A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam

World Development Vol. xx, pp. xxx–xxx, 2014 Ó 2013 Elsevier Ltd. All rights reserved. 0305-750X/$ - see front matter www.elsevier.com/locate/worlddev

http://dx.doi.org/10.1016/j.worlddev.2013.11.003

A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam STEPHAN KLASEN University of Goettingen, Goettingen, Germany TOBIAS LECHTENFELD World Bank, Washingten DC, USA University of Goettingen, Goettingen, Germany

and FELIX POVEL * German Development Bank (KfW), Frankfurt, Germany University of Goettingen, Goettingen, Germany Summary. — This article investigates whether heterogeneous subgroups of female-headed households are worse off than households headed by men. It analyzes the correlates of consumption, shock exposure, and severity, as well as vulnerability to poverty. Using panel data of over 4000 rural households from Thailand and Vietnam, strong evidence of heterogeneity among subgroups of female-headed households are found. In particular, in comparison with male headed households de facto female-headed ones are found to be richer in Thailand, but prone to more severe shocks in both countries. Furthermore, our results suggest that in Thailand single female-headed households are less vulnerable to poverty than households headed by men. However, in Vietnam these households are particularly poor and vulnerable to poverty; we show that this is mostly due to their greater poverty rather than their higher risk exposure. Our findings suggest that differentiation by subgroups of headship is important for policy development and targeting as well as future research. Ó 2013 Elsevier Ltd. All rights reserved. Key words — gender, poverty, shocks, vulnerability to poverty, Asia, Thailand, Vietnam

1. INTRODUCTION

Second, it analyses how different levels of endowments such as assets, land, and education affect levels of consumption, shock exposure, and vulnerability to poverty of female-headed households. Third, it accounts for the heterogeneity among female-headed households by differentiating between de facto female heads (i.e., women whose partner has migrated) and de jure female heads (i.e., households led by single women and widows), and their respective subgroups. The results are derived from a two wave panel of over 4000 households from rural areas in Thailand and Vietnam. These two countries are particularly interesting for such an analysis since their inhabitants have experienced many micro- and macro-level shocks while their poverty headcounts have fallen dramatically over the last two decades, allowing us to implicitly test the inclusiveness of these poverty trends with respect to female headship. Our findings underline the importance of differentiating between different types of female headship. In Thailand women whose husbands have migrated are generally richer than male-headed households. By contrast, in Vietnam

Female-headed households in developing countries deserve special attention since they are said to be disadvantaged regarding access to land, labor, credit, and insurance markets (World Bank, 2011). Furthermore, they may be discriminated against by cultural and social norms and suffering from, for example, high dependency burdens, little economic control over resources, and economic immobility. However, evidence regarding the poverty status of female-headed households in comparison to households headed by men remains ambiguous, often related to the heterogeneous nature of female-headed households (e.g., Chant, 2010; Duflo, 2012; Dre`ze & Srinivasan, 1997). In order to assess the situation of female-headed households in comparison to male-headed ones comprehensively, a static assessment of poverty and welfare differences is not sufficient (Buvinic & Gupta, 1997). Apart from poverty concerns, female-headed households might also be more vulnerable to poverty, as they have limited access to formal and informal credit markets and other instruments of risk management. In fact, numerous authors have asserted that women suffer from greater vulnerability (e.g., Bibars, 2001; Chant, 2010; Moghadam, 2005; World Bank, 2001). However, little empirical evidence on vulnerability and female headship exist to date. 1 This study contributes to this discussion in three ways: First, it combines a broad range of existing empirical evidence to argue that female-headed households might be particularly ill prepared to reduce and mitigate risks and cope with shocks.

* We would like to thank participants at workshops in Hue, Go¨ttingen, and Stellenbosch, conference participants in Frankfurt, Hannover, Manchester, Midrand, St. Gallen, and Paris as well as Dominique van de Walle, Hermann Waibel, and two anonymous referees for helpful comments and discussion. Financial support by the German Research Foundation (DFG) through the research unit FOR 756 is gratefully acknowledged.. 1

