Scholarly article on topic 'Policy Reform toward Gender Equality in Ethiopia: Little by Little the Egg Begins to Walk'

Policy Reform toward Gender Equality in Ethiopia: Little by Little the Egg Begins to Walk Academic research paper on "Law"

CC BY-NC-ND
0
0
Share paper
Academic journal
World Development
OECD Field of science
Keywords
{gender / reforms / "Family Code" / "land registration" / Ethiopia}

Abstract of research paper on Law, author of scientific article — Neha Kumar, Agnes R. Quisumbing

Summary There is growing interest in how reforms in different policy areas can be formulated in order to be consistent in promoting gender equality and empowering women. We use data from the Ethiopian Rural Household Survey (ERHS) to show how two seemingly unrelated reforms—community-based land registration, undertaken since 2003, and changes in the Family Code implemented in 2000—may have created conditions that reinforce each other in improving gender equity. Our findings suggest that the land registration process and the reform of the Family Code had mutually reinforcing effects on women’s rights and welfare.

Academic research paper on topic "Policy Reform toward Gender Equality in Ethiopia: Little by Little the Egg Begins to Walk"

World Development Vol. 67, pp. 406-423, 2015 0305-750X/© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/3.0/).

j.worlddev.2014.10.029

Policy Reform toward Gender Equality in Ethiopia: Little by Little the Egg Begins to Walk

NEHA KUMAR and AGNES R. QUISUMBING*

International Food Policy Research Institute, USA

Summary. — There is growing interest in how reforms in different policy areas can be formulated in order to be consistent in promoting gender equality and empowering women. We use data from the Ethiopian Rural Household Survey (ERHS) to show how two seemingly unrelated reforms—community-based land registration, undertaken since 2003, and changes in the Family Code implemented in 2000— may have created conditions that reinforce each other in improving gender equity. Our findings suggest that the land registration process and the reform of the Family Code had mutually reinforcing effects on women's rights and welfare.

© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/3.0/).

Key words — gender, reforms, Family Code, land registration, Ethiopia

ELSEVIER

www.elsevier.com/locate/worlddev

CrossMarl

http://dx.doi.org/10.10

1. INTRODUCTION

Kes be kes enqullal be-egrwa tihedalech Little by little, the egg begins to walk (Ethiopian saying)

Attention to gender equality remains an important development goal. The importance of gender equality is highlighted in its prominence in the United Nations Millennium Development Goals (MDGs), which have been commonly accepted as a framework for measuring development progress. Of the eight goals, four are directly related to gender: achieving universal primary education, promoting gender equality and the empowerment of women, reducing infant and child mortality, and improving maternal health. Closing gender gaps—which tend to favor males—has also been seen to contribute to women's empowerment. However, the term empowerment refers to a broad concept that is used differently by various writers, depending on the context or circumstance (see Kabeer (2001) and Ibrahim and Alkire (2007) for discussions and a review of concepts).

Other arguments for reducing the gender gap revolve around improving productivity and increasing efficiency, improved outcomes for the next generation, and more representative decision making, which are emphasized by Food and Agriculture Organization (FAO, 2011) and the World Bank (World Bank, 2011) in their flagship publications. Quisumbing et al. (2014) argue that the motivations for closing the gender gap are not mutually exclusive; rather, they reinforce each other. The linkages between women's empowerment and increased productivity and food security are emphasized by Alkire et al. (2013) and Sraboni, Malapit, Quisumbing, and Ahmed (2014), in their analysis of the newly developed Women's Empowerment in Agriculture Index. Closing the gender gap in assets—allowing women to own and control productive assets—both increases women's productivity and increases their self-esteem. A woman who is empowered to make decisions regarding what to plant and what (and how many) inputs to apply on her plot is likely to be more productive in agriculture. Similarly, an empowered woman is likely to be better able to ensure her children's health and nutrition because she is able to take care of her own physical and mental well-being (see

Smith, Ramakrishnan, Ndiaye, Haddad, and Martorell (2003) and studies reviewed therein). Thus, regardless of whether efficiency, equity, or both are stated development objectives, various studies have shown that reducing gender gaps is key to meeting these goals.

If closing the gender gap is such an important development objective, are there complementarities with other development goals that could be exploited? Could different policy reforms have reinforcing impacts on gender equality? This paper explores the complementarity of two different reform processes in Ethiopia that began in the 2000s: the promulgation of the revised Family Code in 2000, and the community-based land registration efforts, which started in 2003. North's (1990) theory of institutional change argues that institutions change incrementally rather than in a discontinuous fashion because they are embedded in formal and informal constraints in societies. Although rules may change overnight because of political or judicial decisions, informal constraints are deeply rooted in customs, traditions, and codes of conduct, and thus change more slowly in response to deliberate policies (North, 1990, p. 6). However, institutional change may be reinforcing as well as path dependent (North, 1990, p. 99).

The possibility of mutually reinforcing policy reforms is relevant to Ethiopia, where gender norms related to property ownership, inheritance, and the division of assets after divorce favor men (Fafchamps & Quisumbing, 2002, 2005). Such gender disparities have important welfare consequences. Dercon

* This research was supported by the Swiss Development Corporation and the International Food Policy Research Institute Strategic Initiative on Gender and Assets, with additional support from the Federal Ministry for Economic Cooperation and Development, Germany and the CGIAR Research Program on Policies, Institutions, and Markets. We thank Klaus Deininger, John Hoddinott, and Stein Holden for helpful discussions on the design of survey modules related to land registration and family law, and Laura Dick and Ruth Meinzen-Dick for discussions on institutional change. We also thank participants at the Ethiopian Rural Household Survey Workshop at Lalibela, with special thanks to Paul Dorosh, Alemayehu Seyoum Taffesse, Stefan Dercon, Valerie Mueller, and Tassew Woldehanna, and two anonymous referees for their comments. All errors and omissions are our own. Final revision accepted: October 27, 2014.

and Krishnan (2000) find that poor women in the southern part of Ethiopia, where customary laws governing settlement at divorce are biased against women, suffer a greater deterioration in their body mass index when illness shocks occur. Fafchamps, Kebede, and Quisumbing (2009) find that the relative nutrition of spouses is associated with correlates of bargaining power, such as cognitive ability, independent sources of income, and devolution of assets upon divorce, and that several dimensions of female empowerment benefit the nutrition and education level of children. These results in Ethiopia are corroborated by findings in Asia and Latin America: in India, Panda and Agarwal (2005) find that women owning immovable property (land or a house) face a significantly lower risk of marital violence than propertyless women. In Nepal, women who own land are significantly more likely to have more say in household decisions, a measure of empowerment, and children whose mothers own land are less likely to be severely underweight (Allendorf 2007). Menon, van der Meulen Rodgers, and Nguyen (2014) find that land titling for women in Vietnam led to improvements in child health and education, and that these effects were stronger than in households with male-only or jointly held land use rights. Joint property rights for spouses and cohabitants in Peru had significant positive effects on women's empowerment, with the effects strongest for increasing women's decisions on large investments and agriculture, and less impact on daily expenditures, as market operations are traditionally seen as a female responsibility anyway (Wiig, 2013).

In this paper we use data from the Ethiopian Rural Household Survey (ERHS) to show how two reforms—the changes in the Family Code implemented in 2000 and the community-based land registration, undertaken since 2003—may have created conditions for gender-sensitive reforms to reinforce each other. To assess whether these reforms have the potential to have differential impacts on households, depending on the sex of the household head, household asset ownership, and other characteristics, we begin by examining how these household characteristics are correlated with changes in women's perceptions regarding allocation of assets upon divorce, and knowledge of and participation in the land registration process. We then analyze whether the two reforms were complementary. We use data from the 1997, 2004, and 2009 rounds of the ERHS, which covered approximately 1,300 households in 15 villages all across Ethiopia. The timing of the survey rounds, before and after these significant policy changes in Ethiopia, enables us to examine the potential complementarity of these reforms.

2. INSTITUTIONAL CHANGE: CONCEPTS AND RECENT ETHIOPIAN EXPERIENCE

(a) A conceptual framework for understanding institutional change

In his 1990 book, Nobel laureate Douglass C. North proposes a theory of institutional change that examines the interaction between formal and informal constraints to economic and political behavior. Among these constraints, one of the most important is property rights (North, 1990, p. 33). According to North (1990, p. 40), formal laws and property rights are only a small portion of the rules that govern society; informal constraints are probably more important and numerous in practice, come from socially transmitted information, and are considered part of culture. Although not discussed by North, gender norms determine how different cultures

define men's and women's property rights and are part of the institutional framework that defines customs surrounding marriage (and marital dissolution), inheritance, and production relations. Because there are a large number of specific (formal, but mostly informal) constraints that affect a particular choice, institutions tend to change very slowly. Significant changes in the institutional framework involve changes in constraints, not only in legal constraints but also in norms of behavior. Only when it is in the interest of those with sufficient bargaining power to alter the formal rules will there be major changes in the formal institutional framework.

Yet, cultures do change over time. North (1990, p. 94) draws from work by Arthur (1989), who argues that small historical events can lead to one technology winning out over another, and thus, for changes to be self-reinforcing. If there are no increasing returns to institutions and markets are competitive, institutions do not matter. But, with increasing returns, all the self-reinforcing mechanisms hypothesized apply. In the context of institutional change, these mechanisms apply when there are: (1) large setup costs when institutions are set up for the first time (or when drastic changes are put into place); (2) significant learning effects for organizations that arise in consequence of the opportunity set provided by the institutional framework; (3) coordination effects directly via contracts with other organizations, and indirectly by induced investment through the polity in complementary activities; and (4) formal rules that result in the creation of a variety of informal constraints that modify the formal rules and extend them to a variety of specific applications. Adaptive expectations occur because increased prevalence of contracting based on a specific institution will reduce uncertainties about the permanence of that rule. In short, the interdependent web of an institutional matrix produces massive increasing returns.

