Women’s Asset Ownership: Evidence from Georgia; Mongolia; and Cavite, Philippines

Since the implementation of the Millennium Development Goals, significant global progress has been achieved in promoting gender equality in education and health. However, progress has not been as remarkable in advancing women’s rights on asset ownership and control, which is critical for securing gender equity in economic participation and opportunity, and delivering on the Sustainable Development Goals. Part of the problem comes from the lack of standards on collecting sex-disaggregated data on the topic. The Evidence and Data for Gender Equality (EDGE) initiative aims to develop standardized methods and guidelines for collecting sex-disaggregated data on asset ownership. This paper provides rich inputs to the methodological guidelines being developed by the United Nations and development partners by drawing on the key findings from the pilot surveys conducted in Georgia; Mongolia; and Cavite, Philippines. Furthermore, survey results suggest substantive gender gaps in ownership across different types of assets and countries. These variations confirm the importance of understanding the social norms governing gender roles in society and legislation that can facilitate or impede women’s asset ownership.


INTRODUCTION
Globally, countries recognize the importance of closing the gaps between men and women. Goal 5 of the Sustainable Development Goals sets out several targets that aim to empower girls and women in terms of economic and political participation, educational attainment, and health outcomes. While substantial headway has been made in some of these areas, significant gender disparities still exist with respect to access to productive resources.
Hence, in any poverty reduction program, promoting greater gender equality in rights to economic resources and ownership of productive assets should be an important ingredient. In fact, numerous studies point to a strong positive correlation between socioeconomic well-being and asset ownership. Existing literature identifies ownership of specific types of assets (e.g., house or land) that determines whether a family is chronically poor or not. Other studies also show that ownership of durable assets can minimize the risk of falling into transient poverty as such assets can facilitate income diversification or be used as collateral to be able to access credit.
There is also a wealth of evidence suggesting that women who have access and command over productive assets also have greater participation in household decision-making (Garikipati 2009;Twyman, Useche, and Deere 2015;and Swaminathan, Suchitra, and Lahoti 2011). This also translates to households faring well in terms of family members' health and nutrition, and children's education outcomes (Quisumbing and Maluccio 2000;and Doss 2006). Furthermore, studies show that women who own productive assets have reduced risk of experiencing domestic violence (Panda and Agarwal 2007;Oduro and Seshie-Nasser 2016;and Bonilla et al. 2017). In other areas, research also demonstrates that more equitable distribution of land rights and security of tenure for women are associated with increased agricultural productivity and better soil conservation practices (Goldstein and Udry 2008;Ali, Deininger, and Goldstein 2014).
While statistics on asset ownership are essential for building evidence to better understand the topic and provide inputs for programs that aim to provide equal opportunities for men and women, sex-disaggregated data needed for monitoring progress on this front is scarce (ADB 2017). Conventional data collection vehicles, including surveys which are conducted by national statistical agencies collect information on assets that are owned by households. By nature, this data collection approach does not shed light on which specific household members actually own the assets. Researchers who are interested in examining gender disparities can only do so by comparing households headed by men with those headed by women. Headship analysis is often a poor proxy for gender analysis as households headed by men and those headed by women are structurally different. Some studies show that for certain categories of assets, gender inequality is overestimated by headship analysis as it ignores women in households headed by men (Deere, Alvarado, and Twyman 2012) while others suggest that gaps in agricultural productivity between plots owned by men and women could be underestimated if analysis is categorized based on the sex of the household head (Peterman et al. 2011).
To address this issue, several initiatives to collect individual-level asset data have been made. However, as most of these attempts were made independently, a common framework to compare the results emerging from these studies is missing due to differences in survey instruments, sampling design, and data collection protocols.
The global initiative Evidence and Data for Gender Equality or EDGE led by the United Nations Statistics Division (UNSD) aims to develop methodological guidelines for collecting sex-disaggregated data on asset ownership and entrepreneurship that can be used for systematic collection of data on this topic. This initiative is in collaboration with the United Nations Entity for Gender Equality and the Empowerment of Women (UN Women), Food and Agriculture Organization of the United Nations (FAO), and other development partners such as the World Bank and the Asian Development Bank (ADB). Pilot surveys employing different methodologies were fielded in seven countries namely, Georgia, Maldives, Mexico, Mongolia, the Philippines, South Africa, and Uganda. Having the capacity to contribute to the development of standardized methodology, availability of plans to either conduct a relevant survey that could accommodate a module or a full-blown survey on asset ownership, and interest in producing better gender statistics, particularly on asset ownership were among the key criteria used in choosing pilot countries. Furthermore, the chosen pilot countries also offer diverse representation in terms of economic landscape, gender norms, and legal property rights. The World Bank provided support to the data collection in Uganda while the National Institute of Statistics and Geography (INEGI) supported the conduct of the survey in Mexico. UNSD supported the conduct of surveys in Maldives, Uganda, and South Africa. ADB contributed to this global effort especially in generating internationally comparable indicators on entrepreneurship and asset ownership by providing administrative, technical, and financial assistance to the national statistics offices of Georgia, Mongolia, and the Philippines in conducting pilot surveys.
This paper presents key results on gendered patterns of asset ownership from the three Asian countries supported by ADB. The findings from the three surveys conducted in Georgia, Mongolia, and the province of Cavite in the Philippines have provided rich inputs to the methodological guidelines on the collection of sex-disaggregated data on asset ownership (ADB 2018). One of the methodological issues that was taken into consideration during the design of the surveys implemented in the pilot countries is whether the information on asset ownership can be collected by proxy or should be selfreported. Findings from the pilot surveys suggest that there are more merits in using self-reported information. Furthermore, data collected from ADB-supported pilot surveys indicate that there are substantive gender gaps in ownership, although these vary by asset types and between countries. These variations point to the importance of understanding the social norms governing men and women's roles and responsibilities in society and the legislative framework that can facilitate or impede women's asset ownership. 1 The remainder of the paper is organized as follows. Section II briefly describes the conceptual framework followed by the EDGE project while section III discusses the main features of the surveys conducted in Georgia; Mongolia; and Cavite, Philippines. In section IV, we highlight the methodological and substantive findings from the three pilot surveys while section V identifies areas for future research. In the final section, we draw broad policy implications of the results.