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

2

WORLD DEVELOPMENT

single-headed households are found to be poorer and more vulnerable than male-headed ones. In both countries, de facto female headed-households are more prone to shocks. We thus suggest that policies to tackle poverty and vulnerability need to consider the heterogeneity of female-headed households and the specific reasons for their divergent economic fortunes. This paper is structured as follows: Section 2 provides an overview about literature related to female-headed households and their well-being compared to male-headed households. Section 3 focuses on concepts and measures of vulnerability and discusses why female-headed households may be particularly vulnerable. Section 4 briefly describes gender differences in Thailand and Vietnam before embarking with the empirical analysis in Section 5. Section 6 summarizes the main results and offers concluding remarks. 2. FEMALE-HEADED HOUSEHOLDS AND POVERTY Starting in the 1990s, the “feminization of poverty” has been intensely debated among researchers and policy-makers (e.g., Chant, 1997, 2008). Reasons for an explicitly gender-related research are, among others, the observed increase of female-headed households (e.g., Budowski, Tillman, & Bergman, 2002), as well as the belief that especially these households suffer from a higher burden of poverty and vulnerability (e.g., Buvinic & Gupta, 1997). The literature about disadvantages of women in developing countries can broadly be grouped into two strands: one which focuses on gender-related differences, i.e., on differences between men and women in general, and another one which concentrates on the comparison of male and female-headed households. When using income-based measures of well-being, it is basically impossible to assess inequalities between males and females within the same households due to the presence of household-specific public goods so that well-being differences between males and females cannot be studied (Marcoux, 1998; Klasen, 2007). Assessment of well-being by headship is, however, still possible and is thus the main focus of this paper. To understand possible differences between different household types, it is nevertheless important to briefly consider the literature on the nature of economic and social disadvantages women face in developing countries. Among disadvantages for women in developing countries the lack of access to markets stand out. First, in many developing countries women in rural areas have little or no access to land. (World Bank, 2007, 2011). In some countries, women’s (co-)ownership is formally ruled out, in others it is possible but still rare. For example, Deere and Leon (2003) find that in some Latin American countries the male share of owners of farm land ranges between 70% and 90%. Underlying factors causing this inequality include inheritance and land titling laws in favor of men. 2 Second, women suffer from a limited access to formal credit markets, linked to the lack of collateral or discrimination in credit access (King, Klasen, & Porter, 2009; Storey, 2004). Husbands or other male relatives may help getting credit by co-signing loans (Fafchamps, 2000). However, such support is not automatically available and much harder to obtain for female-headed households. Third, insurance markets in (rural areas of) developing countries are—if existing at all—often hardly functioning. While both men and women are affected by such a market failure, the latter are likely to suffer more from it in the absence of a functioning safety net and equal property rights. Also, women usually have very limited possibilities to contract health insurance and may get respective access only through spouses employed in formal sector jobs (World Bank, 2001).

Fourth, women typically have less access to the labor market than men and earn less when working (Oostendorp, 2010; World Bank, 2011). Restrictions to female employment, norms limiting women’s employment in general or in particular sectors, high fertility, low female wages, all appear to play a role (Goldin, 1994; Klasen & Gaddis, 2013; World Bank, 2011). Finally, even if female shares in formal employment are comparatively high as is predominantly the case in East and Southeast Asia—women are paid significantly less than men. This wage differential cannot be explained by worker characteristics such as education and experience (e.g., Klasen, 2006; Horrace & Oaxaca, 2001; Oostendorp, 2010; Seguino, 2000). In addition to the disadvantages faced by women in general, there are disadvantages particular to female-headed households. Most importantly, households led by women carry a “double day burden” if their heads have to handle domestic work and the role of breadwinner simultaneously (Moghadam, 2005; World Bank, 2011). Consequently, these women suffer from more pronounced time and mobility constraints than others which possibly impacts negatively on income of their households (Buvinic & Gupta, 1997). Lastly, female-headed households often lack support from both social networks and the state. For example, Bibars (2001) finds that for women in Egypt there is no institutional alternative to a male provider. Chant (2008) underlines that divorced female heads often lack ties with ex-partners’ relatives, as well as with their own families and communities. However, female household heads that are married and whose husband migrated may receive increased remittances income (Buvinic & Gupta, 1997). Thus it is likely to be very important to distinguish between de jure female-headed households (consisting of widowed, divorced, and single women) and de facto ones (where a male head is temporarily absent) as opportunities and constraints might differ substantially. Despite the abundance of reasons why female-headed households may suffer more from deprivation, empirical evidence on the correlation between headship and poverty is ambiguous (Chant, 2008). During the 1980s and early 1990s studies about the “feminization of poverty” have proliferated which conclude that female-headed households are the poorest of the poor, while the number of female-headed households is on the increase in many developing countries. However, this view was quickly criticized as being unsubstantiated, leading Lipton and Ravallion (1995), Chant (2010), and Marcoux (1998) to assert that female-headed households in general are not more likely to be poor than male-headed ones. In meta analyses of this issue, Buvinic and Gupta (1997) review 61 studies concerned with the poverty status of femaleheaded households, using a very broad concept of headship. In 38 of these studies female-headed households are found to be poorer than male-headed ones. However, only certain types of female-headed households are overrepresented among the poor (supported by 15 studies) while others find no evidence that female-headed households are disproportionately among the poor (8 studies). By contrast, Quisumbing, Haddad, and Pena (2001) investigate the poverty status of female-headed households in 10 developing countries using consistent methodologies across countries. Only in two cases they find evidence that female-headed households suffer more from poverty than households headed by men. Clearly, the poverty situation of femaleheaded households varies across countries. Besides country specific contexts, the differentiated picture of gender related poverty research is owed to the heterogeneity of female-headed households. Therefore, in the literature it is increasingly refrained from superficial comparisons between male and female-headed households and switched to the analysis of