One can apply this reasoning to the promulgation and implementation of policy reforms that promote gender equality and identify ways these self-reinforcing mechanisms could manifest themselves. First, institutional reforms, particularly constitutional reform or changes in statutory law, typically take time because of the need to build a constituency to support these changes, whether in legislative bodies or in the electorate. Second, other government agencies, as well as civil society organizations, gear up to implement these reforms, as well as increase knowledge of the reforms through legal literacy campaigns. Third, these reforms may lead to changes in behavior of elected officials and civil servants: for example, courts of law may rule based on the new guidelines regarding the settlement of marital disputes; once both spouses are allowed to own property equally, there may be support for efforts to register land jointly in men's and women's names. Finally, the formal rules that remove gender-based discrimination in property rights may induce parents to change their views toward their sons and daughters, rewrite wills to favor sons and daughters equally, and, seeing that the external environment has become more favorable toward girls, choose to invest in their daughter's human capital by sending them to school.

There is suggestive cross-national evidence of such reinforcing processes toward increased gender equality. Hallward-Driemeier, Hasan, and Rusu (2013a) track the evolution of key constraints to women's and girls' equal rights to property and restrictions on their legal capacity over the past 50 years across 100 countries using a database of legal indicators representing all geographic regions, legal traditions, and income levels. The authors examine which country characteristics and processes are associated with reforms, focusing on income growth, education, patterns of employment, conflict, women's

empowerment (e.g., women parliamentarians, ratification of the Convention to Eliminate Discrimination Against Women (CEDAW), women's age of first marriage) and the strength of the rule of law. Their results show that income does not necessarily predict the adoption of anti-discriminatory reforms, and that the ratification of CEDAW and higher participation of women in national parliaments are significantly associated with more reforms closing gender gaps in women's economic rights. While one could argue that countries with higher levels of female education or higher female participation in the labor force could be more likely to adopt CEDAW (that is, the ratification of CEDAW is also endogenous to political processes reflecting underlying gender norms), the positive associations are remarkable. In another paper using the same 50-year legal rights database, Hallward-Driemeier, Hasan, and Rusu (2013b) find that the elimination of gender gaps in the ability to access and own property, sign legal documents in one's own name, and have equality or non-discrimination as a guiding principle of the country's constitution is associated with greater participation of women in the labor force, greater movement out of agricultural employment, higher rates of women in wage employment, lower adolescent fertility, lower maternal and infant mortality, and higher female educational enrollment. Although causality cannot be established convincingly, these results suggest that reforms promoting gender equality may have self-reinforcing and cumulative effects.

(b) An overview of gender and property rights institutions in Ethiopia 1

Ethiopia, the second most populous country in Africa, is characterized by substantial ethnic and religious diversity, with more than 85 ethnic groups and most major world religions represented, as well as animist belief systems (Webb, von Braun, & Yohannes, 1992). This diversity extends beyond the people and culture of Ethiopia to their environment, as the agroecological zones and, consequently, farming systems vary dramatically around the country. Different religions, with widely divergent views regarding matrimonial issues in general, and the status of women in particular, are well represented and tend to dominate different parts of the country -the Orthodox church of Ethiopia in the north, Sunni Muslims in the east and west, recently converted Protestants in the south, and animist beliefs in parts of the south. This variation is reflected in the diversity of gender norms, which tend to favor men in property rights over assets (Fafchamps & Quisumbing, 2002), with some noticeable regional patterns that indicate that as one moves from north to south in Ethiopia, women's status, and therefore possibly their bargaining power, declines (e.g., Bevan & Pankhurst, 1996; Gopal & Salim, 1999).

Norms regarding marriage and inheritance are important in determining the relative control that men and women have over productive and other assets. Marriage is a fluid state in Ethiopia; divorce is frequent and serial marriages are common (Pankhurst, 1992). First marriages are likely to involve a bond between households, rather than a personal arrangement by the bride and groom (Pankhurst, 1992, p. 122). Analysis of marital histories and assets brought to marriage finds that on average men bring substantially more physical and human capital to the marriage than do women (Fafchamps & Quisumbing, 2002).2 The great majority of the new couple's assets are brought to the marriage by the newlyweds themselves, with grooms bringing substantially more start-up capital than brides. Contrary to the preconception that marriage is the time at which parents endow their offspring with farmland,

most of the land brought by grooms was already theirs prior to marriage, because the state nominally owns the land and assigns use rights (Gavian & Ehui, 1999).

Because marriage is a relatively fluid state in rural Ethiopia, customary practices surrounding marital dissolution also affect the wealth and relative bargaining power of men and women. The literature on bargaining models of the household (e.g., Lundberg & Pollack, 1994; Manser & Brown, 1980; McElroy & Horney, 1981) also suggests that legal and customary dispositions regulating the disposition of assets upon divorce affect the gender distribution of welfare not only after but also during marriage because the bargaining power of married women is thought to depend on their exit option from marriage. A study conducted prior to the passage of the Family Law in 2000 (Fafchamps & Quisumbing, 2002) found that half of the surveyed monogamous households expect the land and house to go to the husband upon divorce; another 40% expect them to be divided equally between husband and wife. Regarding livestock, equal division between husband and wife is the rule, irrespective of whether the livestock is owned jointly or individually by the husband and the wife. Individually owned livestock, however, is more likely to be attributed to its owner upon divorce. The situation in polygamous households is more male dominated in the sense that the husband is much more likely to be given all assets upon divorce. Even there, however, jointly owned livestock is expected to be divided equally in most cases. In two-thirds of the cases, respondents expect the wife to receive custody over young children. Older children, in contrast, are expected either to follow their father or to choose which parent they wish to live with.

The allocation of assets upon divorce varies considerably depending on who is at fault (Fafchamps & Quisumbing, 2002). If it is the husband, the wife is slightly more likely to be granted land and livestock, while if the wife is at fault, asset distribution is dramatically changed in favor of the husband. Even her own livestock is likely to go to her husband. Fault-based divorce is punitive, and is particularly harsh for wives. The concept of fault-based divorce is more prevalent in the south-central region, especially among Protestants and Catholics. Drunkenness, wife-beating, adultery, and failure to support one's wife are most cited as husband faults that justify divorce, while adultery, involvement in crime, disrespect, and disposition of assets without consultation are the most commonly cited faults for wives.

Upon the death of the household head, assets are most likely to go to the surviving spouse, together with child custody. The devolution of livestock to the surviving spouse is essentially unaffected by who owns it. Children inherit in less than half the cases, and when they do, it is usually together with their mother.

There are, however, sharp differences in customs across locations, ethnic groups, and religions. Northern locations are in general more generous toward women. There is systematic variation across ethnic or religious groups, but location-specific norms are the best predictor of the disposition of assets upon marital dissolution. Moreover, communities may have their own ways of protecting women and other vulnerable groups; local councils may also mediate the distribution of assets should a dispute arise.

(c) Institutional change in gender and property rights

While progress toward gender equality has been slow in Ethiopia, recent developments are promising. Prior to 2000, legal reform had a limited impact on local traditions regarding patrimonial issues. For example, although the 1960 Civil Code

gave women more rights than their contemporaries in the United States or United Kingdom, it also maintained the tradition of dispute settlement by personal arbitrators, normally older men within the family or community selected by the disputants. The arbitrators, unfamiliar with or unsympathetic to the new laws, continued to apply old customary laws. The de jure system had nothing to do with the de facto reality that existed for the next 30 years (Gopal, 2001). The major exception was the distribution and control of land, an area in which the Ethiopian state has played a dominant role throughout the centuries.

In 2000, however, the revised Family Code was passed, giving equal rights to spouses during the conclusion, duration, and dissolution of marriage. It also required equal division of all assets between the husband and wife upon divorce (Federal Negarit Gazetta Extra Ordinary Issue, 2000). During 2000-05, Amhara, Oromiya, and Tigray implemented the code (out of nine regions in Ethiopia), but as of 2011, all regions now apply the revised Family Code (Hallward-Driemeier & Gajigo, 2011). Hallward-Driemeier and Gajigo (2011) found that women were relatively more likely to work in occupations that require work outside the home, employ more educated workers, and in paid and full-time jobs in areas where the reform had been enacted, controlling for time and location effects. The relative increase in women's participation in these activities was 15-24% higher in areas where the reform was carried out. Kumar and Quisumbing (2012), analyzing the impact of changes in divorce laws following the reform of the Family Code, find that women who perceive that their husbands will get custody of land or the house upon a divorce perceive less control over their life and personal interests. Children in households where women perceive they will get less in a divorce settlement also do less well in school relative to children of the same age; girls fare even worse than boys in these households.

In 2003, the Ethiopian government also embarked on a process of community-based land registration, which led to joint certification of husbands and wives, giving stronger land rights to women. Property rights institutions governing land have undergone drastic changes over the past four decades. Ethiopia's land tenure system before 1975 was complex, with substantial regional variation, with the main forms being a communal rist system in the North and a largely feudal system that encouraged absentee landlordism in the South.3 High inequality of landownership contributed to the overthrow of the imperial regime in 1975, after which the Marxist government transferred ownership of all rural land to the state, established peasant associations with wide judicial and administrative powers at the village level, and promoted producer cooperatives, villagization, and resettlement programs. Following the overthrow of the Marxist regime in the early 1990s, intentions to move toward a system of private land ownership did not fully materialize: the 1995 constitution highlights that ownership of land is with the state and upholds the right of every Ethiopian who wants to engage in agriculture to receive inheritable use rights to a piece of land for free, a principle that can be enforced through administrative reallocation of land but that will likely conflict with the goal of ensuring land users' tenure security (Deininger et al., 2008, p. 1789). As of 1997, user rights over land were being allocated by Peasant Associations (PA), the local administrative unit in rural areas, although many regions of the country had not experienced land reallocations in recent years. The 1997 round of the Ethiopian Rural Household survey found that, the land user rights held by the surveyed households, two-thirds actually come directly from the PA (Fafchamps & Quisumbing,

2002, Table 4), with land typically allocated to males prior to marriage. Family was thus not the dominant source of land for surveyed households. Women did, however, occasionally receive land from the PA, thereby suggesting a political willingness to depart from rural norms in the allocation of land to women (Gopal & Salim, 1999).

