helpful discussions and suggestions. Any remaining errors or omissions are ours. 1

2 2 1 Introduction Renewed interest in agricultural growth has put issues of land rights in developing countries back on the agenda of policy makers and development agencies. 1 To promote growth, several countries have started land registration programmes which demarcate land held by farmers and issue a formal documentary evidence of the land rights recorded. It is expected that the codification and the enforcement of land rights, which have been formally registered, will increase land tenure security and allow farmers to invest more and/or allocate their land to the best use (Besley, 1995; Feder and Feeny, 1991). In actual fact, however, the issuance of a land certificate begs the question of who owns the land and/or which land rights should be formalized. This chapter describes the access to land and landownership rights for women and investigates how they have evolved during a land registration programme which has demarcated all landholdings held by households across 294 villages between 2008 and 2011 in Benin. We investigate whether the formal registration of land rights led to a deterioration in the position of women. In particular, we check whether some men claimed control and cultivation rights over land previously cultivated by women and assessed the consequence of this on the household welfare. As demonstrated in Ester Boserup s seminal book Woman s Role in Economic Development, women, in the African agrarian system, bear the major responsibility for household food security and constitute a growing force on the labour market. Indeed, the female participation rate to the labour market rose from 57 to 61% in Sub-Saharan Africa between 1980 and 2009 whereas male participation rate decreased from 83 to 81% over the same period (see World Bank, 2012, pg. 200). With limited land rights under customary land tenure, women are often entitled to use land belonging to their husbands to grow food for their family and are allowed to sell the surplus of such crops. It follows that the same landholding can be cultivated by a woman while her husband owns it. From that perspective, and considering that assets are by default separately owned, the is- 1 From 1990 to 2004, Giovarelli et al. (2005) counted 44 projects with land as a primary component that were approved by the World Bank. Three land projects were approved over the period 1990 to This number increased to 19 and to 25 in the and periods, respectively (Giovarelli et al., 2005). In 2010 the Millennium Challenge Corporation (MCC) reported an investment of $ 249 million in property rights and land policy reforms across 11 countries. Those countries include Benin, Burkina-Faso, Ghana, Lesotho, Madagascar, Mali, Mongolia, Mozambique, Namibia, Nicaragua and Senegal.

3 3 suance of a land certificate may prompt a strategic behaviour on the part of landholders with a control or de jure right these are mostly men. Indeed this group will want a legal recognition of their land control rights. Anticipating that the land registration process might give a land certificate to the tenant who is cultivating the land, the holders of land control rights have an incentive to pre-emptively reassert their claim over their land. It remains unclear, however, as to how this anticipation will translate into penalizing or rewarding women s access rights to land and their cultivation activities. This is the question we seek to address in this chapter. To study the effect of the land registration programme on women, this chapter uses panel data which cover a nationally representative sample of households in Benin interviewed before and during the land registration activities. The land registration programme in Benin proceeded in two steps. In the first step, some regions were targeted based on a set of criteria. Then, within each targeted region all the villages were informed about the land registration programme and lotteries were organized to randomly select villages to have their landholdings demarcated. We can therefore distinguish between two potential effects of the registration programme. First, we estimate the effect of the prospect of the land right formalization by comparing the change of access to land and allocation of land rights to women over time between households living in the regions targeted and their peers in the regions not targeted. Next, we estimate the effect of the actual land demarcation activities by comparing households in villages randomly selected for a land demarcation treated villages to households in the other villages control villages. The chapter contributes to the existing literature on the formalization of property rights in two central ways. First, it is one of few studies to examine the effect of a land registration programme on access to land and the land rights held by women. Land registration programmes in Sub-Saharan Africa are relatively new. Related studies have focused on the effects of land reforms under colonisation and the spread of Islam, population pressure and land commercialization. Second, this chapter explores also the welfare gains of such interventions, an aspect that has been neglected so far. Most of the attention is often devoted to the effect of formalization of land rights on investment and growth whereas equity at efficiency concerns are seldom considered. Given that women are central in the structure of agricultural households in Sub- Saharan Africa, the distributional effect of land reforms is important. These concerns could also have more far reaching implications. For instance Aldashev et al. (2012) suggest that the formal law can induce behavioural changes in societies to the point that they can reform customary practices. Kumar and Quisumbing (2012) provide an empirical study which suggests that a