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

A FEMINIZATION OF VULNERABILITY? FEMALE HEADSHIP, POVERTY, AND VULNERABILITY IN THAILAND AND VIETNAM3

different types of the latter (Chant, 2008). On a rather aggregated level it is, as already alluded to above, useful to distinguish between de jure and de facto female-headed households. In case of the former women are the legal and customary heads. Examples are households headed by widows and unmarried, separated or divorced women. The latter have either a self-reported female head whose husband is present or, more typically, a self-reported male head who is absent for most of the time (Quisumbing et al., 2001). Studies analyzing empirically the difference between de jure and de facto female-headed households include, for instance, Chant (2010) who reports that in the Philippines de facto female-headed households had a higher per capita income than de jure female-headed households. According to Moghadam (2005), the majority of de jure female heads of household in developing countries are widows followed by divorced or separated women. Widow heads— especially those who live alone or with other elderly family members (King et al., 2009)—are said to be particularly vulnerable to poverty (e.g., World Bank, 2001). Chen and Dre`ze (1995) ascertain that in India widowhood is a cause of economic deprivation. Widow-headed households tend to have less productive assets and fewer savings than widowers, are less likely to have pension income, and often depend heavily on the economic support of their sons. Besides, single mothers have increasingly gained attention from researchers. In comparison to households in which both parents are present they lack an income earning partner and are likely to have to maintain more dependents at the same time. Consequently, they are often overrepresented among the poor (Chant, 2008). However, there are also households headed by women which may fare fairly well. For instance, de facto households headed by women whose husband migrated in order to work elsewhere may benefit from regularly sent remittances preventing them from falling into poverty (World Bank, 2001). Turning to methodological issues, the use (or neglect) of adult equivalence scales is crucial for the results of any poverty comparison between female and male-headed households. Many female-headed households typically have higher dependency ratios than households headed by men (as they are missing one adult); at the same time, they are often smaller, particularly in the case of de jure female-headed households where there are fewer children (yet) in the case of single women, or some children might no longer be there in the case of older widows or divorced women. Given the different size and structure, assessments of poverty will depend on the assumptions made about equivalence scales, including both adult equivalence of children and presumed economies of scale of larger households (e.g., Lanjouw & Ravallion, 1995). Empirically, economies of scale seem to particularly matter. When economies of scales are neglected the contribution of typically smaller households such as female-headed households to overall levels of poverty might be underestimated (Quisumbing et al., 2001). For example, Dre`ze and Srinivasan (1997) find no evidence suggesting that female-headed households—and particularly households headed by widows—are poorer than male-headed ones if they do not account for economies of scale. However, the incorporation of even fairly small economies of scale in their analysis reveals that poverty rates are much higher among single women (including single widows), widows living with unmarried children, and other female household heads. A second key methodological issue is the endogeneity of female headship status for some types of female headed households. In particular, de jure female-headed households consisting of (young) single women might be female-headed

precisely because they have the economic and social resources to live independently from their parents. Similarly, poverty could drive male members to leave a household in search of work, thereby leaving a de facto female-headed household behind. Conversely, endogeneity is likely to be less of a concern in the case of de jure female-headed households headed by widows and divorcees, as death and divorce are arguably more exogenous and less directly related to the welfare of these households. It is difficult to control for this endogeneity so that most studies discussed above should be seen as primarily descriptive that take these different types of female-headed households as given and consider their economic well-being. Such a descriptive approach is still highly policy-relevant as policy-makers want to know whether certain types of existing female headed households are poorer or more vulnerable and should be the target of specific policy interventions. Lastly, welfare implications are further complicated by different preferences of women, who, when in charge of the household budget, tend to spend more on the welfare of their children than men (Duflo, 2012). To the extent that female-headed households are consumption poorer, this would not necessarily imply lower welfare for women or their children if the fewer resources were spent in a more welfare-enhancing way. 3. FEMALE-HEADED HOUSEHOLDS AND VULNERABILITY Even though evidence concerning the relationship between female headship and poverty is unclear, it may well be that there are other dimensions of poverty where female-headed households in particular are worse off. 3 One such dimension might be vulnerability, i.e., household exposure to adverse events and the threat of future poverty. Vulnerability is a source of deprivation as adverse shocks can throw households into poverty from which they cannot easily recover. To prevent such shocks, vulnerable households often engage in costly ex ante risk-mitigation strategies (such as diversification of income sources and self-insurance; e.g., Dercon, 2005; Morduch, 2004). Recently, researchers started to design and empirically apply measures in an attempt to quantify the vulnerability of households. Pritchett, Suryahadi, and Sumarto (2000) propose the concept of vulnerability to expected poverty which defines vulnerability as the probability that a household will fall below the poverty line in future. The concept incorporates the notion of risks into poverty research. Applications of this concept are found in Chaudhuri, Jalan, and Suryahadi (2002), Christiaensen and Subbarao (2004) and Kamanou and Morduch (2004). 4 Building on this idea and established axioms of poverty measurement, Calvo and Dercon (2005) develop a measure of vulnerability to poverty. It interprets vulnerability as a probability-weighted average of future states of the world specific indices of deprivation, i.e., poverty. The measure ranges from zero to one (most vulnerable). In the empirical analysis below this measure is primarily applied to test whether female-headed households are more vulnerable to poverty. Although the literature on vulnerability has been growing in recent years, so far it rarely focuses on the shock and risk exposure of female-headed households. This negligence is surprising considering that female-headed households are often claimed to be more prone to adverse events and might be less able to cope with them (e.g., Buvinic & Gupta, 1997; Chant, 2010; Moghadam, 2005). Numerous risk-coping strategies are applied by households in developing countries when shocks occur. These include, for example, asset depletion (Fafchamps, Udry, & Czukas,