A 1997 federal proclamation (law) devolved responsibility for land policy to the regions, leading to high inter-regional diversity of key legal provisions, similar to the way that the reform of the Family Law was implemented. Reflecting this relative autonomy of regional land administration, a low-cost land registration and certification program was implemented in 1998-99 in Tigray, the northernmost region, which provided land certificates to more than 80% of the rural farm households (Holden, Deininger, & Ghebru, 2007). The costs were low and affordable because local tools were used in demarcation and measurement of plots, staff with very limited training organized the work, and strong local participation in the implementation was required by the land administration. One-page certificates were issued with the names of the heads of household (the husband for married households), and details about the size, location, and land quality of farm plots, as well as the names of the neighbors for each plot. Female heads of household (widows, divorced and single women) also received certificates in their name for land in their possession. Thus, it can be argued that the land registration strengthened the rights of female-headed households, but not necessarily those of wives in male-headed households. When the land registration was expanded to different regions, the land certificates included maps and, in some regions, photos of the husband and wife (Deininger et al., 2008), which made it more difficult (than signatures) for husbands to sell or rent out land without their wives' consent, particularly in a society with very low literacy rates. The Ethiopian land certification scheme was noteworthy because land administration committees at the kebele level (the smallest administrative unit in Ethiopia) were required to have at least one female member and land certificates were issued after public registration for transparency (Deininger et al., 2008). Thus, it appears that these efforts to strengthen women's land rights are complementary to the changes instituted by the Family Code, which gave equal rights to women and men in terms of marriage, inheritance, and property. We investigate this complementarity in the remainder of this paper.

3. DATA AND SAMPLE DESCRIPTIVES

(a) Data

The ERHS is a panel dataset with seven rounds of data collection. The data collection was coordinated by the Economics Department at Addis Ababa University in collaboration with the Centre for the Study of African Economies at Oxford University and the International Food Policy Research Institute. This paper uses data from the 1997, 2004, and 2009 rounds. The 1997 round contains baseline perceptions of the distribution of assets upon divorce prior to the reform of the Family Code, while the 2004 and 2009 rounds were conducted a year after and six years after the initiation of the land registration effort, respectively. Both authors were involved in the design and analysis of the 1997, 2004, and 2009 rounds of the ERHS.

The ERHS sample consists of about 1,300 households in 15 villages across Ethiopia.4 The ERHS was not designed to be statistically representative of rural Ethiopia as a whole,5 but to represent major agroclimatic zones and farming systems. While sample households within villages were randomly

selected, the villages themselves were chosen to ensure that the major farming systems are represented The ERHS also provides a wealth of longitudinal information on the same households and individuals, who were followed and reinterviewed over 15 years. The location of the sample villages is shown in Figure 1. About a third (32%) of sample households is female-headed, based on self-reported headship, although there is wide variation across the survey villages (Figure 2). The highest rates of female-headship are found in the two northern sites in Tigray (Haresaw and Geblen) and the lowest is in Yetmen in the Amhara region. The rate of female-headship is higher compared to data from the nationally representative Demographic and Health Survey 2011 which reports it as 23.2%. This may be because by 2009, the ERHS households had been followed for over a decade, and as a result the households are older, male heads of households may have died, with their widows (who are typically much younger) taking their place as household heads (also reflected in Table 1).

The surveys collected information on household demographic characteristics, occupation, cropping patterns, perceptions of poverty and well-being, experience with shocks, access to credit, and so on. We present, in Table 1, some of the summary statistics for our sample, disaggregated by the sex of the household head. We also present, wherever possible, the same summary statistics from the DHS 2011 for comparison, also disaggregated by sex of the household head. We disaggregate by sex of household head for two reasons: (1) the relative disadvantage that female-headed households face owing to lower land, assets, human capital, and social capital endowments; (2) the higher vulnerability that female-headed households face to covariate and idiosyncratic shocks. Given these reasons, there is a distinct possibility that reform processes may have differential effects on male- and female-households.

Our previous work on Ethiopia (Kumar & Quisumbing, 2012), as well as data from the DHS, suggests that female-headed households tend to be disadvantaged relative to male-headed households on a number of dimensions. Female heads are, on average, older and less educated than male heads; female heads, on average, have no education, whereas their male counterparts have at least two years of schooling.

Figure 2. Proportion of female-headed households. ¡Source: ERHS 2009.

The gender disparity in schooling is not limited to the education of the head but is also true for the household at large: the average highest education level within female-headed households is 4.76 years, which is about a year and half less than that in male-headed households. Female-headed households also tend to be smaller, with a larger fraction of female members. Because household size is proportional to the amount of labor resources the household controls in a rural area, and because many farm operations (especially plowing) are intensive in male labor, female-headed households tend to be disad-vantaged with respect to labor endowments.6

Female-headed households are also worse off compared to their male counterparts in terms of land and asset ownership. Male-headed households own 2.2 hectares of land, on average, compared to 1.7 hectares for female-headed households. This is true also for the DHS sample, however the difference in amount of land owned is not that stark. Male-headed households also have an average of 9.4 tropical livestock units (TLUs), which is significantly different from female-headed households' holdings of 8.8 TLUs. Sixty percent of male-headed households have at least some oxen, compared to 37% of female-headed households. In the DHS sample, households tend to hold fewer livestock units and the difference in livestock ownership between male- and female-headed households is starker. Despite these differences in land and asset ownership, there is no significant difference between in real

Ethiopian Rural Household Survey Villages

Figure 1. Location of the Ethiopian Rural Household ¡Survey (ERHS) villages. Source: ERHS.

Table 1. Comparison of household characteristics, by sex of household head, Ethiopia ERHS 2009 and DHS 2011

ERHS DHS 2011

Female-headed household Male-headed household p-Value Female-headed household Male-headed household p-Value

Age of head 54.28 52.53 ** 48.22 43.79 ***

Education of head 0.33 2.22 *** 0.66 1.89 ***

Highest grade obtained 4.76 6.28 *** 3.80 4.19 ***

Fraction of female members in household 0.62 0.47 *** 0.72 0.48 ***

Fraction of dependent members in household 0.51 0.52 0.58 0.52 ***

Household size 4.39 6.38 *** 3.61 5.30 ***

Total land owned, hectares 1.73 2.20 *** 2.33 2.72 ***

Total livestock owned, tropical livestock units 8.82 9.39 *** 2.79 4.26 ***

Fraction of households that own any oxen 0.37 0.61 *** n/a n/a

Real per capita consumption in 2004 (ETB, 1994 prices) 94.00 91.00 n/a n/a

Real per capita consumption in 2009 (ETB, 1994 prices) 59.00 60.00 n/a n/a

Fraction of households that are members of an iddir 0.76 0.89 *** n/a n/a

Network size 8.61 11.41 *** n/a n/a

Fraction of households that have a bank account 0.05 0.06 0.04 0.06 **

Number of sources from which a household can borrow 1.32 1.57 n/a n/a

Source: ERHS and authors' computations.

Notes: *** and ** represent statistical significance at 1% and 5%, respectively. ETB = Ethiopian birr.

per capita consumption between male- and female-headed households (Table 1).

Similar to the disparities in land and physical asset ownership, there are also differences in measures of social capital, namely network size and membership in an iddir (burial society or funeral association). Network size refers to the number of people that survey respondents say they can rely on in times of need. Table 1 shows that male-headed households, on average, have larger networks, and that male-headed households are more likely to be members of an iddir. In terms of access to financial institutions and credit, the proportion of households holding a bank account is quite small (about 5%) and is not substantially different for the two groups. However, male-headed households have access to a greater number of sources from which they can borrow.

Female-headed households also tend to be more vulnerable to covariate and idiosyncratic shocks. Dercon, Hoddinott, Krishnan, and Woldehanna (2012) find that per capita consumption in female-headed households is significantly negatively affected by drought shocks (the impact on male-headed households is insignificant), and in other work (Kumar & Quisumbing, 2013) related to the 2007-08 global food price increases, we find that female-headed households are more likely to report experiencing a reduction in asset holdings, household income, or consumption due to high food prices. Most relevant to this paper, Dercon et al. (2012) also find that per capita consumption in female-headed households is significantly lower following the death of a household head, spouse, or another person in the household. Since the Family Law (to be discussed below) protects women's rights to property following marital dissolution, whether in cases of death or divorce, this finding suggests that female-headed households stand to benefit from the additional protection offered by these legal reforms. Moreover, reforms that equalize men's and women's property rights not only improve bargaining power within a marital union but also secure access to assets beyond the marriage (whether after the husband's death or upon a divorce). Therefore, such reforms are not only important for women that are currently in a union, but also for those that are not. Finally, our analysis of the land registration also examines differential participation and awareness of male- and female-headed households because it is possible that the latter

were excluded from the process owing to their generally lower social status or lack of information about the process itself.

4. FAMILY LAW

In this section we use data from the 1997 and 2009 rounds of the ERHS to assess changes in perceptions about the allocation of assets upon divorce.

In the 1997 and 2009 survey rounds, we asked female heads (in female-headed households) or the spouses of male heads (in male-headed households) a series of hypothetical questions designed to elicit perceptions regarding the disposition of assets and custody of younger and older children upon divorce when (1) neither the husband nor the wife was at fault, (2) the husband was at fault, or (3) the wife was at fault. The 1997 responses can be interpreted as baseline perceptions prior to the reform of the Family Code, analyzed in Fafchamps and Quisumbing (2002), while the 2009 responses are post-reform perceptions.7 Although divorce is common, it does not happen often enough to generate enough responses if we asked about each respondent's divorce experiences. Thus, our questions, which were about hypothetical situations, were discussed to obtain respondents' perceptions of local norms regarding divorce.

Figures 3-8 compare these responses of perceptions of asset devolution for 1997 and 2009. Figure 3 shows the fraction of spouse/female heads that perceived equal division of assets and equal custody upon a no-fault divorce. In 1997 about 40% of the women perceived that land would be divided equally between the husband and wife upon a no-fault divorce and this percentage increased to more than 80% in 2009. We observe similar patterns for the allocation of the house and livestock. Figure 4 shows the percentage of women that perceived that all of the assets would be given to the wife in case of a no-fault divorce. A very small fraction of women perceived that other assets would be given in their entirety to the wife in case of a no-fault divorce and there are no significant differences across the two rounds. A very low proportion of female heads/spouses believe that child custody wouldbe shared equally in a no-fault divorce, even if this proportion increased slightly during 1997-2009 (Figure 3). However, a

0.90 0.80 0.70 0.60 0.50

11997 2009

Young children

Older children

Livestock Livestock Livestock HH

from from wife acquired Utensils husband after

marriage

Figure 3. Perceptions of allocation of assets and custody of children: Divided half-half in case of a no-fault divorce. ¡Source: ERHS 2009.