4 4 community based land registration programme and changes in the family code have created conditions for mutually reinforcing gender-sensitive reforms in Ethiopia. The findings of this chapter suggest that in anticipation of the land registration programme, the share of households where a woman has landownership rights has decreased by 13%. We observe that the proportion of households with a landholding that had a female landowner in our sample has fallen from 33% in 2006 before the information about the land registration programme spread to 19% in 2010 two years after the beginning of the information campaigns about the land registration programme. Our identification strategy gives a lower bound estimate of the anticipation effect since the regions that were not targeted had also received information about the formalisation of land rights and are expected to receive the programme during scaling-up at a later stage. 2 Further analysis indicates that the reduction in landownership rights allocated to women was accompanied by a similar increase 12% in the proportion of households that have declared a female member as family helper in the production of crops. Lastly, we observe that after the actual land demarcation activities had occurred, the allocation of landownership right to female household members was slightly higher in villages randomly selected to receive the programme compared to otherwise similar non selected villages. The remainder of this chapter is organized as follows. In Section 2 we provide a brief overview of the studies on female access to land under customary land tenure in Sub-Saharan Africa. We introduce the prevailing land tenure institutions in Benin and present the land registration programme the PFR (Plan Foncier Rural or Rural Land Plan in English). In Sections 3 and 4, respectively, we present the data used in the chapter and describe access to land and the distribution of landownership rights between men and women in Benin in 2006 and This analysis shows the first descriptive evidence regarding land tenure by gender in Benin. In Section 5, we present our econometric approach to measure the effect of the land registration programme on access to land for women and discuss the identifying assumptions. In Section 6 we provide a summary of the findings and discuss their implications. The chapter closes with Section 7, which offers a set of conclusions for researchers and policymakers to consider. 2 There was much discussion in the media before a law was passed in October 2007 to demarcate all landholdings in Benin, more discussion on this later.

5 5 2 Background and Context 2.1 Review of the Literature Land certification programmes have become increasingly popular over the last 2 decades in Africa; in view of legalizing the existence of customary laws that dominate the way land is held. Formalizing land rights has been argued to be a means of strengthening tenure security, promoting investment and productivity, reduce conflict and gain access to credit. It is also argued to be a cost effective intervention. In her seminal book Woman s Role in Economic Development, Ester Boserup presents the acceptance for European settlers, colonial administrators and technical advisers that cultivation is naturally a job for men persuaded them to believe that men could become far better farmers than women... [then]... for the development of agriculture male farming ought to be promoted to replace female farming (see Boserup et al., 2007, pp ). Boserup further demonstrates that this belief materialized through land reforms and other policies which have resulted in the loss of status for women under colonisation in Sub-Saharan Africa. First, she cites Floyd (1959) who reports that in the Bikita Reserve, located in present day Zimbabwe, a land reform of the British administration in 1957 allocated land to men and widows only. Hence in case of divorce, former wives were deprived of the land they had cultivated. Another reform was implemented in 1898 in the then Union of South Africa and introduced the rule of one man one plot with the result that women in a polygynous union had to cultivate together (with their husband s other wives) the land belonging to their husbands. Thus, restricting access to land for women. This reform was followed by a similar transfer of land from women to men in the Taung reserve where, although women were supposedly the main cultivators, the irrigated fields were allocated to men who were the only ones that were taught the techniques of irrigation farming. Last, she argued that in the Kon region of Nigeria women s resistance to the deterioration of their position as farmers coupled with their fear of losing their land to male farmers led to major upheavals in Aside from the reforms described above, that were explicitly focussed on transferring land rights toward men at the expense of female control over land, Lastarria-Cornhiel (1997) describes land privatization programmes in Africa that are gender neutral. She argues that these programmes that appear to be gender-neutral may be biased against women if they fail to ac-

6 6 count for women s land rights, their participation in agriculture, and the gender ideology in the project area. Among others, the author reports that women lose access or cultivation rights when land ceases to be valued for what it can produce and becomes an asset. In particular she explains that the privatisation of land rights allows the land title holders to exert disposal and exclusionary rights in places where those rights were restricted by customary practices. This aspect echoes with the general finding of land capture by bureaucrats, local chiefs and wealthy urban residents who use their knowledge and influence over land titling and land reform to have land titled in their name (Platteau, 1996). Finally, studies related to the access of land for women are more prolific in describing the effect of population pressure and land scarcity on rights to land for women and highlight key features of customary land tenure system (see Meinzen-Dick et al., 1997). Even though customary practices vary a lot across space, common among these systems, however, is that ownership of land is vested in the community in most parts of West Africa. Land is managed collectively and the household has secure right to cultivate or otherwise use land and pass it on to their heirs. In most of the cases, men control allocated land which are passed down to their male heirs whereas women usually have only cultivation rights. For most women access to land depends on their relation to male relatives. A husband for instance has an obligation to provide arable land to his wife to farm and decides which specific piece of land his wife can use and for how long. In case of divorce, the former wife loses her land cultivation rights (Lastarria-Cornhiel, 1997). In case of widowhood, land is transferred to male heirs and women have access to land use rights via their male son. The degree to which those rules are applied varies a lot across space as land quality, social norms, demographics and geography change in Africa. Nevertheless, while men own the land, women have land cultivation rights and own the crops. In a context of growing economic stress, this relationship between spouses and land sparks a constant renegotiation through contestation of rights to land and labour. In Senegal, Carney and Watts (1991) argue that effort to intensify rice production in the Gambia River-Bassin increased land value and initiated intra-household struggles over access to land and control over land and labour. In southern Niger, Diarra et al. (2006) reports cases where the increasing land pressure has enhanced land market values while women s right to cultivate new crops was progressively eroded and the proportion of landless women increased. While land is a major asset in most of Sub-Saharan Africa including Benin, population pressure