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

4

WORLD DEVELOPMENT

1998), borrowing (Udry, 1995), taking up additional occupations (Kochar, 1995), temporal migration (Lambert, 1994), drawing on governmental insurance schemes and/or informal risk-sharing networks (Townsend, 1994), as well as a change in expenditures at the expense of investment in human capital (Jacoby & Skoufias, 1997). However, when being exposed to an adverse event femaleheaded households may not be able to apply these strategies because they lack access to (i) certain assets such as land (asset depletion), (ii) credit markets (borrowing), (iii) labor markets (taking up additional occupations), and (iv) insurance markets (drawing on insurance schemes). Furthermore, they might dispose of less social capital (informal risk-sharing networks) and are restricted in their mobility (temporal migration). Of the aforementioned examples of risk management strategies merely a reduction of expenditures for the education of children seems therefore feasible. Such a coping strategy would, however, increase the likelihood of intergenerational transmission of poverty in female-headed households. Another important aspect of the impact of risks on female-headed households is that the creation of the female-headed household itself was the result of an adverse shock, linked to the endogeneity discussion above. This can happen directly if, for example, the male head dies or abandons the household, or indirectly if the male head migrates in order to help the household to cope with a shock (Quisumbing et al., 2001). On the other hand, it may also be the case that femaleheaded households are less vulnerable. In particular, lower exposure to markets could shield female-headed households from economic risks such as price shocks on the output or input side. Similarly, de facto female-headed households might particularly be able to rely on the absent household member to reduce the impact of shocks through remittances. Despite the different arguments regarding the vulnerability of female-headed households, empirical work regarding this matter is scarce. A detailed analysis using comprehensive micro-level data on shocks, risks, poverty, and vulnerability may help to provide new insights into these aspects. 4. GENDER DIFFERENCES IN THAILAND AND VIETNAM The empirical analysis focuses on Thailand and Vietnam, two countries which experienced profound economic transitions, high average growth rates, as well as great success in poverty reduction during the last decades. On the other hand, both countries have been exposed to the Asian crisis during the second half of the 1990s (particularly Thailand) and are feeling the current global economic crisis, which was accompanied by food price shocks and inflationary pressures in both countries. In addition, the region regularly suffers from natural disasters including typhoons and floods, while delayed monsoon patterns have also caused devastating droughts in recent years. In addition to idiosyncratic risks at the household level such as health shocks and violence, this volatile economic and climatic environment renders the empirical analysis of poverty and vulnerability particularly relevant for Thai and Vietnamese households (see Klasen & Waibel, 2013). The review of the country-specific literature reveals that gender differences in terms of poverty and opportunity seem to be less pronounced in Thailand and Vietnam than elsewhere. For example, in composite indices of gender inequality from the years under investigation here, such as the 2008 World Economic Forum’s Gender Gap Index (WEF, 2008) or UNDP’s 2008 Gender Inequality Index (UNDP, 2010), the countries are between ranks 50 and 70 of a sample of over 130 countries,

just below OECD and transition countries, and above most countries in Asia and Africa. This is supported by the country-specific literature. For example, Nguyen, Albrecht, Vroman, and Westbrook (2007) find for Vietnam that sex of household head is not correlated with the income quintile of a household. Also, during 1992–98 poverty reduction was more successful in the case of female- than in case of male-headed households, which is mostly due to the high share of femaleheaded households living in urban areas (Glewwe, Gragnolati, & Zaman, 2002). Moreover, gender equality in gross enrollment rates which was already quite advanced in 1985 further improved during Vietnam’s economic transition. (World Bank, 2001). Finally, Vietnamese women are overrepresented in non-agricultural wage work mainly due to their high employment shares in manufacturing industries (World Bank, 2007). However, the country seems to be well suited for an analysis of potentially marginalized and highly vulnerable groups such as female-headed households: First, the global crises (food, fuel, and finance) of 2007 and 2008 are likely to affect especially export-oriented manufacturing industries wherefore women might be more vulnerable to it than men. Second, a great deal of poverty reduction occurred in urban areas—but what happened to rural (female-headed) households, who might also be differentially affected by the food and fuel crises? Third and related to the preceding point, Vietnam’s economic development is accompanied by increasing levels of inequality between rural and urban areas and female-headed households might be left behind (e.g., Nguyen et al., 2007). In Thailand, gender gaps seem to be rather small, too. Prior to the Asian crisis in 1997 the share of women employed in the industrial and service sector increased steadily (World Bank, 2001). When the food, fuel, and financial crises materialized male employment was affected more severely than female employment because most jobs were lost in the male-dominated construction sector. Also, men’s wages were hit (slightly) harder than the ones of women during the crisis (Behrman & Tinakorn, 1999). However, Deolalikar (2002) reveals that residence in female-headed households is associated with a higher incidence of poverty. In summary, despite the rather low static gender gaps in poverty and well-being in Thailand and Vietnam, the risky environment in both countries poses the question whether (different types of) female- and male-headed households are more or less vulnerable to suffer from shocks and associated vulnerability to poverty.