1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0.00

1997 2009

Young children

Older children

Livestock Livestock Livestock HH

from from wife acquired Utensils husband after

marriage

Figure 4. Perceptions of allocation of assets and custody of children: Given to the wife in case of a no-fault divorce. Source: ERHS 2009.

0.50 0.40 0.30 0.20 0.10 0.00

11997 2009

Young Older children children

Livestock Livestock Livestock HH

from from wife acquired Utensils husband after

marriage

Figure 5. Perceptions of allocation of assets and custody of children: Divided half-half in case of divorce when wife is at fault. Source: ERHS 2009.

Young Older children children

House Livestock Livestock Livestock HH

from from wife acquired Utensils husband after

marriage

Figure 6. Perceptions of allocation of assets and custody of children: Given to the husband in case of divorce when wife is at fault. Source: ERHS 2009.

large proportion believes that custody of younger children The next two figures show women's perceptions of allocation remains with the wife, and this increased substantially between of assets in case of a divorce when the wife is at fault. Figure 5 the two rounds (Figure 4). shows the percentage of women that perceived that assets would

Young Older Land House Livestock Livestock Livestock HH Utensils children children from from wife acquired

husband after

marriage

Figure 7. Perceptions of allocation of assets and custody of children: Divided half-half in case of divorce when husband is at fault. ¡Source: ERHS 2009.

0.80 0.60 0.40 0.20 0.00

11997 2009

Young Older children children

House Livestock Livestock Livestock HH Utensils from from wife acquired husband after

marriage

Figure 8. Perceptions of allocation of assets and custody of children: Given to the wife in case of divorce when husband is at fault. ¡Source: ERHS 2009.

be divided equally in case of a divorce when the wife is at fault. In 1997, the fraction of women that perceived equal division of assets in such a scenario is quite small with the exception of livestock acquired after marriage and household utensils. This, however, changes dramatically in 2009 where close to 30% of women perceive equal division of a number of assets. Figure 6 presents the corresponding fraction of women who perceived that the assets would be given to the husband in case of a divorce when the wife is at fault. As expected, in 1997 a large fraction of women perceived that most assets would be given to the husband in such a situation but again these percentages have decreased drastically in 2009. While there is a slight increase in those who believe that custody of children would be shared equally (Figure 5), in most cases where the wife is at fault, there is a slight increase in those believing the husband would have custody of the younger children, while there is a decrease the proportion of those who think the husband would have custody of the older children. The last set of figures shows women's perceptions regarding the allocation of assets upon a divorce when the husband is at fault. For most asset categories, in 1997 a large fraction of women perceived that the asset would be divided equally between the husband and the wife in case of a divorce when the husband is at fault (Figure 7). These percentages are greater than the corresponding ones for a divorce that occurs when the wife is at fault. Interestingly, these women do not necessarily perceive that the assets would be given to the wife when the husband is at fault in 1997, although these perceptions change a little in favor of women in 2009 (Figure 8). Respondents believe, however, that the wife would have custody of children when the husband is at fault, with the proportion of those holding this belief increasing markedly in 2009.

These above-mentioned figures clearly show that regardless of who is at fault when a divorce occurs, there is a shift in perceptions toward splitting property evenly between the husband and

the wife, with the exception of children, who are perceived to stay with the wife. This change, probably driven by the changes in the Family Law that occurred in 2000, tends to be observed throughout the sample, albeit with some regional variation.

We construct a variable that indicates whether female-household heads/spouses in male-headed households perceived that allocations of land and livestock acquired after marriage shifted toward equal allocations across spouses in case of a no-fault divorce (Table 2). On average, a large fraction of households (44% and 35%, respectively) moved toward perceiving a more equal distribution of land and livestock in case of a no-fault divorce, although there is substantial regional variation. In Tigray, the fraction of households that moved toward perceiving a more equal distribution of assets is relatively small, about 14%, primarily because initial conditions were already more egalitarian. In 1997, local norms regarding the distribution of assets after divorce were already more equal in Tigray, with about 40% of households perceiving that land is allocated equally between the couple upon a no-fault divorce. On the other end of the spectrum lies SNNPR, where almost two-thirds of the households changed their response to reflect perceptions of more equal allocation. This is also due to initial conditions: a very small proportion of households reported perceptions of equal division in 1997. These statistics show that not only did the greatest shifts in perceptions toward more equal allocations occur in the regions where the distribution was most unequal, but improvement was perceived even in the regions with relatively gender-fair post-divorce allocations.

We examine whether individual and household characteristics affect changes in perceptions regarding the allocation of land and livestock upon divorce using regression analysis. Because perceptions are subjective, they may be influenced by individual and household characteristics; persons from wealthier households, for example, may have more to lose if a

Table 2. Summary statistics: Changes in perceptions regarding allocation of land and livestock upon divorce, 1997-2009

Percentage of households whose perceptions shifted toward All Female-headed Male-headed

household household

Equal allocation of land upon a no-fault divorce 44 40 46

Equal allocation of livestock acquired after marriage upon a no-fault divorce 35 34 36

Tigray

Equal allocation of land upon a no-fault divorce 13 17 7

Equal allocation of livestock acquired after marriage upon a no-fault divorce 14 18 9

Amhara

Equal allocation of land upon a no-fault divorce 30 33 28

Moved toward equal allocation of livestock acquired after marriage upon a no-fault divorce 21 24 20

Oromiya

Equal allocation of land upon a no-fault divorce 52 48 54

Equal allocation of livestock acquired after marriage upon a no-fault divorce 35 33 36

Equal allocation of land upon a no-fault divorce 62 54 66

Equal allocation of livestock acquired after marriage upon a no-fault divorce 58 61 57

Source: ERHS and authors' computation.

Table 3. Regression results for changes in perceptions regarding allocation of land and livestock upon divorce

Variables Moved to split land half-half Moved to split livestock half-half

All Male-headed Female-headed All Male-headed Female-headed

household household household household

(1) (2) (3) (4) (5) (6)

Age of household head 0.000 0.002 -0.003 0.001 0.001 0.003

(0.002) (0.002) (0.003) (0.002) (0.002) (0.003)

Sex of household head (1 = male, 0 = female) 0.041 0.048

(0.049) (0.048)

Number of years of schooling of the head 0.019*** 0.021*** -0.004 0.011* 0.009 0.015

(0.007) (0.008) (0.023) (0.007) (0.007) (0.022)

Highest grade obtained in household -0.011* -0.012 -0.008 -0.018*** -0.017*** -0.015

(0.006) (0.007) (0.010) (0.005) (0.007) (0.010)

Total plot area in hectares, 2004 survey -0.019* -0.025** 0.003 -0.032*** -0.038*** -0.010

(0.010) (0.011) (0.027) (0.009) (0.009) (0.028)

Fraction of total land that is good or medium quality 0.112** 0.136* 0.087 0.150*** 0.198*** 0.109

(0.057) (0.074) (0.090) (0.054) (0.066) (0.089)

Dummy for land quartile 1, 2004 survey -0.001 0.075 -0.141 -0.022 0.027 -0.061

(0.047) (0.057) (0.093) (0.045) (0.055) (0.090)

Dummy for land quartile 2, 2004 survey -0.035 -0.011 -0.118 -0.051 -0.055 -0.037

(0.047) (0.055) (0.098) (0.045) (0.051) (0.096)

Dummy for land quartile 3, 2004 survey -0.008 -0.029 0.014 0.012 -0.001 0.028

(0.045) (0.051) (0.101) (0.042) (0.046) (0.095)

Total livestock holdings, 2004 survey (tropical livestock units) -0.012** -0.011* -0.020* -0.017*** -0.015*** -0.028***

(0.005) (0.006) (0.010) (0.005) (0.005) (0.010)

Member of an iddir, 2004 survey 0.134*** 0.131** 0.109 0.090** 0.056 0.079

(0.043) (0.057) (0.072) (0.042) (0.054) (0.070)

Network size, 2004 survey -0.000 -0.001 0.005 -0.000 -0.001 0.007

(0.001) (0.001) (0.006) (0.001) (0.001) (0.006)

Observations 972 659 313 966 660 306

R-squared 0.072 0.087 0.082 0.088 0.136 0.087

Source: ERHS and authors' computation.

Notes: ***, **, and * represent statistical significance at 1%, 5%, and 10%, respectively. The number of observations in columns (1) and (4) represents the sample for which we have all set of covariates and outcome variables. The number of observations in columns (2) (3) and (5) (6) represents the male-(female-) headed households in the sample.

post-divorce allocation was not in their favor. Alternatively, division and (b) whether her perception regarding the distribu-individuals controlling more assets within the marriage may tion of livestock after a divorce shifted toward equal division, perceive (perhaps wishfully) that the divorce allocations post- as a function of the sex and age of the head, the number of reform would still favor them. Table 3 presents regressions years of schooling of the head, the highest grade obtained in on (a) whether the female head/spouse's perception regarding the household, the fraction of good quality land, plot area in the distribution of land after a divorce shifted toward equal 2004, a dummy for relative wealth in the village (as measured

Table 4. Regression results for changes in perceptions regarding allocation of land and livestock upon divorce with village fixed effects

Variables Moved to split land half-half Moved to split livestock half-half

All Male-headed Female-headed All Male-headed Female-headed

household household household household

(1) (2) (3) (4) (5) (6)

Age of household head -0.000 0.001 -0.002 -0.000 0.000 0.003

(0.001) (0.002) (0.003) (0.001) (0.002) (0.002)

Sex of household head (1 = male, 0 = female) 0.050 (0.047) 0.018 (0.044)

Number of years of schooling of the head 0.006 0.005 0.000 0.002 -0.003 0.024

(0.007) (0.007) (0.022) (0.006) (0.007) (0.020)

Highest grade obtained in household -0.004 -0.007 0.007 -0.003 -0.007 0.008

(0.006) (0.007) (0.010) (0.005) (0.006) (0.010)

Total plot area in hectares, 2004 survey -0.044*** -0.035*** -0.092*** -0.028*** -0.027*** -0.031

(0.010) (0.010) (0.033) (0.009) (0.009) (0.035)

Fraction of total land that is good or medium quality 0.002 0.026 -0.029 -0.017 0.028 -0.039

(0.054) (0.071) (0.089) (0.055) (0.072) (0.083)

Dummy for land quartile 1, 2004 survey -0.034 0.004 -0.171** -0.008 0.019 -0.025

(0.042) (0.052) (0.086) (0.041) (0.051) (0.084)

Dummy for land quartile 2, 2004 survey -0.033 -0.031 -0.071 -0.003 -0.021 0.063

(0.043) (0.051) (0.092) (0.040) (0.044) (0.090)

Dummy for land quartile 3, 2004 survey -0.015 -0.026 0.007 0.020 0.022 0.006

(0.040) (0.045) (0.094) (0.037) (0.043) (0.085)

Total livestock holdings, 2004 survey (tropical livestock units) -0.002 0.005 -0.016 0.000 0.007 -0.016*

(0.006) (0.007) (0.010) (0.005) (0.006) (0.008)

Member of an iddir, 2004 survey -0.123* -0.129 -0.150 -0.170*** -0.259*** -0.111

(0.069) (0.093) (0.115) (0.064) (0.084) (0.107)

Network size, 2004 survey -0.002 -0.002 -0.000 -0.001 -0.001 0.005

(0.001) (0.001) (0.006) (0.001) (0.001) (0.005)

Observations 972 659 313 966 660 306

R-squared 0.243 0.287 0.233 0.283 0.318 0.338

Source: ERHS and authors' computation.