1
The findings from Georgia, Mongolia, and the Philippines, along with EDGE surveys conducted in other countries, are used as inputs to the development of UNSD's methodological guidelines on the collection of sex-disaggregated data on asset ownership.

II. ASSET OWNERSHIP AND CONTROL: A GENDER PERSPECTIVE
The EDGE conceptual framework (see figure below) provides a comprehensive overview of the diverse factors underpinning asset ownership and why it is important from a policy perspective. Laying it out explicitly makes it clear that the entire spectrum ranging from asset acquisition, ownership, and use, is gendered. Thus, any attempt to reduce the gender gap has to grapple with how gender interacts along this spectrum. Most surveys, whether collecting household or individual asset data, simply request respondents to identify if the asset is held by the household and if so, to identify individual owners. This is labeled as reported ownership and can be differentiated from documented ownership, where ownership claim is based on having one's name on some type of legal document (formal title deed, purchase agreement, and property tax records, and so on). This type of ownership may provide greater security of tenure than reported ownership. For example, in the Philippines, the system does not provide security of tenure to informal settlers who do not possess any ownership documents and who illegally occupy public and private lands (Porio and Crisol 2004). On the other hand, indigenous people (IP) in other Asian countries hold land through "customary or community-based tenure systems" but their legal rights are only limited to a small portion of the land, unless they are able to own the land through other means (e.g., in Cambodia and the Philippines, land can be officially allocated for IPs and local groups [Rights and Resources Initiative 2015]) and have their names written as owners in legal documents. In countries where formal property rights are recognized to some degree, documented ownership are also necessary for alienation rights, right to sell, and right to bequeath.
In many developing countries, particularly in Asia and Africa, the full bundle of rights may not be vested in one individual (ADB 2017). Someone could be a reported owner, but not a documented owner. Women may be legal owners of assets, but due to gendered norms, men may actually control and manage the assets and often, never report women as owners. This is likely to arise, especially with immovable property such as land and house. In tenure systems where land is leased over long periods, alienation rights maybe restricted allowing intergenerational transfers but not sale. Such is the land tenure system in Maldives prior to the enactment of the Land Act of 2002 when the land is solely owned by the State, which provides permission to individuals to occupy the land for residential, commercial or other use. The law has, since then, granted private citizens to acquire the land by purchasing it from the State, provided that owners' and heirs' rights are considered in the acquisition (IMF 2008). Similarly, in Nigeria, the state owns the land, and the head of the state provides legal rights of tenancy. The tenant, however, is restricted from selling, leasing, or transferring the land without permission from the head of state (International Center for Nigerian Law 1990).
Another important factor affecting control over assets, including alienation rights, is whether ownership is joint or individual. This form of ownership is heavily influenced by the marital regime of the country in interaction with social norms. A marital regime regulates ownership of marital assets or assets within a consensual union. There are three common types of marital regimes: separation of property, partial community property, and full community property. Under a separation of property regime, properties brought into marriage are separately owned and any property acquired and inherited during marriage is maintained. Under a partial community property regime, properties acquired during marriage by either spouse is treated as jointly owned by both spouses. On the other hand, under a full community property regime, all individual properties brought into, acquired, and inherited during marriage is treated as the joint property of both spouses. Thus, inheritance is treated on par with marital assets in full community property, while in partial community property, inheritance is kept separate from marital assets. The influence of social norms cannot be underrated. Often, laws are egalitarian, but are rendered ineffective in the face of regressive attitudes. This is typically seen in the context of natal inheritance for women. In some Latin American countries, for instance, there is a disconnect between equality in women's and men's property rights as stipulated in the law and men's privileges in inheritance of properties (Hamilton 2004). Again, while laws may specify no discrimination between sons and daughters, patriarchal attitudes often force women to sign away their rights to their brothers.
An asset, as conceptualized here, is based on the 2008 System of National Accounts (SNA) set of definitions and is understood as "a store of value representing a benefit or a series of benefits accruing to the economic owner by holding or using the entity over a period of time." Thus, all assets, physical and financial, fall within the purview of data collection. The framework further highlights the complementarity of wealth data to asset ownership data in providing rich insights to socioeconomic planners when designing policies that could potentially have an impact on the well-being of individuals.