7 7 and the induced scarcity raised the economic value of land for households. In such a scenario, privatization of land could ignite conflicts within households, leading to land capture and may end up penalizing secondary users, particularly women. 2.2 The Plan Foncier Rural or Rural Land Plan (PFR) Benin, formerly Dahomey, was a French colony from 1890 to The country is bordered by Togo to the West, by Nigeria to the East and by Burkina Faso and Niger to the North. The administrative organisation of Benin has 12 departments which are subdivided into 77 communes. These communes comprise of 545 arrondissements which are further stratified into 5,290 villages and neighbourhoods; the smallest administrative units of the country. In October 2007, the Beninese Parliament passed a law recognizing the land certificate issued during a land registration programme as a legal proof of land rights. The law requires all villages to undergo land registration and provides legal recognition of land rights established or acquired in accordance with customary rules. For many practitioners, this law is a milestone for land tenure in Benin. For the first time, the law gives to landholders in rural area an alternative to a land title which only recognizes ownership and private property rights. 3 The land certificate provides legal recognition for land rights agreed upon by the neighbours and the community of the land certificate holder. With a land certificate, rural households may seek compensation in case of land conflicts or expropriation at the court of justice. 4 Once the law was enacted, the Millenium Challenge Account (MCA) offered to register all land held by households in 300 villages between 2007 to 2011 as part of its project Access to Land. 5 The process of land registration produces a map of all the landholdings located in the village and issues a land certificate for each demarcated plot. The map of the village is called Plan Foncier Rural or Rural Land Plan and the certificate are the Certificat Foncier Rural or Rural Land Certificate (see Figure 1 for an example of PFR). 3 The transaction costs in time and financial resources for obtaining a land title are indeed high. Reviewing demands of land titles received by the Direction des Domaines, de l Enregistrement et du Timbres the Beninese authority in charge of the issuance of private property right land titles between 2006 and 2007, MCA (2009) finds that the average costs of a land title is FCFA 500,000 this is almost the double of the average gross domestic product per capita of Benin between 2006 and 2007 and the process takes over 2 years to complete. 4 The ability to enforce land rights at a court may however be limited by the social cost or stigma imposed on individuals who seek mediation outside their customary system (Aldashev et al., 2012). 5 MCA lobbied for the law to be enacted before the beginning of its project Access to Land. The project Access to Land provides formal land registration to households. Overall, the activities of the MCA encompass three other projects. The project Access to Financial Services is directed at poverty reduction through the creation of jobs and increased income as a result of expanded activities by micro, small, and medium size enterprises. The project Access to Justice aimed at improving the judiciary procedures. Last, the project Access to Markets is designed to promote access to markets by reducing transaction costs related to accessing the port of Cotonou.

8 8 Figure 1: Rural Land Plan of the Village of Serou in Northern Benin. Source: Designed and drawn by the Institut Géographique National. To determine the 300 villages that would receive a land registration programme, 40 communes were chosen out of the 77 communes in Benin in January 2007 (before the passing of the law in October 2007). These communes were selected by a board of officials in charge of land administration in Benin and the 300 villages were supposed to be randomly selected across these 40 communes to receive a systematic registration of household landholdings (MCA, 2007). A list of ten criteria was defined and the communes were ranked accordingly. These 10 criteria gave priority to communes with the following characteristics: (i) They have a high degree of poverty but good economic prospects; (ii) they are also targeted by the projects Access to Financial Services and Access to Justice of MCA; (iii) they demonstrate a willingness to receive a land registration programme; (iv) they face a high pressure on land (v) they have agriculture as the dominant activity; (vi) they have a dynamic civil society, (vii) they show a high commitment for poverty alleviation policies; (viii) they have a good record from earlier land administration programmes and; (ix) a high proportion of migrants. The last criteria was to maximize an equal