5. EMPIRICAL ANALYSIS The empirical analysis relies on a household panel survey with a focus on household dynamics and vulnerability conducted in two consecutive years in 2007 and 2008. Data stem from over 4000 households in six rural provinces in Thailand and Vietnam. These include the Thai provinces of Buriram, UbonRachathani and NakhonPhanom and the Vietnamese provinces of Ha Tinh, ThuaThien-Hue, and DakLak. The provinces are predominantly rural and rank in the lowest income quintile in each country. 5 The survey instrument collects information about (i) household member characteristics such as demographics, education, and health; (ii) shocks and risks; (iii) agriculture; (iv) off-farm and self-employment; (v) borrowing, lending, public transfers, and insurance; (vi) expenditures; (vii) assets; and (viii) housing conditions. Especially the shock section of the questionnaire, which addresses numerous income, health, and social events experienced by households, is crucial to the analysis. Among

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

A FEMINIZATION OF VULNERABILITY? FEMALE HEADSHIP, POVERTY, AND VULNERABILITY IN THAILAND AND VIETNAM5

others, it aims at shedding light on the severity of adverse events by asking about their impact on income, assets, and additional expenditures (Klasen et al., 2013). Country specific differences in headship are analyzed for several outcome indicators (per capita consumption, shock exposure, shock severity, and (ex ante) vulnerability to poverty). Each outcome is analyzed with regard to three different types of headship. The analysis sets off with aggregate headship, a binary indicator equal to unity for female-headed households, and zero otherwise. Second, de jure and de facto femaleheaded households are differentiated. The third specification further distinguishes between households whose female heads are widows or singles (among de jure female-headed households) or whose female heads’ husband is absent (among de facto female-headed households). The reference group always consists of male-headed households. 6 The two survey waves enable us to mitigate issues of endogeneity and reverse causality by regressing outcome variables from 2008 on their ex-ante correlates from 2007. 7 This is our preferred approach of exploiting the panel structure of the data, because the alternatives have certain drawbacks: An analysis of changes in, for example, poverty status over time is likely to capture a lot of noise rather than true movements in and out of poverty due to the short panel interval. Household fixed-effects are not a suitable option either, because many covariates, in particular the female-headship dummies, change only for very few households, thereby basing inference on a very small sample. As a result, we can address some endogeneity issues related to other covariates, but not the endogeneity of the female-headship status itself. In that sense, our results will be descriptive but, as we argued above, still highly policy-relevant as we are able to say which types of female-headed households are descriptively more affected by poverty or vulnerability. All models are estimated with a vector of control variables: Education (highest level of completed schooling) and age of the household head are included, as well as age-squared to allow for non-linearities. Additional household controls are household size (in adult equivalents accounting for economies of scale) and dependency ratio to capture household composition; a wealth index (constructed with the first principle component of various indicators of the quality of housing); the size of land holdings (in natural logs); agricultural income (crop vs. livestock), and non-farm income activities (measured by nonexclusive sector dummies). 8 Other control variables such as indicators of ethnic minority status and remittances have been used selectively in order to control for otherwise omitted variables of interest (see also below). Location fixed effects are used to account for unobserved spatial heterogeneity. Summary statistics of all covariates by household type and country are provided in Tables 8 and 9 in the appendix. The first dependent variable is the log of per adult equivalent (World Bank scale) consumption per day of household i (ln(cons)i,2008; in USD PPP). We use adult equivalence conversion for different age-gender group and apply an economies of scale parameter of 0.8 (i.e., the household income is divided by the number of adult equivalents to the power of 0.8). This measure of consumption accounts for systematic differences between female- and male-headed households that may affect the results (as discussed above); but the results are not strongly affected by using slightly larger or smaller assumptions of economies of scale (e.g., 0.7 or 0.9). Consumption from 2008 and covariates from 2007 are used to estimate the following least squares specification separately for each country: lnðconsÞi;2008 ¼ a þ bFHH 0i;2007 þ cX 0i;2007 þ ei ;

ð1:1Þ

where FHH 0i;2007 denotes a vector of dummies for different types of female headship, X 0i;2007 is a vector of controls including household characteristics and village dummies. ei is a random error assumed to be independent and identically distributed. We use survey weights, report robust standard errors, and test for the joint significance of the female headship dummies. The coefficient of interest b indicates whether headship is significantly correlated with consumption. Changes in b due to the omission or inclusion of other covariates point to the underlying reasons of differences in consumption between female- and male-headed households. Second, shock exposure of female-headed households is examined by estimating the probability that household i experiences an adverse event during 2007–2008 (Pr(shocki,2008 = 1)). We subsume all types of shocks in the dependent variable irrespective of whether they are preventable or not. We do so because all sorts of adverse events contribute to the vulnerability of households and we are interested, among others, in households’ (in)ability to prevent shocks from happening. Also, in some cases it is not possible to distinguish between preventable and non-preventable shocks. 9 The dependent variable is modeled as a function of female headship dummies ðFHH 0i;2007 Þ, district dummies and household characteristics ðX 0i;2007 Þ, including dummies for economic activity, from 2007: Prðshock i;2008 ¼ 1Þ ¼ f ðFHH 0i;2007 ; X 0i;2007 Þ