III. METHODOLOGY OF ASSET SURVEYS
The EDGE data collection covered the entire range of physical and financial assets. The physical assets included immovable property (dwellings, agricultural land, and other real estate); livestock; agricultural equipment; nonagricultural enterprise and enterprise assets; consumer durables and valuables (jewelry, precious stones, and so on). The financial assets module elicited information on savings in formal and informal sectors, equity, pension funds, insurance instruments, as well as money lent by the respondent. For a range of assets, the surveys also collected valuation data and how the asset was acquired. The value of an asset summarizes the quality of the asset and its attributes (size of the dwelling or plot, desirability of the location, etc.). Further, differences in the quantity of an asset owned across respondents are captured by value data, unlike an ownership count. The respondents reported the value of their assets at current market price. Table 1 provides details on the ownership, valuation, and mode of acquisition questions in the surveys. with respect to household assets; the second respondent was the spouse or partner, if any, of the primary respondent. These individuals are the principal couple. The third respondent was chosen randomly from the remaining adults. The total respondents interviewed were 5,937 in Georgia; 5,592 in Mongolia; and 3,456 in Cavite, Philippines.
A. Who Is an Asset Owner?
There is increasing recognition that one should pay attention to who is being interviewed in surveys. Household surveys conventionally collect most of their information from one member in the household, typically the head or someone identified by the household as being most knowledgeable of the relevant survey topic. Surveys like the Demographic and Health Surveys (DHS) that focus on reproductive health and sensitive issues such as prevalence of domestic violence collect data from several individuals within the household. Even so, monetary information relating to income, consumption expenditure, or asset inventory to construct wealth quintiles are obtained from one person.
Recent evidence suggests that proxy data provided by one respondent differs from data that is collected from individual household members or self-reported data. In Malawi, the wife's income tends to be underestimated by husbands (Fisher, Reimer and Carr 2010); in Tanzania, with respect to labor force participation rates, Bardasi et al. (2011) find that male employment rates are negatively biased due to proxy reporting, but with no impact on female employment rates. There may be several reasons for such divergence in estimates; there could be an information asymmetry within households either deliberately or due to gendered norms that may prevent such communication, particularly between men and women.
With regard to economic resources such as assets, gendered perceptions of ownership may also bias ownership estimates. In Uganda, Kilic, and Moylan (2016) report substantive differences in both men's and women's asset ownership estimates on comparing proxy versus self-reported data. The authors consider several assets and conclude that proxy reporting underestimates the individual asset ownership rate irrespective of the sex of the owner.
The EDGE surveys discussed in this paper collected individual-level asset information through self-reporting as well as proxy reporting which enables a comparison of these two approaches. A maximum of three respondents were interviewed in each household, wherein each respondent provided information on their own assets as well as ownership information on assets held by other household members. Thus, they self-report and also act as proxies for other household members. The survey calculates both self-assigned ownership (SAO) using self-reported data and ownership assigned by any respondent (OAAR) using proxy information from all household respondents.
The SAO approach ignores the information provided (proxy reporting) by the other respondents about the ownership of assets and is based on the premise that the respondent is in the best position to provide accurate information about the assets in which they have an ownership stake. On the other hand, the OAAR approach consolidates the information provided by all respondents to form a single set of information for the household. Statistically, it is the union of two or three sets of information. The OAAR approach is the "most inclusive" approach in that it considers all information provided by all eligible respondents in the household regarding assets ownership. "Inclusive" here refers to the broadest definition of ownership, i.e., as long as a person is identified as an owner by one eligible respondent in the household, that person is considered an owner, irrespective of what the other household members report.
To illustrate the differences between the two approaches, consider the tabulation of hypothetical responses for reported ownership of an asset in Table 2. The household has five adult household members (columns numbered 1 through 5, with their sex in parentheses), three of whom were selected as respondents. Under the OAAR approach, a household member is considered an owner as long as he/she has been identified as an owner by at least one respondent. Hence, the fourth and fifth members are considered as owners, even though they were not interviewed, since the second respondent identified them as such. Meanwhile, under the SAO approach, only the first two members are classified as owners since it is based on self-identification. However, the information provided by selfreported owners are attached with inflation factors so that they still sum up to the total target population.

B. Indicators of Asset Ownership
Drawing from the conceptual framework presented earlier, three sets of indicators-incidence of asset ownership, distribution of form of ownership, and the gender wealth gap-are presented by sex to explore inequalities in asset ownership from a gender perspective (see Box). Rather than rely on a single indicator, a portfolio of indicators is relevant to fully appreciate why gender disparities exist and to suggest possible policy levers to help remedy the problem.
Incidence of asset ownership is the proportion of individuals in the total population owning a particular asset and provides an easy-to-understand snapshot of the prevalence of asset ownership in a country. The data requirements are relatively light and the indicator itself is simple to calculate. However, there are also some downsides to this indicator; we do not know if a person owns an asset exclusively or jointly with other persons, or owns several pieces of the same type of asset (e.g., owns two parcels of land). Further, there is no indication of the quality of the asset owned.
To address these gaps, two additional measures are presented: the distribution of forms of asset ownership and gender wealth gap indicator. The first measure reveals how the asset is owned; whether by a single individual or jointly by several individuals. The joint ownership could be between the principal couple, with household members or even nonhousehold members. Often, whether an asset is owned individually or jointly is influenced by inheritance and marital regimes, which in turn could impact rights over the assets. The distribution measure can also look at distribution of alienation rights and mode of asset acquisition. The EDGE surveys collected data on asset value, based on the current sales price of the asset, to calculate a gender wealth gap that would account for the quality and quantity of the assets as well as individual or joint ownership. Valuation data though, is harder to obtain as there may be missing asset markets or lack of awareness of market prices by the respondent or a reluctance to share such information.