9 9 representation of Southern, Center and Northern regions of the country. Remaining 37 communes comprising of rural, urban or peri-urban areas were not targeted by the Access to Land programme of MCA. 6 Following the selection, a series of information campaigns was carried out in each of the 40 communes chosen by the MCA to raise awareness regarding the land registration programme. Given the large number of villages to reach, the information campaigns proceeded from one village to the other following a staggered pattern. The information campaign involved, among other things, the provision of information regarding the programme and its benefits, answered questions raised by people and invited the village to apply for a chance to receive a land registration programme and a rural land plan of all land occupied the village inhabitants. Overall, most rural villages (in the 40 communes chosen by the MCA) have received an information campaign between January 2008 and January 2010; 36 villages were reached between January and August At the same time, a government body responsible for the management of watershed and forests (ProcGRN) started their own information campaigns in five forest dominated communes that were not targeted by the MCA. 7 Taking January 2010 as reference point, Figure 2 illustrates the number of days since the information campaign has reached each arrondissement, showing that all villages in the targeted 40 communes received an information campaign. Within the 40 communes, the final selection of 300 villages was randomized using public lotteries at the commune level. To evaluate the impact of the programme, the MCA had initially planned to use data collected from a nationwide household survey by the National Office of Statistics (INSAE), enquête modulaire integrée sur les conditions de vie (EMICoV from now on), in 2006 as baseline. However not all the villages were covered by the EMICoV survey in Thus for a given targeted commune, the rural villages that requested for a PFR were divided into two tracks. On the one hand, there were the villages that were covered by the EMICoV 2006 and on the other hand, there were the villages not covered by it. Each track is then assigned ex ante a certain number of villages to be preselected. The pre-selection is done using a score assigned to each 6 Though those communes were not targeted by the rural land registration programme they were expected to benefit from the scaling up of the land registration programme by the government in the future. Besides, the project Access to Land includes some financial and technical support intended for facilitating the issuance of land titles for landholders in urban areas. 7 ProcGRN stands for Programme de Conservation et de Gestion des Ressources Naturelles. It is a national agency in charge of protecting watershed and other natural resources against the expansion of neighbouring villages. It was operating in 46 villages at the time of the PFR carried out by the MCA.

10 10 MCA ProcGRN MCA & ProcGRN Number of days Number of days Figure 2: Implementation of MCA and ProcGRN PFRs Information Campaigns across Arrondissements in Source: Author s illustration based on EMICoV 2006, EMICoV 2010 and administrative data from MCA. village based on the following criteria: (i) The degree of poverty and availability of economic opportunities; (ii) compliance with the family code especially the formal law in favour of land bequest to women; (iii) the availability of government agencies in support of agricultural and economic development for women and poor; (iv) the existence of acute land issues 8. All the villages were graded on a scale from 0 to 100 based on the criteria above. For all the preselected villages within each commune-track combination, a public lottery was performed to select the final set of 300 villages that would receive the PFR. Figure Figure 3 provides a graphical overview of the selection procedure. In order to check the impact of the information campaign, we compare the targeted villages which includes all the villages located in 45 communes chosen by the MCA and the ProcGRN for a land registration programme and the non-targeted villages which includes the remaining 32 communes. From the selection, it is clear that the sample of targeted and non-targeted communes in not random (refer to Table A-8). In the following sections, we discuss different 8 The preselected villages were, by design, the poorest villages where women were more likely to have a landholding. Most of the preselected villages were assisted by government agencies to facilitate access to credit, improved seeds and fertilizer. Those measures, it was expected, would help bolster the economic outcomes of the land registration programme and the scaling up of the programme by the government.

11 11 strategies to account for this issue in our empirical analysis (see Figure 3 for a lexicon of the different groups of interest in this study). Figure 3: Levels of Treatment and Vocabulary. Information Campaign Targeted (1,714) Not targeted (3,576) Preselected (575) Not preselected (1,093) Treated (300) Control (275) Source: Author s illustration based on data collected during the field survey. Note: The number of villages in each category is reported in parentheses. Although intricate, the rules applied to identify the 300 villages that will receive a land registration program allow for a matching of each treated village covered by EMICoV with a corresponding control village. However, the number of control villages available to each village selected for a PFR is not balanced. For example there are 108 treated and only 78 control villages in the sample covered by EMICoV Within each village selected for a land registration programme, the land certificates are provided in 4 steps. First, a socio-legal inquiry is conducted that aims at identifying and clarifying rights and claims to land. This process is extremely thorough and involves investigating the cross-cutting registry of current customary rights over land, along with information on local issues, means of access to land, right holders and relevant authorities in a particular place. The second step involves the mapping of these rights on each identified plot. At that step, plat are demarcated with cornerstone in presence of neighbours and boundary disputes are discussed and settled. Plot plans are created and used to issue the land certificates. The third step is the publicity phase where claims are publicized and can be contested. The final step involves delivery of the land use certificate (Chauveau, 2003; Le Meur, 2008).