ð1:2Þ

A probit estimator is used and survey weights are applied. Marginal effects and robust standard errors are reported. We test for the joint significance of the female headship dummies. Again, we scrutinize the underlying reasons for differential shock exposure by omitting or including other covariates than the female headship dummies. In addition, we run the same regressions with modified dependent variables capturing only income, market, agricultural supply, health, and social shocks, respectively. Third, we analyze whether female-headed households, which are struck by a shock, tend to suffer from more severe events than male-headed households. For this purpose the subsample of households that experienced a shock during 2007–2008 is used. We thus focus on household-specific differences in the ability to mitigate and cope with, respectively, the impact of a shock. The correlates of income and asset losses, as well as additional expenditures that are triggered by adverse events are estimated using a Tobit model. 10 This estimation technique allows for the dependent variable to be left censored. For example, shocks may lead to negative additional expenditures which are censored at zero in the data. That is, observed loss is given by: ( lossi;2008  if lossi;2008  > 0 ; ð1:3Þ lossi;2008 ¼ 0 if lossi;2008   0 where lossi;2008  is the actual, in case of a negative value unobserved loss of household i due to a shock during 2007–2008. The corresponding equation in the Tobit model is: lossi;2008  ¼ bFHH 0i;2007 þ dshock 0i;2008 þ cX 0i;2007 þ ei ;

ð1:4Þ

where FHH 0i;2007 is a vector of female headship dummies and X 0i;2007 a vector of village dummies and household characteristics from 2007. Shock type dummies (shock 0i;2008 ) are also included to control for the different nature of shocks. 11 Survey weights are applied, robust standard errors are reported, lagged explanatory variables are used, and joint significance of female headship dummies is tested for. Again, we examine the underlying reasons for differential shock exposure by omitting or including other covariates than the female headship dummies.

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

6

WORLD DEVELOPMENT

Eqn. (1.2) measures the probability that a household experiences a shock, Eqn. (1.4) the impact of such events on different welfare dimensions. In order to assess the overall vulnerability of female-headed households we combine both shock exposure and shock severity by deriving household specific measures of vulnerability to poverty (VTPi) as proposed by Calvo and Dercon (2005): VTP i ¼ 1 

Ni X

! pij 

xaij

; with

j¼1

0  xij  1;

to the underlying reasons of differences in vulnerability to poverty between female- and male-headed households. Table 1 presents the sample composition with regard to headship in 2007 (wave 1). In the sample from three rural provinces, 20.8% of all Thai households are female-headed. Of these, 79.6% households are de jure and 20.4% are de facto femaleheaded. Further disaggregation of subgroups shows that de jure female-headed households are largely widow headed (83.0%), with the remainder (17.0%) being single (unmarried or divorced) female-headed households. The de facto group consists entirely of households of absent male spouses. Fewer female-headed households are found in rural Vietnam, where only 15.1% of all households belong to this category. Of these, 81.4% are de jure female-headed households, which themselves consist of 76.0% widows and 24.0% single female heads. The de facto group constitutes 18.6% of all female-headed households. The results for the second round in 2008 are very similar and can be found in Table 6 in the appendix. A marginal decrease (increase) in Thailand (Vietnam) of the share of female-headed households of less than 1% is observed, which is too small to apply econometric differencing methods with any precision. The composition of female-headed households also remains unchanged. Attrition equally affects all headship types and is generally very low at 2.1% (Vietnam) and 2.5% (Thailand). Table 2 shows the result of OLS regressions which measure the country specific correlation between equivalized consumption in 2008 and female headship in 2007 controlling for other covariates. No significant differences between (aggregated) femaleand male-headed households are found—neither in Thailand, nor in Vietnam (columns 1 and 2). Splitting female headship into de facto and de jure provides a richer picture (columns 3 and 4): In Thailand de facto female-headed households spend on average 12.8% more per adult than male-headed households (significant at 10% level). In Vietnam, de jure female-headed households spend significantly less (8.2%) than male-headed ones (significant at 5% level). The latter result can be attributed to single female-headed households whose average spending is 22.6% below the consumption of male-headed households (significant at 1% level; column 6). The correlation between other covariates and consumption is as expected. In Thailand the significant and positive correlation between consumption and de facto female headship is robust to the omission and inclusion of different controls (Table 10 in the

ð1:5Þ

Ni X pij ¼ 1 and 0  a  1; j¼1

where pij denotes the probability of state of the world j to occur, ~y xij is a state-specific degree of deprivation which equals zij , and ~y ij is a censored outcome measure. That is, all outcomes where yij is above the poverty line z are censored at z and consequently do not change the vulnerability measure. VTPi ranges between zero and one. There is a total of Ni possible states of the world. The closer (further away) a moves to (from) one the less (more) risk aversion is assumed. District specific probabilities are predicted, as well as household specific severities of different states of the world. The product of these predictions is used to calculate VTPi. The risk aversion parameter a is set equal to 0.5 and a modest $2 per day per capita (USD PPP) is used as poverty line. We use the same poverty lines in order to allow for cross-country comparisons of our results. The vulnerability to the poverty score from 2008 and covariates from 2007 are used to estimate the following least squares specification separately for each country: VTP i;2008 ¼ a þ bFHH 0i;2007 þ cX 0i;2007 þ ei ;