Measuring Gender Differences in Asset Ownership
To examine patterns in ownership and detect any gender disparity, three measures were used in the Evidence and Data for Gender Equality (EDGE) pilot surveys. These were incidence of asset ownership, distribution of asset ownership, and the gender wealth gap.

Incidence of Asset Ownership by Sex
Incidence of asset ownership measures what percentage of adult female are owners as well as what percentage of adult male are owners.
The incidence of ownership was computed by type of ownership (reported or documented) for all assets covered in the survey, and using the type of ownership approach (self-assigned ownership and ownership assigned by any respondent) for select assets.

Distribution of Asset Ownership
This measure looks at the distribution of asset ownership by forms of ownership (i.e., exclusively owned by men, exclusively owned by women, owned by principal couple, jointly owned by all household members, and other forms of joint ownership.), enabling us to compare the proportion of owners who have exclusive or joint rights over an asset. A sample formula for the percentage of exclusive male (or female) owners of a specified category of assets out of all the owners of that asset type is as follows.
The distribution of ownership may also be examined by ownership rights (right to sell or bequeath an asset); form of ownership and right (exclusive or joint right to either sell or bequeath an asset); and mode of asset acquisition (founded or purchased, inherited, etc.), to name a few.

Gender Wealth Gap
The value of an asset is derived from its current market price. In the succeeding analysis of wealth gap, we will focus on the value of dwellings only because of high nonresponse rates on values of other types of assets. The share of the value of dwellings owned by men and by women is then computed using the formula below. All indicators are calculated for individuals 18 years and older. The incidence indicator is estimated for all asset categories, while other indicators are presented for select assets. The results presented in this paper are mainly based on the SAO approach. Nevertheless, we also present a comparison of the incidence estimates derived from the SAO and OAAR approaches.

IV. COUNTRY CONTEXT AND KEY RESULTS
The three countries that conducted the EDGE pilot survey vary in terms of geography, demographic features, administrative structure, and economy. The pilot countries also have diverse social and legal factors that govern asset ownership and corresponding actions such as acquisition, bequeathing, and sale. These consist of constitutional provisions and existing laws on inheritance and marital regimes, as well as social customs and practices.
Georgia. Georgia is situated at the crossroads of Europe and Asia in the Caucasus and its total population is estimated at 3.72 million in 2016. Its gross domestic product (GDP) per capita is estimated at $3,852.5 in nominal terms and $9,267.3 in purchasing power parity (PPP) terms in 2016. The male-female population ratio is 1.11:1 in 2015 (National Statistics Office of Georgia 2018). The country ranks 70th out of 180 countries in the Human Development Index (HDI) and is categorized in countries as having a high level of human development. The HDI for women is 0.754 and the HDI for men is 0.77, indicating a small gap of human development between the two sexes (UNDP 2016).
Gender equality is an important area of discourse in Georgia. The Georgian government has ratified the United Nations (UN) Convention on the Elimination of All Forms of Discrimination against Women (CEDAW) in 1994, and enacted the Constitution of Georgia in 1995 and the Law on Gender Equality in 2010, which all includes gender equality provisions. Article 38 of The Constitution of Georgia states that "Citizens of Georgia shall be equal in social, economic, cultural and political life irrespective of their national, ethnic, religious or linguistic belonging." Article 10 of the Law on Gender Equality, on the other hand, stipulates gender equality in family relations and states that "spouses shall have equal rights to own, acquire, manage, enjoy, and administer property" (ADB 2018).
Mongolia. Mongolia is located between the Russian Federation in the north and the People's Republic of China in the south. Mongolia's total population reached 3.06 million in 2015, with women comprising more than half (51.6%). The country has a high level of human development based on the HDI and ranks 92nd out of 180 countries. The HDI for women is 0.744 and the HDI for men is 0.725, indicating a small gap in disparities of development that favors women (UNDP 2016). Cavite, Philippines. Cavite is a province located in Region IV-A CALABARZON, which is adjacent to the National Capital Region. Its nearness to the nation's metropolis has been favorable to the economic development of the province. The province had a total population of 3.68 million in 2015, making it the fastest growing province in the region (Philippine Statistics Authority 2015). It is considered as a highly industrialized province, with around 1,250 industrial establishments registered in 2015, with businesses in export, logistic services, facilities, and information technology (Provincial Government of Cavite 2015).
The Philippines has enacted several legislations in its efforts to address gender equality and women empowerment in the country. It has ratified CEDAW, the Philippine Constitution, Magna Carta of Women (MCW), and the Family Code of the Philippines, which all provide protection and welfare for Filipino women in various concerns. The MCW stipulates equal rights of spouses in terms of "ownership, acquisition, management, administration, enjoyment and disposition of property." The Family Code of the Philippines, on the other hand, provides supplementary clause on property regimes that may be followed in marriage settlements among spouses, i.e., absolute community, conjugal partnership of gain, complete separation of property, or any other regime (Philippine Statistics Authority 2018).
The key characteristics of the male and female respondents for Georgia, Mongolia, and Cavite are presented in Table 3. The average age of surveyed respondents ranges from 39 to 50 years old. In terms of educational level, about four in 10 respondents have attained secondary education. In all these three areas, female respondents are more educated than male respondents in general. More than 60% of the respondents in the three pilot areas are married. Meanwhile, at least 65% of male respondents in the three countries are employed. In Georgia and Cavite, half of the female respondents are employed, while the remaining half are either unemployed or not in the labor force. A.