12 12 3 Data To assess the effect of the land registration programme on access to land for women and on household welfare, this chapter draws on several sources of information. First, we use administrative data, from the authority in charge of implementing the land registration activities in the 300 randomly selected villages, and document the date at which the information campaigns reached each village in the 40 communes targeted by the MCA. Second, we use information about land allocation at the household level collected in a survey undertaken by the World Bank and the Institute for Empirical Research in Political Economy (IREEP) in Third, we use information from the EMICoV surveys in rural areas about household characteristics, expenditures and landholding during the twelve months preceding the survey conducted in 2006, 2007 and The first EMICoV survey was conducted from August to November It covered 17,511 rural households and is representative at the commune level and the 17,511 are spread across 135 strata. Within each stratum, counting zones (CZ) were chosen proportionally to their size in the 2002 General Census of Population and Housing the Households. Overall, 750 CZ were selected during the survey. In each CZ, 24 households were randomly chosen to be interviewed. The households surveyed by EMICoV in 2006 were reinterviewed from October to December 2007 and from February to May In the remainder of the chapter, EMICoV 2006, EMICoV 2007 and EMICoV 2010 are used in reference to each of the previously mentioned rounds of EMICoV (please refer to Figure 4 for a detailed timeline of the roll out of the various surveys and the activities of MCA). Although the EMICoV survey tried to follow the same household over time, the construction of the panel was faulty. Overall, 80% of the 10,025 rural households interviewed in 2006 were also interviewed in 2007 and 2010 (see Table A-1). The attrition rate is mostly driven by households who relocate. Those households were more often female headed or headed by a migrant and had fewer members (see Table A-2). When a household relocates, it is replaced, when possible, by the new occupants of the old dwelling without a change to the identification number. 9 Since the data does not contain personal information on which households can be uniquely identified, it is not possible to ascertain that the identification number follows the same household 9 There are often no new occupants and the household is replaced by households randomly selected from the list of households enumerated in the CZ during the population census in In that case a new identifier is attributed to the new household.

13 13 Figure 4: Timeline of the Program Roll-out and Survey. EMICoV 2006 I.a Aug. Nov. EMICoV 2007 I.b Oct. Dec. EMICoV 2010 Feb. May. WB IREEP survey Mar. Apr Rural Land Act passes recognizing PFRs Oct Information Campaign Jan Jun Public lotteries to select 300 PFR villages Jan. Dec. PFR land surveys in 300 villages Jun Feb # Activity Time period Issuance of Land Certificates from Feb I. Data collection I.1 EMICoV 2006 I.a Aug. Nov I.2 EMICoV 2007 I.b Oct. Dec I.3 EMICoV 2010 II Feb. May 2010 I.3 WB-IREEP Mar. Apr II. PFR Implementation Activities II.1 Enactment of Law recognising the PFRs 16 th Oct II.2 PFR Information campaigns Jan Jun II.3 Lottery and selection of 300 PFR villages Jan. Dec II.4 Rolling out of PFR activities Jun Feb II.5 Issuance of Land Certificates From Feb Source: Author s illustration based on data collected during the field survey. over time. 10 Last, Table A-3 compares some key household characteristics between the identical sample households for which the gender of the household head remained unchanged between EMICoV 2006 and EMICoV 2010 and balanced sample households with the same identifier in EMICoV 2006 and EMICoV 2010 there are no significant differences. Thus, the results of the panel analysis are only valid for a the population of households that were successfully tracked or replaced over time. 10 Among other, this implies that the proportion of households headed by a migrant will be smaller in Assume, as Fenske (2010) shows in Côte d Ivoire, that migrants in Benin are also less likely to have a secure access to land. If a wife of a migrant is also discriminated against, the faulty panel structure will tend to increase the proportion of women with access to land between 2006 and 2010 in Benin. Indeed, compared to 2006, migration will decrease the headcount of women without access to land in 2010 will be smaller.

14 14 In 2010, the sample size of the EMICoV survey was expanded to include more villages where the PFR was ongoing. The survey conducted by the World Bank and IREEP in 2011 follows a subset of households randomly drawn from the extended EMICoV 2010 frame. It covers therefore households that were not originally surveyed in Overall 2,896 households were interviewed, with detailed information on their plots. The households were located in 244 villages, 150 of which were selected to receive the PFR while 94 were randomly selected to serve as control villages (resulting in 48 distinct lottery pools). 11 Since the survey only includes the villages that took part in the lotteries, it does not include any village from the 37 communes that were not targeted by MCA. 11 Table A-4 shows that the households that were added to the sample in 2010 share many similarities with those that were already covered in This helps to allay concern about endogenous expansion of the EMICoV sample in 2010.

15 15 4 Descriptive Evidence Research documenting the state of land use and land administration in Benin is limited this section helps to provide some evidence in that direction. Table 1 provides a summary description of household characteristics in 2006 (please refer to Subsection A-2. in the Appendix for details on the survey questions). Panel A: Household characteristics Table 1: Descriptive Statistics of Households in Obs Min Mean Median Std. dev. Max Female headed hh Age of hh head (years) Education (years) Household size Panel B: Access to land a HH with landholding HH sold a land in last 3 years HH cultivated a plot Total land size (ha) Land size per landholder (ha) Landholder acquired land through: - bequest purchase land rental Landholder can sell land HH with female landholder Land size (ha) HH with female land tiller cultivated her plot helped a fellow member Panel C: Daily consumption 2005 $ b Total consumption per cap Own food production from the hh head from the spouse Note: The table shows descriptive statistics of the households in our sample. The data used is the EMICoV 2006 survey. The sample is restricted to households which are covered in both the EMICoV 2006 and the EMICoV a Access to land is based on responses provided by the household head about the landholdings controlled by the household members (refer to Subsection A-2. for questions related to the measure of land access used in this study). b Prices are converted in 2005 International $ using the consumer price indexes and the purchasing power parity conversion factor for private consumption collected from the World Development Indicators (WDI). Indicates dummy variables. A typical household interviewed in 2006 is male headed, has 5 members, and has spent on