Notes: ***, **, and * represent statistical significance at 1%, 5%, and 10%, respectively. The number of observations in columns (1) and (4) represents the sample for which we have all set of covariates and outcome variables. The number of observations in columns (2) (3) and (5) (6) represents the male-(female-) headed households in the sample.

by the dummy variables for land quartiles within the village, with quartile 1 being the lowest 25%), total livestock holdings in 2004, whether the household belonged to an iddir in 2004, and network size in 2004. We use 2004 values to control for the possibility that 2009 household characteristics may be codetermined by factors affecting perceptions regarding divorce distributions. The regressions are estimated separately for currently married females (spouses in male-headed households) and female-headed households; it is possible that actual experiences of divorce or marital dissolution could create differences between responses of currently and previously married women. Regression results (Table 3) show that among wives in male-headed households, those with large quantities of land or livestock are less likely to have changed their perceptions regarding whether or not land or livestock will be equally allocated after divorce. This suggests that households in which the husbands have more at stake are less likely to report having shifted their perceptions toward equal division upon divorce. This variable is not significant for female-headed households in the case of land, but is significant in the case of livestock. Female-headed households that own more land are less likely to have experienced a shift in perceptions toward equal division of livestock upon divorce. It is possible that wealthier males, precisely because they have more wealth to lose, will resist efforts to achieve greater equality between ex-spouses after divorce.

Are these results robust to the inclusion of village fixed effects that may capture unobserved, time-invariant social norms regarding the division of property upon divorce?8 To

test this, we add village fixed effects (Table 4) to the regressions. Although some results change, some key results remain. Having larger areas of land still reduces the probability that perceptions of the spouses of male heads and female heads will shift toward equal division of land and livestock upon divorce, but the impact of landholdings on the perceptions of female-headed households regarding the division of livestock is no longer significant. Neither do livestock holdings influence perceptions regarding the division of land or livestock after marital dissolution when village fixed effects are included. This shows that within-village variation in land area owned by female-headed households and livestock ownership in general is driving the results as opposed to variation across villages. We do, however, observe that membership in an iddir reduces the probability that the perceptions regarding the division of land or livestock after a divorce shift toward equal division. Because iddirs are traditional risk-sharing institutions, it is possible that they subscribe to traditional norms that may dampen the effect of reforms, although this remains a topic to be investigated further.

5. THE LAND REGISTRATION PROCESS

In this section we use data from ERHS 2009 to examine whether male- and female-headed households differ in terms of land owned and cultivated, and whether male or female household heads differ in their awareness of and participation in the land registration process. In 2009 we interviewed all

Table 5. Characteristics of land owned and cropped

Female-headed household Male-headed household p-Value

Total plot area, hectares 1.60 2.00 **

Total cropped area, hectares 1.19 1.69 ***

Fraction of total land that is cropped 0.71 0.85 ***

Fraction of cropped land that is good or medium quality 0.83 0.89 ***

Fraction of total land that is good or medium quality 0.83 0.88

Fraction of cropped area operated by women 0.82 0.01 ***

Fraction of plot area operated by women 0.84 0.01

Fraction of cropped area registered 0.95 0.97 **

Fraction of total land area registered 0.96 0.97

Source: ERHS and authors' computations.

Notes: ***, **, and * represent statistical significance at 1%, 5%, and 10%, respectively.

heads of households on their awareness, participation and perceptions of the land registration process. Table 5 indicates that male-headed households hold more land (have larger plot sizes), of which a larger proportion is cultivable, compared with female-headed households. The larger areas and proportions of land cultivated may be partly because of better land quality and the fact that larger plot sizes are more viable for cultivation. Women in male-headed households are very rarely in charge of operating land, but the converse cannot be said for female-headed households, where about one-fifth of the time men are operating the land.9 This may occur due to cultural norms that prohibit women from plowing land because it is perceived to be too strenuous or culturally inappropriate (Frank, 1999). Male-headed households are also more likely to have a larger fraction of their land registered.

Next, we explore the differences in awareness, participation, and perception of the land registration process between the two types of households, for the entire sample and separately by region (Table 6). In Tigray, only about 3% of the household heads in our sample reported any awareness about the land registration process; therefore, we do not have useful estimates of participation by these households.10

Male heads of households were much more likely to have heard of the land registration process. Almost all male household heads (90%) had heard of the process, compared to about three-quarters of female heads. There is, however, some regional variation. In Oromiya, female household heads were just as likely as the male heads to have heard about the process. We find that, throughout our survey villages, male household heads were more aware of public information meetings held before the land registration process, were more likely to have attended such meetings (as well as a greater number of meetings), and were more likely to have received some written material about the program.

Most household heads acknowledged that their plot boundaries were well demarcated before the land registration process started, and about a quarter to a third of the households reported facing land disputes before the registration process. They perceive the land title as a protection against encroachment and agree that the number of land disputes decreased after the land registration process was complete. In Oromiya, household heads do not value the title so much as a means of protection against encroachment (42-48%, compared to the sample average of 62-65%), probably because their plot boundaries were clearly demarcated even before the process was implemented. All household heads, regardless of sex, believe that the title increases their incentive for planting trees (more so for male heads) and increases the probability of receiving compensation in case of appropriation. Both male and female household heads also believe that having a land

certificate improves the position of women. All in all, the data in Table 6 suggest that most household heads perceive the land registration process as valuable.

The major difference between male- and female heads of households lies in their knowledge of and participation in the program. We use information on the awareness of the land registration process, about public meetings held before the registration process started, whether household members attended these meetings, the number of meeting attended and if they received any written materials on the process to construct an index of participation that ranges from 0 to 5, where 0 represents no awareness or participation and 5 represents a high level of awareness and participation. This is presented graphically in Figure 9 for the entire sample and then for each region, disaggregated by the sex of the household head. This shows that male-headed households on average have a higher index of participation compared to female-headed households. This disparity is most striking in SNNPR.

We estimate alternative regression models that examine the determinants of awareness about the land registration process, participation in the process by way of attending meetings, and the index of participation in the land registration process, with lagged household characteristics as regressors, as well as a variable indicating the presence of female members in the Land Administration Committee (LAC).11 We estimate the regressions for awareness of and participation in the land registration process as linear probability models and the regressions for index of participation as an ordered logit model. Household characteristics, such as sex of the household head, education levels, size and quality of land, and relative wealth within the community, can affect the household's extent of awareness and participation in the land registration program. These characteristics include the age and sex of the household head, years of schooling of the head and the highest grade obtained in the household, area and quality of land, asset holdings, networks, iddir membership, access to credit, and relative wealth in the village (as measured by the dummy variables for land quartiles within the village, with quartile 1 being the lowest 25%). With the exception of household demographic characteristics, which refer to the current round, we use lagged household characteristics because current household characteristics (for example, asset holdings) could be correlated with participation in the land registration effort. We also include two categorical variables that capture the respondent's perception regarding his or her power to change the course of his or her life. This variable is similar to those included in surveys like the Gallup World Poll and World Values Survey (for details see http://www.gallupworld-poll.com/content/24046/About.aspx and http://www.world-valuessurvey.org/index_html). While the power to change

Table 6. Land registration process: Knowledge and participation

Whole sample Amhara Oromiya SNNPR

Female-headed Male-headed p-value Female-headed Male-headed p-value Female-headed Male-headed p-value Female-headed Male- headed p-value

household household household household household household household household

Are aware of the land 0.75 0.9 *** 0.9 0.96 *** 0.96 0.96 0.83 0.95 ***

registration process

Public information meetings were 0.79 0.91 0.83 0.91 0.74 0.9 *** 0.86 0.93 *

held before the land registration

program started

Any member of the household 0.81 0.89 0.83 0.87 0.8 0.9 ** 0.83 0.91 *

attended any of these meetings

Number of these meetings 2.19 2.71 2.28 2.74 2.07 2.66 ** 2.3 2.79 **

attended

Received any written material on 0.15 0.22 *** 0.17 0.18 ** 0.03 0.08 ** 0.3 0.47 ***

this program

The plot borders were clearly 0.88 0.88 0.84 0.82 0.92 0.94 0.89 0.86

demarcated before the land

registration

Faced border disputes before the 0.28 0.26 0.31 0.22 0.27 0.28 0.29 0.28

land registration

The plot borders were clearly 0.97 0.96 0.96 0.93 0.96 0.97 0.99 0.96 *

demarcated during the land

registration

The land registration reduced the 0.39 0.38 0.42 0.41 0.27 0.24 0.55 0.56

number of border disputes

during the process

The land registration reduced the 0.39 0.39 0.45 0.42 0.26 0.25 0.49 0.54

number of border disputes after

the process was completed

Having a certificate protects 0.62 0.65 0.8 0.8 0.42 0.48 0.72 0.76

against encroachment on land by

neighbors

Need for a new land demarcation 0.35 0.37 0.32 0.44 0.3 0.3 0.46 0.41

to make borders clearer

Have sufficient witnesses that can 0.94 0.92 0.93 0.91 ** 0.92 0.9 0.98 0.94 **

confirm the borders of their plots

in case it was contested

Interested in planting trees on 0.77 0.81 0.81 0.86 0.7 0.73 0.83 0.87

any of their plots

Having the land certificate 0.74 0.81 0.79 0.85 0.65 0.74 ** 0.83 0.87

increases their incentive to plant

Having a certificate will increase 0.92 0.92 0.93 0.9 0.95 0.92 0.87 0.94 *

the possibility of obtaining

compensation in case land is

appropriated

Having a land certificate 0.94 0.95 0.98 0.99 0.9 0.93 0.97 0.94

improves the position of women

G w N D

Source: ERHS and authors' computations.