Incidence of Asset Ownership: Reported and Documented
The incidence measure gives the proportion of the total adult population who are asset owners. The incidence gap or the gender asset gap is the difference in ownership rates between men and women (Box). Table 4 presents the incidence for immovable assets (dwelling, agricultural land, and other real estate) using reported and documented ownership concepts and results of corresponding t-tests.
The levels of reported ownership of property varies across the countries reflecting differing institutional structures. Dwelling is widely owned, while agricultural land shows the least prevalence rate (except for Georgia). Dwelling is owned by 80% of men and 76% of women in Georgia, by 60% of men and 33% of women in Mongolia, and by 34% of men and equal proportion of women in Cavite. Given its relatively urban nature, Cavite province, at less than 5%, has the lowest landownership rates. Mongolia is only marginally higher, but in Georgia, nearly 50% of the men and more than one-third of the women report landownership. This is mainly due to the receipt of private land by rural households in Georgia after the collapse of the Soviet Union, while Mongolia has more communal land. Other real estate refers to residential and nonresidential buildings excluding dwelling and nonagricultural land. Real estate ownership is lowest in Cavite. In Mongolia, it is higher than ownership of agricultural land, but lower than ownership of dwelling.
With few exceptions, irrespective of the type of ownership or context, men are significantly more likely than women to own immovable property. The gap in men's and women's property ownership is smallest in Cavite (in fact, there is no statistically significant gender gap for reported dwelling and real estate ownership) and largest in Mongolia. Focusing on reported ownership, men in Mongolia are twice as likely as women to own their dwellings, four times as likely to own land, and a little more than one-and-a-half times as likely to own other real estate.
The trends in documented ownership mirror that of reported ownership, although the ownership rates are much lower for both men and women. This suggests that many reported owners do not have their names on documents, with sharper differences in Georgia (ranging from 4 to 43 percentage points) than in the other countries. Although the reasons for this discrepancy was not obtained during the survey, one could speculate that there are transaction costs to changing names on such ownership documents. During field interviews, respondents are asked about the existence of legal documents with the names of the owners but the existence of such documents are not further verified. Presumably, disparity between the incidence of reported and documented ownership could even be larger than estimated if the names of the respondents are checked in the ownership documents. Notes: Significant t-test result implies that the incidence of ownership among men is statistically higher than the incidence of ownership among women. *** = 1% significance, ** = 5% significance, * = 10% significance. Source: Asian Development Bank estimates using the Evidence and Data for Gender Equality (EDGE) pilot surveys. Table 5 presents the incidence of ownership of assets other than immovable properties and corresponding results of t-tests. The importance of livestock in rural areas is reflected in fairly high ownership rates in Mongolia and Georgia. Approximately 40% of both men and women adult population in Georgia own livestock; whereas at 33% for men and 18% for women, the gender gap in ownership is highest in Mongolia.
Electronic copy available at: https://ssrn.com/abstract=3343988 Ownership of large agricultural equipment is low in Mongolia and negligible in Cavite; but it is not uncommon to hold small agricultural equipment. This low ownership could be due to two reasons: low ownership of agricultural land, and presence of an active rental market for large equipment due to their high costs of acquisition. Small agricultural equipment tends to be owned by all household members in Georgia and hence, the ownership module was not canvassed. However, ownership pattern among men and women seems more distinct for large agricultural equipment, with only 1% of Georgian women owning such equipment compared to 6% of Georgian men.
As expected, the incidence of ownership of consumer durables is highest among all assets in all the three countries, with low or no gender gaps. In fact, there is no statistically significant gender gap on ownership of consumer durables in Georgia and Cavite. The estimates of incidence of ownership of financial assets are lower than expected, particularly for Georgia. This finding could be attributed to the limitations on how the financial assets were conceptualized in the survey instruments and during field operations. In particular, respondents mix pension funds with the existing state pension allowances and exclude bank accounts that are opened for the purpose of receiving state pension allowances in the list of financial assets. On the other hand, the results suggest that women are slightly more likely to own financial assets than men in Mongolia and Cavite.
The gender gaps for the incidence of nonagricultural enterprises are more evident in Georgia and Mongolia. Interestingly, in Cavite, the gender gap is reversed in favor of women. Notes: Significant t-test result implies that the incidence of ownership among men is statistically different from the incidence of ownership among women. *** = 1% significance, ** = 5% significance, * = 10% significance. Source: Asian Development Bank estimates using the Evidence and Data for Gender Equality (EDGE) pilot surveys. Electronic copy available at: https://ssrn.com/abstract=3343988

B. Forms of Ownership and Alienation Rights over Assets
In addition to examining the incidence of ownership of assets among men and women, it is also important to study whether an asset is owned individually or jointly as it can have implications for an individual's control and rights over the asset. However, there is no linear mapping between ownership and control over assets; there are many mediating factors that are often encapsulated as "context" and refers to macro (laws, norms, and so on) and micro (household structure, marital relations, and so on) processes.