16 16 average $ 1.20 per day per person, hence is considered poor with respect to the $ 1.25 absolute poverty line rule. Most households have at least one landholding 80% of households are landholder and 85% of those households with land do cultivate their plot(s) Access to Land for Women between 2006 and 2010 In 2006, land held by household members was often acquired through bequest in 71% of the households with a landholding land was acquired as a bequest but land purchase and rental were also prevalent.table 1 shows that, in 2006, the landholder could sell or mortgage its land in 46% of households with a landholding. On the gender dimension, one third of the households with a landholding have reported at least one woman as landholder. Female landholders are on average older than their peers without a landholding and have fewer years of schooling (see Table A-6 ). They are also more likely to be divorcees or widows and head of their household. Compared to their male counterparts, female landholders are slightly younger and have fewer years of formal education. It appears that 49% of them acquired their landholding(s) through bequest against 71% of male landholders (see Table A-7). This is in line with the idea that women are less likely to inherit a piece of land. Unlike their male counterparts, female landholders are more likely to obtain a landholding through a land rental or a sharecropping contract in 2006, 21% of female landholders had acquired a landholding through a rental contract against 10% for male landholder. Lastly, women are twice more likely to work on a landholding that they do not exert control over 33% of female landholders helped cultivate a plot of a fellow household member versus 18 % for male landholders. Table 2 describes the household characteristics and changes in access to land and per capita consumption over time. From 2006 to 2007, the proportion of female headed households has 12 We find that the average landholding size for the household as a whole is 12.9 ha and the average plot size is 5.90 ha with considerable variation it stands at 3.13 ha in the Southern departments, 3.72 ha in the Center and ha in the North. These figures are large. In comparison the average land size estimated in 2011 using the survey undertaken with the WB and IREEP was estimated at 6.73 ha. But that figure is also likely upward biased. Indeed for a sub-sample of the households surveyed by the WB-IREEP survey, land area was measured using a GPS. When we focused on those households that have exactly one landholding, and taking into account rounding errors, we find that land size was correctly reported for 43% of landholdings (see Figure A-2). Otherwise, the average measurement error is estimated at 3.46 ha. Measurement errors are smaller for parcels with a female manager and decrease with years of education of the parcel manager (see Table A-5). On top of that upward bias in land size, several households covered by EMICoV 2006 and 2010 have not reported the size of their landholding. Out of the 6,275 households that have at least one landholding, land size is reported for 5,417 households. In addition, the survey questionnaire asked for land size to be reported in hectares and square meters, a confusion among which could also explain such large figures. Thus, we refrain from using land size as a variable in our analysis.

17 17 Table 2: Household Characteristics in 2006, 2007 and Year of the survey Diff Female headed hh *** 0.02*** (0.01) (0.01) (0.01) (0.00) (0.00) Age of hh head (years) *** (0.25) (0.23) (0.22) (0.13) (0.15) Education (years) *** -0.49*** (0.07) (0.06) (0.07) (0.04) (0.04) Household size *** -0.40*** (0.08) (0.07) (0.07) (0.05) (0.06) HH with landholding *** (0.01) (0.01) (0.01) (0.01) (0.01) HH sold a land in last 3 years *** -0.01*** (0.00) (0.00) (0.00) (0.00) (0.00) Landholder cultivated a plot *** 0.03** (0.01) (0.01) (0.01) (0.01) (0.01) Number of landholdings *** (0.04) (0.04) (0.03) (0.05) (0.05) Landholder acquired land through: - bequest *** 0.04** (0.01) (0.01) (0.01) (0.01) (0.01) - purchase *** (0.01) (0.01) (0.01) (0.01) (0.01) - land rental *** -0.05*** (0.01) (0.01) (0.01) (0.01) (0.01) Landholder can sell land *** 0.47*** (0.01) (0.02) (0.01) (0.02) (0.02) HH with female landholder *** -0.13*** (0.01) (0.01) (0.01) (0.01) (0.01) - Number of landholdings *** (0.03) (0.04) (0.03) (0.05) (0.05) HH with female land tiller (0.01) (0.01) (0.01) (0.02) (0.02) - cultivated her plot *** -0.11*** (0.01) (0.01) (0.01) (0.01) (0.01) - helped a fellow member *** 0.03 (0.01) (0.01) (0.01) (0.02) (0.02) Consumption per capita (2005 $) *** (0.03) (0.02) (0.03) Own food production (2005 $) *** (0.01) (0.00) (0.01) Number of households Note: The table compares descriptive statistics of household characteristics in 2006, 2007 and 2010 using the EMICoV survey. The sample is restricted to households which are covered in both the EMICoV 2006 and the EMICoV Columns and report the differences of the averages between 2007 and 2006 and between and 2006 respectively. In 2007, no question related to consumption and expenditure was asked and the consumption variables are therefore missing. Questions related to access to land are based on responses provided by the household head about the landholdings controlled by the household members (refer to Subsection A-2. for details). Lastly, prices are converted in 2005 International $ using the consumer price indexes and the purchasing power parity conversion factor for private consumption collected from the World Development Indicators (WDI). Standard errors are in parentheses and are clustered at the CZ or primary sampling unit level. The significance levels are reported for t-tests on the equality of the means over time for each variable. They are denoted as follows: * p<0.10, ** p<0.05, *** p<0.01. Indicates dummy variables.