ð1:6Þ

FHH 0i;2007

where is a vector of dummies for different types of female headship, X 0i;2007 is a vector including village dummies and household characteristics, and ei is a random error assumed to be independent and identically distributed. Survey weights are applied, robust standard errors are reported, and the joint significance of the binary indicators for subgroups of female headship is tested for. The coefficient of interest b indicates whether headship is significantly correlated with vulnerability to poverty. Changes in b due to the omission or inclusion of other covariates point

Table 1. Headship and sample size (Wave 1, 2007) Country

Female

Male

Thailand

451 20.8%

1724 79.2%

Vietnam

323 15.1%

1867 84.9%

Thailand Vietnam

Thailand Vietnam

De Jure

De Facto

359 79.6% 265 81.4%

92 20.4% 58 18.6%

Widow

Single

Absent husband

298 83.0% 202 76.0%

61 17.0% 63 24.0%

92 100.0% 58 100.0%

Note: Percent by subgroup is using survey weights.

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

A FEMINIZATION OF VULNERABILITY? FEMALE HEADSHIP, POVERTY, AND VULNERABILITY IN THAILAND AND VIETNAM7 Table 2. Correlates of consumption per adult equivalent OLS: ln (consumption)

(1)

(2) Female head

Female head

(3)

(4)

De Facto vs. De Jure

Thailand

Vietnam

0.046 (0.0343)

0.049 (0.0355)

De Facto FHH De Jure FHH

Thailand

Vietnam

0.128* (0.0655) 0.023 (0.0390)

0.078 (0.0638) 0.082** (0.0403)

Head: Middle Edu Head: Secondary Edu Head: Tertiary Edu Age of head Age square Housing index ln (land) Non-farm sector Crops sector Livestock sector Constant

Households Adjusted R-squared Headship joint signif. Headship prob > F

Vietnam

0.074 (0.0637) 0.035 (0.0373) 0.226*** (0.0836) 0.120*** (0.0192) 0.059*** (0.0171) 0.092** (0.0363) 0.205*** (0.0363) 0.314*** (0.0474) 0.494*** (0.0528) 0.024*** (0.0062) 0.000*** (0.0001) 0.113*** (0.0232) 0.081*** (0.0145) 0.057** (0.0236) 0.229*** (0.0527) 0.055 (0.0335) 7.519*** (0.1640) 2130 0.360 3.101 0.030

0.119*** (0.0192) 0.056*** (0.0172) 0.098*** (0.0363) 0.212*** (0.0361) 0.323*** (0.0466) 0.507*** (0.0515) 0.023*** (0.0062) 0.000*** (0.0001) 0.114*** (0.0230) 0.083*** (0.0142) 0.054** (0.0237) 0.230*** (0.0529) 0.050 (0.0339) 7.540*** (0.1629)

0.087*** (0.0183) 0.072*** (0.0166) 0.104** (0.0418) 0.257*** (0.0702) 0.389*** (0.0801) 0.706*** (0.0907) 0.007 (0.0075) 0.000 (0.0001) 0.147*** (0.0130) 0.111*** (0.0119) 0.064** (0.0271) 0.139*** (0.0463) 0.080** (0.0326) 7.808*** (0.2036)

0.119*** (0.0191) 0.059*** (0.0171) 0.090** (0.0369) 0.204*** (0.0364) 0.314*** (0.0472) 0.495*** (0.0527) 0.023*** (0.0062) 0.000*** (0.0001) 0.113*** (0.0232) 0.082*** (0.0144) 0.058** (0.0238) 0.228*** (0.0533) 0.051 (0.0337) 7.523*** (0.1640)

2113 0.248 1.763 0.187

2130 0.357 1.909 0.170

2113 0.249 2.046 0.134

2130 0.358 2.999 0.054

2113 0.249 2.085 0.106

FHH, single

Head: Primary Edu

Thailand

0.088*** (0.0183) 0.069*** (0.0161) 0.107** (0.0420) 0.258*** (0.0701) 0.389*** (0.0803) 0.710*** (0.0912) 0.006 (0.0074) 0.000 (0.0001) 0.148*** (0.0131) 0.111*** (0.0119) 0.059** (0.0264) 0.136*** (0.0467) 0.080** (0.0327) 7.846*** (0.1995)

FHH, widow

Dep. ratio

(6) FHH subgroups

0.131** (0.0655) 0.006 (0.0406) 0.104 (0.0724) 0.086*** (0.0184) 0.073*** (0.0164) 0.103** (0.0415) 0.258*** (0.0700) 0.387*** (0.0804) 0.702*** (0.0904) 0.007 (0.0075) 0.000 (0.0001) 0.147*** (0.0130) 0.112*** (0.0120) 0.064** (0.0272) 0.138*** (0.0465) 0.080** (0.0327) 7.802*** (0.2045)

FHH, absent husband

HH size (adult equiv)

(5)

Notes: Survey weights used; robust standard errors in parentheses; outcome: log of consumption in 2008 in USD PPP per day per adult equivalent (World Bank scale) with economies of scale (0.8); all covariates are from 2007; income sector dummies are not exclusive (no reference group); reference group for educational attainment: no education; village dummies included but not reported. Significance levels: * p < 0.1. ** p < 0.05. *** p < 0.01.