Forms of Ownership
Joint ownership can be with spouses or partners, other household members, and often with nonhousehold members too. These ownership patterns are generally influenced by the legislative framework (inheritance and marital regimes) as well as current social norms and practices. Marital regime refers to the set of laws that govern property ownership within marriage or cohabitation. Georgia and Mongolia follow a partial community of property marital regime which means that all property acquired during the marriage is treated as joint property of the couple. But property that was acquired before the marriage and any inheritances are treated as individual property. The Philippines follows a full community property regime where all property, irrespective of when and how they were acquired is treated as jointly owned by the couple. Thus, one could expect ownership by the principal couple to dominate in the province of Cavite in the Philippines compared to the other two countries.
In all three countries, there is no legal discrimination in inheritance between sons and daughters, but prevailing norms reveal biases in favor of men. For instance, Georgian men are more than twice as likely to acquire dwelling units as an inheritance than Georgian women, and this reflects the traditional perceptions that men are more privileged in inheritance (National Statistics Office of Georgia 2018). Meanwhile, existing law in Mongolia provides that all adult household members should be registered as owners in land titles, but members have the prerogative to renounce this right. Women's insufficient knowledge of their legal rights on landownership leads to more men owning majority of land in Mongolia (National Statistics Office of Mongolia 2018). The Philippines' Magna Carta for Women upholds equality on ownership, acquisition, and disposition of property among men and women. However, a "closer analysis of certain provisions of the laws and its interaction with customs and tradition reveals constraints on women's asset ownership and accumulation" (Philippine Statistics Authority 2018).
Considering owners of dwelling (Table 6), we see a striking impact of the marital regime playing out with ownership by the principal couple as the most prevalent in Cavite, Philippines (63% and 33% for reported and documented, respectively). The corresponding figures for Georgia are 17% and 10%, respectively and for Mongolia are 22% and 15%, respectively. Individual ownership by men is dominant in Mongolia (43% and 44% for reported and documented, respectively); in Georgia, there is a switch from all household members as reported owners to exclusive male owners for documented ownership (31%), suggesting that the perception of ownership is more inclusive than the documented reality. Turning to agricultural land, one sees more variation across ownership types. Exclusive male ownership of at least 15% cuts across ownership types and countries. In Cavite and Mongolia, landownership rates are very low; thus, the forms of ownership figures should be interpreted with caution due to the small sample size. Reported and documented ownership with non-household members is common in Cavite. This could be due to the relatively urban nature of the province where urban households co-own agricultural land in rural areas with extended family members. The proportion of exclusive male owners is higher for documented than reported dwelling owners in Mongolia, and mainly comes at the expense of women as exclusive owners, and principal couple owners. On average, the gender gap in exclusive ownership is highest in Mongolia for reported and documented owners of immovable property and is also high for documented ownership for agricultural land in Georgia. Interestingly in Georgia, far fewer women are reported as exclusive owners of property than what the documents seem to indicate.

Alienation Rights
Certain patterns emerge with respect to alienation rights (right to sell and bequeath) over property (Tables 7 and 8). Two broad trends are clear regarding rights, assets, and locations. First, a higher proportion of women owners than men owners have no alienation rights. Second, women owners' exclusive right to sell or bequeath is lower than men owners; however, in most instances, women owners' consultative rights are on par if not higher than men owners' consultative rights.
Electronic copy available at: https://ssrn.com/abstract=3343988 A larger proportion of Mongolian men and women owners have exclusive alienation rights over sale and bequeathes compared to owners in the other countries. With respect to dwelling for example, nearly three-quarters of Mongolian men owners have exclusive rights to bequeath their dwelling compared to 25% and 38% in Georgia and Cavite, respectively. Similar trends are observed with women owners as well. About 52% have an exclusive right to bequeath their dwelling, compared to 19% and 34% in Georgia and Cavite, respectively (Table 8). This can be attributed to the relatively strong individual ownership patterns that one observes in Mongolia.
Georgian women owners are the most disadvantaged in terms of having no rights over their property. Approximately 20% of women owners do not have any right to sell their dwelling or land, while a quarter do not have any bequeathing rights over these assets. However, the results also suggest that consulting rights are more prevalent in Georgia for both men and women. This could presumably be related to the manner in which the property was acquired and is discussed in the following section.

Do Forms of Ownership and Rights Move Together?
The implications of individual or joint ownership is likely to vary across asset types and countries. While it is likely that joint ownership involving women is better than no ownership, it cannot be presumed all women owners will benefit to the same extent or enjoy equal rights over their assets as their co-owners. Exploring the relationship between forms and alienation rights for dwelling (Tables 9 and 10), the results suggest there exists a positive relationship between form of ownership and right to sell or bequeath. Focusing first on reported ownership, exclusive ownership confers the strongest rights irrespective of the sex of the owner. When there is ownership by the principal couple, there is greater consultation in decisions, but women do not seem to enjoy equal rights as their partners. Women are more likely to have no rights, while men are more likely to have exclusive rights even when the asset is jointly held by the couple. Interestingly, men most likely have no rights over the assets when it is held jointly with other household or nonhousehold members. A comparison of rights between reported and documented ownership is revealing; consistently, across the three countries, having one's names on documents strengthen rights for women who own assets jointly, either with their partners or others. They are likely to have greater consultative rights as suggested by the substantive reduction in the proportion of women owners with no rights. This effect also holds for men; in the case of other joint ownership, men with documents have greater consultative rights than men who are reported owners. In sum, the results suggest that documented ownership does indeed have a stronger relationship with right to sell and right to bequeath than reported ownership.  C.