18 18 slightly increased. Along that change, the average years of education of the household head and the household size have both decreased. This trend is further accentuated in 2010 as the proportion of female headed households has increased. It is consistent with the replacement of departed male heads by their wives who, on average, are younger and have fewer years of education. At the time of the EMICoV 2007, from October to December 2007, the Land Law was enacted. Awareness about the land registration programme was however prevalent before the enactment of the Law of October The passing of the law allowed the MCA to speed up its programme and to start the information campaigns targeted at the villages starting from January In this context, Table 2 shows that the proportion of households with a landholding has not changed between 2006 and 2007 but the share of households where the landholder has actually cultivated a plot has slightly decreased from 85% in 2006 to 81% in In 2007, there were also fewer households that sold land or where the landholder has a landholding acquired through land rental. In most households with a landholding the landholders still acquired their land through bequest. The proportion of households with such land is estimated at 77% against 71% in 2006 and the proportion of households where the landholder obtained land through land rental has fallen from 13% to 9% between 2006 and Since a given household can simultaneously have a landholder with land acquired through inheritance, purchase or rent, a smaller prevalence of land rental contracts does not imply a higher prevalence of other modes of land acquisitions. It just means a restricted number of de facto landholders. It is likely that de jure landholders that rented out a plot in 2006 claimed it back in 2007 and asserted themselves as de facto landholder. This scenario may explain the decrease of the proportion of households with female landholder from 33% in 2006 to 25% in The average number of landholdings held by female landholders, however, has not decreased; suggesting an extensive rather than an intensive reduction of land ownership patterns for female landholders between 2006 and This is further stressed by the observation that female members were more likely to help a fellow household member cultivate his plot and less likely to cultivate their own plot (due to the loss of status as being landholder) in 2007 as compared to Hence, the reduction of female landholders between 2006 and 2007 was not accompanied by a decrease of the participation rate of women in agricultural activities. As in 2006, 67% of households with a landholdings had a female member cultivating a plot in 2007.

19 19 In 2010, at the time of the follow-up round of the EMICoV, the information campaigns had already reached all villages in the 45 communes 40 communes selected by the MCA and five forest dominated communes where the PRocGRN had ongoing land registration activities targeted for land registration. Moreover, the land registration process had already started in some of the 300 villages randomly selected by the MCA. Table 2 shows that there is a reduction in the percentage of households with a landholding between 2006 and Households with a landholding were more likely to have it cultivated in 2010 and consumption per capita has fallen from $ 1.13 to $ Since we follow the same households over time, the decrease of the share of households with a landholding indicates that several households have either sold or have lost control rights over their landholdings. The latter seems more likely since only 1% of households sold some land between 2007 and 2010 and because the average number of landholdings has also reduced significantly between 2007 and Besides, in most households with a landholding in 2010 the landholder had acquired its holding through bequest 76% against 71% in 2006 while the proportion of households where the landholder acquired land through rental agreements decreased from 13% to 8%. Next, the proportion of households where the landholder declared they could either sell or mortgage their land has significantly increased between 2006 and % of the households had a landholder that could sell land in 2010 versus 46% in This suggests that most landholders observed in 2010 were de jure landowners that have the right to sell or mortgage their land. Consequently, the distribution of land ownership rights across gender is more skewed in 2010 compared to 2006 and In 2010, the households with a female landholder represented 19% of the households with a landholding. This is smaller than the 25% share of households with female landholders observed in However, overall participation of women in agricultural activities has not changed and remained stable over time even though there is a decline in households where women cultivate their own plot which is offset by an increase in the percentage of households where women are helpers in agricultural activities. Figure 5 offers a visual representation of the variation of the share of households with a female landholders across arrondissements in 2006, 2007 and 2010 and time spent since the PFR information campaign has reached each arrondissement. In 2006, access to land for women varies across space. In 2007, right after the enactment of Land Law of October 2007, the proportion of households with female landholders decreased especially in the northern regions of the country. This is further seen in Table 3. In 2010, there were even fewer households with a female