Notes: ***, **, and * represent statistical significance at 1%, 5%, and 10%, respectively.

one's life could potentially be affected by the Family Law reform (and therefore be endogenously determined), our previous work (Kumar & Quisumbing, 2012) suggests that changes in perceptions of divorce distributions do not affect respondents' perceptions of their ability to change the course of their lives, justifying its inclusion in this regression. We run this regression with and without village fixed effects to control for unobserved village-level characteristics.

The regressions reported in Table 7 were run for the pooled sample with a dummy for the sex of the household head, but this variable was not significant. The regression estimates, show that, on average, membership in an iddir and the presence of female members in the LAC increases knowledge of and attendance at meetings during the land registration process. This is reasonable because the iddir is a type of social network that facilitates information-sharing in addition to its insurance objectives. The presence of female members in the LAC may

Table 7. Regression results for knowledge and participation in land registration process with village fixed effects

Variables All Male-headed household Female-headed

household

Knowledge Attendance Index Knowledge Attendance Knowledge Attendance

(1) (2) (3) (4) (5) (6) (7)

Age of household head 0.001 -0.001 -0.005 0.000 -0.002 0.001 0.000

(0.001) (0.001) (0.006) (0.001) (0.002) (0.001) (0.003)

Gender of household head (1 = male, 0 = female) 0.034 (0.023) 0.020 (0.040) 0.296 (0.191)

Number of years of schooling of the head -0.001 -0.000 -0.016 -0.001 0.004 -0.010 -0.018

(0.004) (0.005) (0.031) (0.003) (0.006) (0.016) (0.016)

Highest grade obtained in household -0.004 0.012*** 0.021 -0.005* 0.009* -0.004 0.019*

(0.003) (0.004) (0.024) (0.003) (0.005) (0.006) (0.011)

Total plot area in hectares, 2004 survey -0.008 -0.004 -0.068 0.004 -0.007 -0.009* 0.027

(0.007) (0.011) (0.043) (0.005) (0.011) (0.005) (0.043)

Fraction of total land that is good or medium quality 0.032 0.047 -0.063 0.039 0.049 0.048 -0.011

(0.031) (0.057) (0.298) (0.041) (0.064) (0.043) (0.133)

Total livestock holdings, 2004 survey (tropical livestock -0.001 -0.004 -0.017 -0.002 0.000 -0.002 -0.023*

units)

(0.002) (0.005) (0.021) (0.002) (0.005) (0.004) (0.012)

Dummy for land quartile 1, 2004 survey -0.011 -0.070* -0.696*** 0.007 -0.049 -0.057 -0.144*

(0.021) (0.038) (0.196) (0.022) (0.045) (0.059) (0.083)

Dummy for land quartile 2, 2004 survey 0.007 -0.038 -0.183 0.003 -0.049 -0.016 -0.084

(0.020) (0.034) (0.191) (0.019) (0.040) (0.057) (0.074)

Dummy for land quartile 3, 2004 survey 0.022 -0.062* -0.255 0.010 -0.028 0.021 -0.260***

(0.018) (0.032) (0.179) (0.019) (0.035) (0.054) (0.087)

Presence of female members in the LAC 0.912*** 0.469* 6.090*** 0.933*** 0.179 0.911*** 1.066***

(0.036) (0.262) (0.668) (0.048) (0.379) (0.064) (0.184)

Household head perceives to have some power to change 0.021 0.104** 0.276 0.056 0.085 -0.057 0.161*

the course of his/her life

(0.032) (0.051) (0.224) (0.040) (0.064) (0.044) (0.084)

Household head perceives to have a lot of power to 0.039 0.057 0.247 0.060 0.033 -0.000 0.142*

change the course of his/her life

(0.031) (0.052) (0.223) (0.040) (0.065) (0.039) (0.083)

Member of an iddir, 2004 survey 0.062* 0.071 0.610** 0.045 0.047 0.031 0.208

(0.038) (0.071) (0.256) (0.043) (0.082) (0.054) (0.159)

Network size, 2004 survey 0.000 -0.001 -0.002 -0.000 -0.000 0.003 0.000

(0.000) (0.001) (0.004) (0.000) (0.001) (0.003) (0.006)

Household member has a bank account -0.023 0.008 0.312 -0.005 -0.038 -0.039 0.007

(0.039) (0.056) (0.383) (0.043) (0.068) (0.070) (0.129)

Observations 1,062 793 1,064 737 602 325 191

R-squared 0.675 0.127 0.650 0.115 0.748 0.342

Source: ERHS 2009 and authors' computation.

Notes: ***, **, and * represent statistical significance at 1%, 5%, and 10%, respectively. The number of observations in columns 1-3 represents the sample for which we have all set of covariates and outcome variables. Column 2 has fewer observations than columns 1 and 3 because the outcome variable— attendance—is defined conditional on knowledge. The number of observations in columns (4) (6) and (5) (7) represents the male- (female-) headed households in the sample.

3.5 3 2.5 2 1.5 1 0.5 0

I Female _ Headed

Male Headed

Whole Sample

Tigray Amhara Oromiya SNNPR

Figure 9. Index of participation in land registration process. ¡Source: ERHS

Table 8. Comparing characteristics of villages with Land Administrative Committees with and without women members

No women in LAC Women in LAC p-value

Average age of household head, 2009 53.05 54.02 0.60

Fraction of male headed households, 2009 0.76 0.66 0.08*

Average years of schooling of head, 2004 1.85 1.55 0.61

Average of highest grade obtained in household, 2004 6.08 6.76 0.35

Average household size, 2009 6.42 6.24 0.62

Average land area owned, 2004 2.19 0.97 0.02**

Real per capita consumption in 2004 (ETB, 1994 prices) 97.57 81.63 0.45

Average fraction of land that is of good quality, 2009 0.90 0.75 0.04**

Average fraction of land registered, 2009 0.96 0.95 0.57

Average livestock holdings, tropical livestock units, 2004 4.38 2.47 0.10*

Fraction of heads that perceive to have some power to change the course of their life, 2009 0.33 0.31 0.71

Fraction of heads that perceive to have a lot power to change the course of their life, 2009 0.56 0.65 0.12

Average fraction of households that are member of iddirs, 2004 0.9 0.64 0.11

Average network size, 2004 9.1 6.66 0.17

Average fraction of households that have a bank account, 2004 0.04 0.06 0.47

No. of villages 12 6

Source: ERHS 2009 and authors' computation.

Notes: ** and * represent statistical significance at 5% and 1%, respectively.

provide a channel of information that women may feel more comfortable accessing, and also improves their participation. Surprisingly, households with higher schooling levels are less likely to know of the land registration process, possibly because these households may be less likely to be intensively involved in agriculture. Being in the top land quartile makes the household more likely to have attended a meeting during the land registration process compared to those in the third quartile. These households, by definition, own a larger amount of land and have more to gain out of attending such meetings. Household heads who think they have some power to change their circumstances are more likely to attend these meetings compared to those who think they have no control over circumstances. This is an interesting finding suggesting that individuals that feel they have no control over their circumstances do not find it useful to attend the meetings - these individuals probably do not perceive much benefit from such changes in tenure security. The index is higher for households that live in villages with at least one female member in the LAC, are members of an iddir, and are in the top land quartile within the village.

The coefficients on interaction terms when we run a model (not reported) with all covariates interacted with the sex of the household head are jointly significant, indicating that the impact of these variables varies by sex of the household head for the awareness and participation in the land registration process. For ease of exposition and interpretation, we estimate the regressions separately for male-and female-headed households (reported in Table 7).

In terms of knowledge of the land registration process, the characteristics that differ across male- and female-headed households are highest grade obtained in the household and total plot area. For male-headed households, education has a negative effect, whereas this effect is positive (although not significant) among the female-headed households. Also, female-headed households with smaller amounts of land are more likely to have heard about the land registration process, which is not the case among the male-headed households. This interesting distinction between male- and female-headed households indicates that the more vulnerable female-headed households who cultivate small land holdings are more likely to have heard of the land registration process. For the attendance regressions, the main difference comes from total livestock holdings, being in the third land quartile in the village, and presence of female members in the LAC. Female house-

hold heads with large livestock holdings (and in the third land quartile) are less likely (than those in the fourth quartile) to have attended a meeting. Heads from households with larger livestock holdings are less likely to attend meetings because of the high opportunity cost of livestock cultivation, which tends to be labor-intensive, and because these households are more diversified into livestock products. An interesting finding is that the presence of female members in the LAC encourages participation by female-headed households and does not appear to discourage participation by male-headed households. This indicates that having female members in the LAC has a net positive impact on attendance at meetings relating to land registration.

Are there systematic differences between villages that have women members on their LACs and those from villages that do not? Summary statistics presented in Table 8 show that villages that have women members on their LAC have also on average a greater fraction of households headed by women, with smaller plots of land, and fewer livestock units. In particular, average household land holding in villages that do not have women in their LAC is more than double as compared to that in villages that have women members in the LAC and the former are also more likely to have better quality land. This shows that women tend to be present in LACs in relatively poorer communities with higher prevalence of female-headship, and smaller and poorer quality landholdings.