Mode of Acquisition
Asset acquisition patterns shed some light on the complex interactions between legislation and broad gendered norms governing men's and women's social roles. Assets may be acquired in one of several ways: through market transaction, state transfer, inheritance or gift or within marriage (which could involve any of the other mechanisms).
A broad observation is that, with the exception of agricultural land in Mongolia, markets account for a substantive portion of property acquisition for men and women (Table 11). There are some exceptions to this pattern. Interestingly for other real estate, a greater proportion of women owners in Cavite have purchased their property than men owners (67% versus 55%, respectively).
Acquiring property due to marital law or custom is important for Georgian women across property types (39% for dwelling, 33% for land, and 21% for other real estate), whereas for men, gifts or allocations range from 38% to 46% as a means of property acquisition. This reflects the existing marital law in Georgia, where a woman enjoys equal rights to own her spouse's property after marriage (ADB 2018). 2 Inheritance, whether natal or marital, while not totally unimportant, is also not a typical means of acquiring property, except in acquiring agricultural land in Cavite (39% and 42% for men and women, respectively). The data suggest a gender bias in inheritance toward men in Georgia and Mongolia (excluding land); in Cavite on the other hand, a greater proportion of women owners report inheriting their dwelling and land.
Government programs as a means of property are particularly relevant in Mongolia. Presumably, due to the communal nature of land, purchase is least common with almost 60% of men and 45% of women reporting government allocation as their mode of acquisition. Such programs also account for roughly one-third of other real estate owned by men and women.
In general, with regard to dwellings, our results suggest that the modes of acquisition are not gender biased in Cavite. On the contrary, the proportion of female owners who acquired their dwellings by purchasing the asset is higher than that of men. Acquisition of dwelling through marital custom is also slightly biased toward women, while inheritance shows a slight male bias in Mongolia. In Georgia, one finds more significant gender biases. Women are less likely than men to inherit or receive a gift from household members, but more likely to acquire via marital law and custom.

2
Article 10 of the Law on Gender Equality stipulated gender equality in family relations and stated that "spouses shall have equal rights to own, acquire, manage, enjoy, and administer property" (The Law of Georgia on Gender Equality 2010. https://matsne.gov.ge/en/document/download/91624/3/en/pdf). Electronic copy available at: https://ssrn.com/abstract=3343988

D. Gender Wealth Gap
Valuation data, although challenging to collect, provide greater depth to our understanding of the gender gap in economic resources. Asset values in the EDGE survey were based on the market price of the asset on the day of the interview as reported by the respondents. We briefly discuss concerns around missing data as it could potentially impact reliability and usefulness of estimates derived from the data.
Missing values are not uniform across countries. For example, with dwellings, the data show variation in the proportion of missing values across countries as well as by sex. On average, women are less likely than men to provide a value for the dwelling. At 15% for men and 18% for women, the nonresponse for dwelling valuation is lowest in Mongolia, followed by Cavite (48% for men and 60% for women); then Georgia (65% for men and 72% for women). About 51% of the respondents in Georgia and 74% of the respondents in Cavite who did not provide sales value of dwelling declared that there are no market or sales transactions in the community. 3 While imputation is a possibility, it requires detailed information on the asset attributes that are likely to affect its value.
Keeping these data limitations in mind, it is nevertheless useful to examine the descriptive results based on reported market values. In this paper, we only consider the gender wealth gap calculated from the market value of the dwelling unit based on reported and documented ownership of assets (Table 12). 4 For dwellings that are jointly owned, the value of a dwelling is apportioned equally among all owners.
Considering reported ownership, at 49%, women's share of dwelling wealth is almost equal to that of men in Georgia, followed by Cavite at 45%, and Mongolia at 37%. Contrasting women's share of wealth to their representation among owners provides some insights. In Georgia and Cavite, women represent more than half of all reported dwelling owners (53% and 51%, respectively), but their share of dwelling wealth is lower than 50%, suggesting that the dwellings owned by women may be less valuable than those owned by men. Compared with reported ownership, the gap becomes more pronounced for documented ownership in Georgia and Cavite, while it is more or less the same in Mongolia. It may be noted that similar patterns are observed in Table 4 where the gender gaps in the incidence of ownership of dwelling, regardless of the type of ownership, is almost the same in Mongolia, whereas in Georgia, and in Cavite, Philippines, the gender gaps in documented ownership are much higher than the gender gaps in reported ownership. 3 Among those who did not report values of owned dwellings, at least half think that the market for selling houses is absent in their respective areas. 4 Although wealth gap between men and women can be calculated for other types of assets too, there are technical issues associated with doing such calculations. For instance, consider a case when a male respondent reported ownership of a specific parcel of land and another female respondent reported owning a parcel of land too. There is not enough information collected from the survey that can indicate whether these two respondents are referring to the same land or not (asset matching), making it difficult to estimate the total value of assets. The same line of argument applies to other types of assets except main dwelling. Thus, we focus on main dwelling only. Comparing SAO and OAAR approaches, the results suggest that estimates of incidence of reported and documented owners are generally higher using the OAAR approach than using the SAO approach. This is slightly different from the findings in Uganda (Kilic and Moylan 2016), but it is worth noting that the country context of this research is hugely different and varied in terms of legal provisions governing assets and social norms. Hence, the results from the two studies are not directly comparable.
The largest differences between the two estimation approaches are found in Georgia, where the SAO approach gives lower estimates for dwelling and agricultural land incidence rates for men and women (Table 13) implying that individuals who do not consider themselves as owners are still being identified as such by other household members. Meanwhile, estimated incidence of reported and documented ownership for dwelling in Mongolia using the SAO approach is slightly higher than the numbers produced using the OAAR approach.
In Cavite, the incidence of reported owners of all immovable assets and the incidence of documented owners of agricultural land and other real estate using SAO and OAAR approaches are not statistically different. Similar findings are noted in the incidence of reported owners of agricultural land and other real estate, and documented owners of dwelling and agricultural land in Mongolia, as well as documented owners of other real estate in Georgia. Notes: Significant t-test result implies that the incidence of ownership using SAO approach is statistically different from the incidence of ownership using OAAR approach. *** = 1% significance, ** = 5% significance, * = 10% significance. Source: Asian Development Bank estimates using the Evidence and Data for Gender Equality (EDGE) pilot surveys. There are several disadvantages in using the OAAR approach. First, there may be inadequate sharing of information within households. In addition, there are existing gender norms about asset ownership that may result to bias in proxy responses about the ownership status. There are also countries which do not clearly differentiate the ownership rights among household members and hence, responses of individual household members on who owns a particular asset may differ. Furthermore, other household members may not be fully aware of reported ownership of a particular asset as opposed to documented ownership. From a policy perspective, it might also be useful to collect self-reported data on asset ownership because the success of intervention program is more likely to be driven by people's own perception of what assets they own rather than what other people think they own (United Nations Department of Economic and Social Affairs 2019).