20 20 Figure 5: Proportion of Households with Female Landholders across Arrondissements and Information Campaigns 2006 Share (%) (a) Aug. Nov Share (%) (b) Oct. Dec MCA & ProcGRN 2010 Share (%) (c) Feb. May 2010 Number of days (d) Jan Jan Source: Author s illustration based on EMICoV 2006, EMICoV 2010 and administrative data from MCA. landholder and access to land seems to have decreased the most in the arrondissements that were targeted by the information campaigns. However, the share of households with a female landholder has also decreased in the arrondissements located in communes that were not targeted by the MCA. Even though the communes are clearly demarcated one from another on a map, it is often not the case on the ground. We observe that access to land for women had changed by 2007, before the start of the information campaigns. This is arguably due to the fact that the law that was passed in October 2007 stated that eventually all the village will undergo a land registration programme which will issue land certificate to landholders in Benin. Le Meur (2008) also makes the point that the passing of law of 2007 was, to some extent, the result of an information campaign that was meant at building support for the rural land certification and the land registration programme of the MCA. This could potentially have informed the Beninese citizens about land registration the information campaigns made it more salient starting from Further evidence of this is provided in our data where we observe that land access had changed between 2006 and 2007 for both targeted and non-targeted villages (results not shown). Thus in the remainder of this chapter, we do not distinguish between the enactment of the law of October 2007 and the PFR information campaign that reached the villages between January 2008 and January The

21 21 rest of the analysis focusses on the change in land ownership patterns between 2006 and Table 3: Household Characteristics in 2006 and 2010 and Information Campaigns. Targeted Not targeted diff diff.(1) diff.(2) (1) - (2) Female headed hh *** *** (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) Age of hh head (years) *** *** (0.31) (0.26) (0.20) (0.41) (0.37) (0.22) (0.29) Education (years) *** *** (0.07) (0.07) (0.05) (0.13) (0.13) (0.08) (0.09) Household size *** *** 0.20* (0.08) (0.08) (0.07) (0.14) (0.11) (0.10) (0.12) HH has a landholding ** ** (0.01) (0.02) (0.01) (0.02) (0.03) (0.01) (0.02) HH sold a land in last 3 years *** *** 0.00 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.01) Landholder cultivated a plot * (0.01) (0.01) (0.01) (0.02) (0.02) (0.02) (0.03) Number of landholdings *** *** -0.26*** (0.06) (0.03) (0.06) (0.07) (0.05) (0.08) (0.10) Landholder can sell land *** *** 0.01 (0.02) (0.01) (0.02) (0.03) (0.01) (0.03) (0.04) HH with female landholder *** *** -0.06** (0.02) (0.01) (0.02) (0.02) (0.02) (0.02) (0.03) - number of landholdings *** * (0.04) (0.03) (0.07) (0.06) (0.06) (0.08) (0.10) HH with female land tiller * (0.02) (0.02) (0.02) (0.02) (0.03) (0.03) (0.04) - cultivated her plot *** *** -0.05* (0.02) (0.01) (0.02) (0.02) (0.01) (0.02) (0.03) - helped a fellow member ** ** (0.02) (0.02) (0.02) (0.02) (0.03) (0.03) (0.04) Consumption per cap. (2005 $) *** *** 0.15** (0.04) (0.02) (0.04) (0.06) (0.05) (0.06) (0.07) Own food production (2005 $) *** *** 0.04** (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.02) Number of households Note: The table compares household characteristics across locations targeted and not targeted by the land registration information campaigns. The statistics reported under the heading Targeted refer to the sub-sample of households located in one of the 45 communes chosen by MCA and ProcGRN for a land registration programme and whose village received an information campaign. Statistics reported under the heading Not targeted are based on the sample of households living in the remaining 32 communes not chosen and subsequently did not receive any information campaign about land registration. The sample is restricted to households which are covered in both the EMICoV 2006 and the EMICoV Columns diff. (1) and diff. (2) describe variation of household characteristics in each sub-sample between 2006 and Column diff. (1) - (2) shows time trend differential across both sub-samples between 2006 and Questions related to access to land are based on responses provided by the household head about the landholdings controlled by the household members (refer to Subsection A-2. for details). Standard errors are in parentheses and are clustered at the CZ or primary sampling unit level. Significance levels for coefficients in columns diff.(1) and diff.(2) are reported for t-tests of the equality of the means over time for each of the variables for households in targeted and not targeted communes. Significance levels for coefficients in column diff. (1) - (2) are reported for the test of equality between diff.(1) and diff.(2). Significance levels are denoted as follows: * p<0.10, ** p<0.05, *** p<0.01. Indicates dummy variables.