Do policy reforms reinforce each other? Table 9 presents regressions on the change in perceptions regarding distribution of land and livestock upon divorce, with additional variables that capture the land reform registration effort as regressors.12 This is an augmented version of the regressions presented in Table 3, which attempts to test whether the land registration effort has an incremental impact on divorce perceptions. Controlling for all previously included regressors, we find that awareness about the land registration process is positively correlated with the shift in perceptions toward equal division of land and livestock upon divorce, especially for male-headed households, suggesting that interventions can reinforce each other. Having at least one female member in the LAC also is positively correlated with the shift in perception toward equal allocation of land among female-headed households and livestock allocation for all samples. It is possible that the positive impact of women in the LAC reflects village characteristics associated with having a female LAC member in the

Table 9. Regression results for changes in perceptions regarding allocation of land and livestock upon divorce

Variables Moved to split land half-half Moved to split livestock half-half

All Male-headed Female-headed All Male-headed Female-headed

household household household household

(1) (2) (3) (4) (5) (6)

Age of household head 0.000 0.002 -0.003 0.001 0.002 0.002

(0.002) (0.002) (0.003) (0.001) (0.002) (0.003)

Sex of household head (1 = male, 0 = female) 0.030 (0.048) 0.058 (0.046)

Number of years of schooling of the head 0.019*** 0.023*** -0.006 0.008 0.007 0.011

(0.007) (0.008) (0.022) (0.006) (0.007) (0.023)

Highest grade obtained in household -0.007 -0.009 0.001 -0.007 -0.009 -0.000

(0.006) (0.007) (0.010) (0.005) (0.006) (0.010)

Total plot area in hectares, 2004 survey -0.031*** -0.032*** -0.025 -0.062*** -0.061*** -0.063**

(0.011) (0.012) (0.029) (0.011) (0.012) (0.031)

Fraction of total land that is good or medium quality 0.044 0.062 0.044 0.037 0.087 0.016

(0.057) (0.074) (0.091) (0.054) (0.068) (0.086)

Dummy for land quartile 1, 2004 survey -0.015 0.058 -0.173* -0.048 -0.005 -0.114

(0.047) (0.056) (0.094) (0.043) (0.053) (0.087)

Dummy for land quartile 2, 2004 survey -0.040 -0.017 -0.132 -0.060 -0.064 -0.065

(0.046) (0.054) (0.099) (0.042) (0.048) (0.094)

Dummy for land quartile 3, 2004 survey -0.017 -0.028 -0.012 -0.001 -0.002 -0.014

(0.045) (0.050) (0.102) (0.040) (0.045) (0.092)

Total livestock holdings, 2004 survey (tropical livestock units) -0.013** -0.012* -0.022** -0.019*** -0.017*** -0.029***

(0.005) (0.006) (0.011) (0.005) (0.005) (0.010)

Presence of female members in the LAC 0.081** 0.038 0.214*** 0.301*** 0.262*** 0.375***

(0.041) (0.048) (0.079) (0.035) (0.042) (0.069)

Member of an iddir, 2004 survey 0.011 -0.021 0.043 -0.059 -0.072 -0.063

(0.053) (0.069) (0.089) (0.049) (0.062) (0.079)

Network size, 2004 survey -0.001 -0.001 0.003 -0.001 -0.001 0.004

(0.001) (0.001) (0.006) (0.001) (0.001) (0.006)

Aware of the land registration process 0.203*** 0.329*** -0.020 0.123** 0.168** 0.014

(0.060) (0.079) (0.099) (0.051) (0.066) (0.081)

Observations 970 657 313 964 658 306

R-squared 0.094 0.117 0.108 0.166 0.198 0.173

Source: ERHS and authors' computation.

Notes: ***, **, and * represent statistical significance at 1%, 5%, and 10%, respectively. The number of observations in columns (1) and (4) represents the sample for which we have all set of covariates and outcome variables. The number of observations in columns (2) (3) and (5) (6) represents the male-(female-) headed households in the sample.

first place. However, when we include village fixed effects (Table 10), we find that the positive impact of the presence of females in the LAC on shifting perceptions toward an equal split in both land and livestock for both male- and female-headed households is robust to the inclusion of village fixed effects. This indicates that, even controlling for local norms regarding the distribution of assets upon divorce, as well as the relatively poor local environments of villages with females on the LAC, the presence of females in an important village-level committee may provide support to women who are asserting their legal rights, whether in the area of land registration or in divorce negotiations.13 This suggests that increasing women's representation in village committees may have spillover effects that lead to improvements in gender equality. Moreover, since females tend to be on the LAC in poorer villages, the significance of this variable indicates that female LAC members may have a pro-poor effect on the land registration process.

6. SUMMARY AND POLICY IMPLICATIONS

Drawing on North's theory of institutional change, we use data from rural Ethiopia to show how the changes in the

Family Code implemented in 2000 and the community-based land registration, undertaken since 2003 may have created conditions for self-reinforcing reforms that favor gender equity.

Ethiopia's land registration process increased tenure security among women, and, if properly implemented, has the potential for far-reaching impacts. Similar to previous studies, however, our analysis finds gender gaps in awareness and information about the process. In particular, male-headed households are, on average, more likely to have heard about the land registration process, to have attended meetings (and a greater number of meetings), and to have received some written material with information about the process. A noteworthy finding is that the presence of female members in the LAC encourages participation by female-headed households, who are more likely to be excluded by the process, but does not discourage participation by male-headed households.

Although the reform of Family Code occurred a few years before the beginning of the land reform effort, our analysis finds additional impacts of the land registration effort on the evolution of perceptions of the distribution of assets upon divorce. We find that awareness about the land registration process is positively correlated with the shift in perceptions toward equal division of land and livestock upon divorce, particularly for wives in male-headed households. The presence of

Table 10. Regression results for changes in perceptions regarding allocation of land and livestock upon divorce with village fixed effects

Variables Moved to split land half-half Moved to split livestock half-half

All Male-headed Female-headed All Male-headed Female-headed

household household household household

(1) (2) (3) (4) (5) (6)

Age of household head -0.000 0.001 -0.002 -0.000 0.000 0.003

(0.001) (0.002) (0.003) (0.001) (0.002) (0.002)

Sex of household head (1 = male, 0 = female) 0.048 (0.047) 0.016 (0.043)

Number of years of schooling of the head 0.006 0.005 0.000 0.002 -0.003 0.024

(0.007) (0.007) (0.023) (0.006) (0.007) (0.020)

Highest grade obtained in household -0.003 -0.006 0.007 -0.003 -0.007 0.008

(0.006) (0.007) (0.010) (0.005) (0.006) (0.010)

Total plot area in hectares, 2004 survey -0.046*** -0.036*** -0.092*** -0.029*** -0.028*** -0.031

(0.011) (0.012) (0.033) (0.009) (0.010) (0.035)

Fraction of total land that is good or medium quality -0.001 0.021 -0.029 -0.020 0.022 -0.039

(0.053) (0.070) (0.090) (0.055) (0.071) (0.083)

Dummy for land quartile 1, 2004 survey -0.034 0.002 -0.171** -0.009 0.017 -0.025

(0.042) (0.052) (0.086) (0.041) (0.051) (0.084)

Dummy for land quartile 2, 2004 survey -0.033 -0.030 -0.071 -0.003 -0.021 0.063

(0.043) (0.051) (0.092) (0.040) (0.045) (0.090)

Dummy for land quartile 3, 2004 survey -0.015 -0.023 0.007 0.018 0.021 0.006

(0.040) (0.045) (0.094) (0.037) (0.043) (0.085)

Total livestock holdings, 2004 survey (tropical livestock units) -0.003 0.004 -0.016 0.001 0.007 -0.016*

(0.006) (0.007) (0.010) (0.005) (0.006) (0.008)

Presence of female members in the LAC 0.406*** 0.541*** 0.516** 0.203* 0.234* 0.375*

(0.127) (0.135) (0.241) (0.104) (0.135) (0.194)

Member of an iddir, 2004 survey -0.126* -0.129 -0.150 -0.173*** -0.261*** -0.111

(0.069) (0.092) (0.115) (0.064) (0.085) (0.108)

Network size, 2004 survey -0.002 -0.002 -0.000 -0.001 -0.001 0.005

(0.001) (0.001) (0.006) (0.001) (0.001) (0.005)

Aware of the land registration process 0.066 0.113 0.000 0.058 0.114 -0.005

(0.069) (0.089) (0.110) (0.058) (0.078) (0.086)

Observations 970 657 313 964 658 306

R-squared 0.243 0.287 0.233 0.282 0.317 0.338

Source: ERHS and authors' computation.

Notes: ***, **, and * represent statistical significance at 1%, 5%, and 10%, respectively. The number of observations in columns (1) and (4) represents the sample for which we have all set of covariates and outcome variables. The number of observations in columns (2) (3) and (5) (6) represents the male-(female-) headed households in the sample.

female members in the LAC also has a positive effect on the shift in perceptions toward a more equal distribution of assets upon divorce. This "female in LAC" effect is robust to inclusion of village fixed effects, which implies that even after controlling for local norms regarding the distribution of assets upon divorce, and the possibility that female members are more likely to be appointed in poorer villages with higher rates of female-headship, the presence of females in an important village-level committee may provide support to women and also may be a source of information regarding the new Family Code.

These findings are particularly important when viewed in the light of recent recommendations in a report on land tenure and property rights with special emphasis on vulnerable groups (Katz, 2010). This report, which identifies "women" as one of many vulnerable groups, argues that policy makers have to pay extra attention to these vulnerable groups when designing land reforms so that they will not be adversely affected by such reforms. Among the recommendations from the report's analysis of gender and land policy at a global level is the importance of a gender progressive legal framework, especially in relation to marital property rights and inheritance rights. This is relevant to the Ethiopian context, where changes to the Family Law preceded the land registration. There is also evidence that the two reforms, taken indi-

vidually, had positive gendered impacts. Studies on the land registration process indicate that it was largely beneficial to women and that increased tenure security enabled them to rent out their land (Holden et al., 2007). The research on family law shows that it led to greater participation by women in the labor force and that adverse perceptions of divorce outcomes have long-term consequences on children's schooling outcomes (Hallward-Driemeier & Gajigo, 2011; Kumar & Quisumbing (2012)). Our analysis suggests that, taken together, the reform of the Family Code and the community-based land registration process may have mutually reinforcing effects on women's rights and welfare. Despite the long history of gender discrimination in property rights in Ethiopia, these reforms, and recent increased attention to women in agricultural development programs, illustrate that perhaps, little by little, progress is being made—or, to quote an Ethiopian saying, the egg is beginning to walk. While this example is obviously rooted in the Ethiopian context, it raises the possibility that similar reform efforts may be complementary in other countries as well. Given the potential gains derived from eliminating the gender gap in access to assets and resources (FAO, 2011; World Bank, 2011), exploiting complementarities in the reform process may be an untapped opportunity to accelerate progress in closing the gender gap worldwide.