V. AREAS FOR FURTHER RESEARCH
While findings from the EDGE pilot surveys conducted in Georgia; Mongolia; and Cavite, Philippines provide rich inputs to the development of methodological guidelines on the collection of sexdisaggregated data on asset ownership, there are several areas that warrant further research. From a data collection perspective, it is insightful to look deeper into the gender disparities on intrahousehold reporting of asset ownership and understand the cost and benefit of covering multiple respondents versus limiting it to the primary respondent only, principal couple only, household head only, and other types of respondents in the surveyed households. In a household with principal couple respondents, for instance, the perception of men could be compared to the perception of women and investigate if women are less likely to identify themselves as owners due to gender norms. It is also important to study whether the asset owners are correctly or incorrectly identified by different household members and understand the sociodemographic profile of these asset owners. From an analytical lens, it is important to explore how total value of assets owned by individual members of the household could be estimated in light of data issues confronted. In particular, high item nonresponse rates were observed for survey questions collecting information on values of different types of assets. While missing values could be imputed, such calculations would require additional information which may not be readily available. Furthermore, estimation of values of co-owned assets requires allocation assumptions, which are highly subjective. As explained earlier, matching of assets is also needed to avoid duplication in the estimates. The possibility of coming up with indices of asset ownership, similar to the HDI or multidimensional poverty index, is another area of research that could also be explored. From a socioeconomic point of view, there are also interesting issues for further investigation. For instance, it is instructive to examine the profile of women who own specific types of assets and identify what policy makers can learn from them when designing intervention programs targeted at women who lack access to such types of productive assets.
In addition to physical assets, it is also important to examine the gender dimension of nonphysical assets such as human capital and how it interacts with ownership of physical assets. For instance, in some societies, women have slightly better educational outcomes than men as families; however, men hold more physical assets, which are often acquired through intergenerational transfer (Behrman, Pollak, and Taubman 1995;Quisumbing 1994). For a holistic understanding of gender issues in welfare and well-being, it is important to understand such dynamics through further research.

VI. DISCUSSION AND POLICY IMPLICATIONS
To provide inputs for intervention programs that aim to narrow the gap between men and women with respect to access and control over productive resources, collecting relevant sex-disaggregated data is an indispensable step. Lack of comparable individual-level data on ownership of assets is a serious constraint in shaping policy and programs that promote gender equality. However, in addition to the long duration needed to administer a household survey, financial and technical capacity constraints, and the need to contextualize data collection with varying cultural notions of how property or assets may be owned, one of the major reasons why national statistical agencies do not routinely collect sexdisaggregated data on asset ownership is the lack of standardized methodological approaches for collecting data on this topic.
The EDGE pilot surveys conducted by the national statistics offices of Georgia, Mongolia, and the Philippines, in collaboration with ADB, are powerful case studies as they have demonstrated that with the availability of standardized methods and guidelines, collecting sex-disaggregated data on asset ownership is feasible. The major contribution of these pilot surveys (along with other methodological surveys and approaches piloted in Maldives, Mexico, South Africa, and Uganda) is the inputs they provided for the development of methodological guidelines by the United Nations Statistics Division efforts under the global EDGE initiative. The resulting guidelines are based on field experience, and with minimal adaptation, can be applied by national statistics offices of other countries that have diverse geographies and social contexts.
There are several valuable lessons that can be drawn from conducting the EDGE pilot surveys. First, one needs a diverse array of indicators (e.g., incidence, distribution, forms, wealth) to facilitate a comprehensive gender analysis of asset ownership. Depending on what is being examined, the mode of data collection can be aligned with its objectives while taking each of the data collection modes' strengths and limitations into careful consideration. Second, with clarity on the survey objectives and information needs, it is possible to prioritize collecting data for a specific indicator or basket of indicators and accordingly align the survey design, survey questionnaires, and sampling methodology. Third, collecting information on values of assets through household surveys is not a simple task and it may be necessary to combine survey data with other administrative information, or to adopt sophisticated data imputation procedures.
Overall, the pilot surveys have produced extremely rich sex-disaggregated data on asset ownership and provided operationally relevant lessons for future data collection efforts. For this initiative to become a sustainable part of national statistics offices' regular data collection activities, both data producers and data users have to work together and exert a conscious effort to ensure that such data is produced regularly, is of the highest quality, and is disseminated in a timely manner. It is also incumbent on policy makers, researchers, and the development community to capitalize on the availability of such type of data to monitor the progress of gender equality.