1. The review of gender and property rights institutions draws on previous work by one of the authors, in Fafchamps and Quisumbing (2002).

2. The 1997 round of the ERHS included detailed marital histories and questions about assets brought to marriage by each spouse. The 1997 round also included questions on what respondents perceived to be the disposition of assets upon marital dissolution (whether death or divorce). These are discussed in Fafchamps and Quisumbing (2002). The same questions were asked in the 2009 round, after the passage of the Family Law and the land registration.

3. This historical review draws heavily from Deininger, Ali, Holden, and Zevenbergen (2008).

4. The number of observations reported in the regression tables is about 1,000 because we have the complete set of covariates for that many households.

5. The ethnic and religious mix of the sample, for instance, does not match what we know of rural Ethiopia: Oromos are underrepresented; Protestants are overrepresented. The small number of Oromo sites is in part due to civil unrest at the time when the initial sample was drawn. Several villages from the Oromo region have been added to the 2000 survey round.

6. In comparison to the DHS, households in our sample are older and larger, reflecting the fact that the first 8 villages in the panel were surveyed in 1989, and additional 7 villages added in 2004. Although the female heads in DHS sample have more education than the female heads in our data, household education levels are higher in our sample.

7. We only collected perceptions data from female spouses or female heads of households because the modules on family history and disposition of assets upon marital dissolution upon divorce were only collected from women in the 1997 round. This was because the rest of the questionnaire, which focused on consumption, agricultural production, and other

information related to rural livelihoods, was administered primarily to men and we did not want to create excessive burden on respondents. While it would have been ideal to have this information for men and women alike, this is one of the constraints imposed by the dataset.

8. We realize that including village fixed effects only allows us to capture time-invariant unobservables, and not, for example, changes in gender norms over time. Moreover, because we are using changes in perceptions as a function of previous period characteristics, we are, strictly speaking, not estimating a panel data model.

9. Some of this may be driven by respondent bias, since the agriculture module of our survey was administered to the household head.

10. While these low numbers may raise doubts about survey implementation, these results are not surprising to those familiar with the land registration process. The land registration process in Tigray, which started much earlier, was very rushed and took place without photos, public awareness campaigns, or area measurement. This implies that land records were often out of date and that most of the farmers viewed the land certificate issued by this process as one of many certificates that they might have received in the past. That is, they may not have associated the land certificate they had in their possession with the community land registration effort (Klaus Deininger, personal communication, February 13, 2010).

11. This index is created by aggregating responses to questions relating to knowledge and participation in the land registration process (these questions are shown in Rows 1-5 in Table 5).

12. Even though the land registration process occurred after the enactment of new Family Law we examine the impact of measures related to the land registration process as of 2009 on changes in perceptions regarding the Family Law during 1997-2009.

13. We note, however, that including these fixed effects does not correct for time-varying village-specific unobservables.

REFERENCES

Alkire, S., Meinzen-Dick, R., Peterman, A., Quisumbing, A., Seymour, G., & Vas, A. (2013). The women's empowerment in agriculture index. World Development, 52, 71-91.

Allendorf, K. (2007). Do women's land rights promote empowerment and child health in Nepal? World Development, 35(11), 1975-1988.

Arthur, W. B. (1989). Competing technologies, increasing returns, and lock-in by historical events. Economic Journal, 99, 116-131.

Bevan, P., & Pankhurst, A. (1996). Report on the sociological dimension of the Ethiopian rural economies project. Centre for the Study of African Economies, Oxford University and Department of Sociology, Addis Ababa University, Oxford, mimeograph.

Deininger, K., Ali, D. A., Holden, S., & Zevenbergen, J. (2008). Rural land certification in Ethiopia: Process, initial impact, and implications for other African countries. World Development, 36(10), 1786-1812.

Dercon, S., Hoddinott, J., Krishnan, P., & Woldehanna, T. (2012). Burial societies in rural Ethiopia. In E. Mwangi, H. Markelova, & R. Meinzen-Dick (Eds.), Collective action and property rights for poverty reduction. Philadelphia: University of Pennsylvania Press for the International Food Policy Research Institute.

Dercon, S., & Krishnan, P. (2000). In sickness and in health: Risk sharing within households in rural Ethiopia. Journal of Political Economy, 108(4), 688-727.

Fafchamps, M., Kebede, B., & Quisumbing, A. (2009). Intrahousehold welfare in rural Ethiopia. Oxford Bulletin of Economics and Statistics, 71(4), 567-599.

Fafchamps, M., & Quisumbing, A. R. (2002). Control and ownership of assets within rural Ethiopian households. Journal of Development Studies, 38(6), 47-82.

Fafchamps, M., & Quisumbing, A. (2005). Assets at marriage in rural Ethiopia. Journal of Development Economics, 77(1), 1-25.

FAO (Food and Agriculture Organization of the United Nations). (2011). State of food and agriculture 2010-2011: Women in agriculture-closing the gender gap for development. Rome: FAO.

Federal Negarit Gazetta. (2000). No. 1/2000. Extra ordinary issue. The revised Family Code Proclamation No. 213/2000.

Frank, E. (1999). Gender, agricultural development and food security in Amhara, Ethiopia: The contested identity of women farmers in Ethiopia. Addis Ababa: United States Agency for International Development.

Gavian, S., & Ehui, S. (1999). Measuring the production efficiency of alternative land tenure contracts in a mixed crop-livestock system in Ethiopia. Agricultural Economics, 20, 37-49.

Gopal, G. (2001). Law and legal reform. In A. R. Quisumbing & R. S. Meinzen-Dick (Eds.), Empowering women to achieve food security. 2020 FOCUS. Washington, DC: International Food Policy Research Institute.

Gopal, G., & Salim, M. (1999) 'Gender and law: Eastern Africa speaks', conference organized by the World Bank and the Economic Commission for Africa. Washington, DC: World Bank.

Hallward-Driemeier, M., & Gajigo, O. (2011). Strengthening economic rights and women's occupational choice: The impact of reforming Ethiopia's family law. Paper presented at the Centre for Study of African Economies 25th Anniversary Conference 2011: Economic Development in Africa, Oxford, UK.

Hallward-Driemeier, M., Hasan, T., & Rusu, A. B. (2013a). Women's legal rights over 50 years: Progress, stagnation or regression? Policy Research working paper no. WPS 6616. Washington, DC: World Bank Group.

Hallward-Driemeier, M., Hasan, T., & Rusu, A. B. (2013b). Women's legal rights over 50 years: what is the impact of reform? Policy Research working paper no. WPS 6617. Washington, DC: World Bank.

Holden, S., Deininger, K., &Ghebru, H. (2007). Impact of land certification on land rental market participation in Tigray Region, Northern Ethiopia. Paper presented at the Nordic Development Economics Conference, Copenhagen, Denmark, June 18-19. Also Munich Personal RePEc Archive Paper 5211, Munich University Library.

Ibrahim, S., & Alkire, S. (2007). Agency and empowerment: A proposal for internationally comparable indicators. Oxford Development Studies, 35(4), 379-403.

Kabeer, N. (2001). Reflections on the measurement of women's empowerment: Theory and practice. In A. Sisask (Ed.), Discussing women's empowerment: Theory and practice (pp. 17-54). Stockholm, Norway: Novum Grafiska AB.

Katz, E. (2010). Land tenure, property rights, and natural resource management: Land tenure and property rights reform in the developing world: who is vulnerable? Prepared for the United States Agency for International Development, USAID Contract Number EPP-00-06-00008-00, Property Rights and Resource Governance (PRRGP) Task Order under PLACE Indefinite Quantity Contract.

Kumar, N., & Quisumbing, A. R. (2012). Beyond "death do us part": The long-term implications of divorce perceptions on women's well-being and child schooling in rural Ethiopia. World Development, 40(12), 2478-2489.

Kumar, N., & Quisumbing, A. R. (2013). Gendered impacts of the 20072008 food price crisis: Evidence using panel data from rural Ethiopia. Food Policy, 38, 11-22.

Lundberg, S., & Pollack, R. A. (1994). Noncooperative bargaining models of marriage. American Economic Review, 84(2), 132-137.

Manser, M., & Brown, M. (1980). Marriage and household decision making: A bargaining analysis. International Economic Review, 21(1), 31-44.

McElroy, M. B., & Horney, M. J. (1981). Nash-bargained household decisions: Toward a generalization of the theory of demand. International Economic Review, 22(2), 333-349.

Menon, N., van der Meulen Rodgers, Y., & Nguyen, H. (2014). Women's land rights and children's human capital in Vietnam. World Development, 54, 18-31.

North, D. (1990). Institutions, institutional change, and economic performance. Cambridge: Cambridge University Press.

Panda, P., & Agarwal, B. (2005). Marital violence, human development and women's property status in India. World Development, 33(5), 823-850.

Pankhurst, H. (1992). Gender, development, and identity: An Ethiopian study. London: Zed Books.

Quisumbing, A., Meinzen-Dick, R., Raney, T., Croppenstedt, A., Behrman, J. A., & Peterman, A. (Eds.) (2014). Gender in agriculture and food security: Closing the knowledge gap. Dordrecht and Rome: Springer and Food and Agriculture Organization of the United Nations.

Smith, L. C., Ramakrishnan, U., Ndiaye, A., Haddad, L., & Martorell, R. (2003). The importance of women's status for child nutrition in developing countries. Research report 131. Washington, DC: International Food Policy Research Institute.

Sraboni, E., Malapit, H. J., Quisumbing, A., & Ahmed, A. (2014). Women's empowerment in agriculture: What role for food security in Bangladesh? World Development, 61, 11-52.

Webb, P., von Braun, J., & Yohannes, Y. (1992). Famine in Ethiopia: Policy implications of coping failure at national and household levels. Research report 92. Washington, DC: International Food Policy Research Institute.

Wiig, H. (2013). Joint titling in rural Peru: Impact on women's participation in household decision-making. World Development, 52, 104-119.

World Bank. (2011). World development report 2012: Gender equality and development. Washington, DC.

Available online at www.sciencedirect.com

ScienceDirect