appendix). Including further control variables such as a dummy, which equals one if households are net receivers of remittances, does not change this correlation considerably (results not shown). This suggests that remittance receipt itself is not the driver of this higher well-being. 12 It could therefore be that de facto female-headed households are drawn from a better-off group of households in Thailand. Furthermore, widow-headed Thai households are significantly consumption poorer than male-headed ones if we do

not include other control variables. Plugging in further household head characteristics and indicators for wealth indicates that the former tends to be less educated and less wealthy than the latter; thus this lower education and lower asset base are explaining their lower consumption (columns 5 and 7). In Vietnam, the correlation between consumption and singleheaded households remains significantly negative if different controls are omitted or included. However, columns 6 and 8 suggest that these households tend to have less educated and

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

8

WORLD DEVELOPMENT

possibly younger heads and are less wealthy than male-headed ones. All in all, in terms of consumption we do not find compelling evidence that female-headed households are generally worse off than male-headed ones, which is in line with, for example, Quisumbing et al. (2001). Rather there seems to be a great deal of heterogeneity both among female-headed households with de facto female-headed households being better off Thailand and de jure, particularly single, ones being worse off in Vietnam.

After having established the correlation between different types of female headship and consumption the focus now shifts to shock exposure of female-headed households. Table 7 in the appendix shows that the incidence of adverse events differs between both countries: The 66.5% of Vietnamese households that report an adverse event exceed the 46.6% of Thai households that are affected by shocks. Generally, agricultural supply shocks which include, among others, adverse weather events such as storms and droughts are most common in both countries

Table 3. Correlates of shock exposure Probit: any shock

(1)

(2)

(3)

Thailand

Vietnam

Thailand

Vietnam

0.014 (0.0290)

0.012 (0.0310) 0.022 (0.0594) 0.011 (0.0321)

0.014 (0.0682) 0.011 (0.0337)

Female head

Female head

(4)

De Facto vs. De Jure

De Facto FHH De Jure FHH FHH, absent husband

Head: Primary Edu Head: Middle Edu Head: Secondary Edu Head: Tertiary Edu Age of head Age sq. Housing index Ln (land) Non-farm sector Crops sector Livestock sector

Observations Pseudo R2 Model Chi2 Model prob > Chi2 Headship joint signif Headship prob > Chi2

FHH subgroups Thailand

Vietnam

0.014 (0.0682) 0.014 (0.0377) 0.004 (0.0588) 0.017 (0.0144) 0.013 (0.0145) 0.012 (0.0321) 0.048 (0.0319) 0.025 (0.0389) 0.121** (0.0533) 0.003 (0.0052) 0.000 (0.0000) 0.040** (0.0175) 0.030*** (0.0099) 0.022 (0.0229) 0.188*** (0.0377) 0.022 (0.0250) 2122 0.206 510.385 0.000 0.164 0.983

0.035** (0.0157) 0.002 (0.0154) 0.019 (0.0369) 0.028 (0.0643) 0.060 (0.0691) 0.212*** (0.0810) 0.007 (0.0060) 0.000 (0.0001) 0.015 (0.0116) 0.027*** (0.0101) 0.011 (0.0258) 0.040 (0.0390) 0.010 (0.0273)

0.017 (0.0144) 0.014 (0.0144) 0.012 (0.0318) 0.048 (0.0317) 0.025 (0.0387) 0.121** (0.0531) 0.003 (0.0052) 0.000 (0.0000) 0.040** (0.0175) 0.030*** (0.0099) 0.022 (0.0228) 0.188*** (0.0376) 0.023 (0.0249)

0.035** (0.0157) 0.002 (0.0155) 0.018 (0.0370) 0.028 (0.0643) 0.059 (0.0691) 0.213*** (0.0810) 0.007 (0.0061) 0.000 (0.0001) 0.015 (0.0116) 0.027*** (0.0101) 0.012 (0.0260) 0.040 (0.0390) 0.010 (0.0273)

0.017 (0.0144) 0.014 (0.0145) 0.012 (0.0320) 0.048 (0.0319) 0.025 (0.0389) 0.121** (0.0533) 0.003 (0.0052) 0.000 (0.0000) 0.040** (0.0175) 0.030*** (0.0099) 0.022 (0.0229) 0.188*** (0.0377) 0.023 (0.0249)

2116 0.044 124.934 0.000 0.221 0.638

2122 0.206 510.360 0.000 0.138 0.710

2116 0.044 124.963 0.000 0.249 0.883

2122 0.206 510.361 0.000 0.140 0.933

2116 0.044 125.221 0.000 0.507 0.917

FHH, single

Dep. ratio

(6)

0.024 (0.0595) 0.005 (0.0346) 0.042 (0.0681) 0.035** (0.0158) 0.002 (0.0155) 0.018 (0.0370) 0.027 (0.0643) 0.060 (0.0691) 0.214*** (0.0810) 0.007 (0.0061) 0.000 (0.0001) 0.015 (0.0116) 0.027*** (0.0101) 0.012 (0.0260) 0.040 (0.0390) 0.010 (0.0273)

FHH, widow

HH size (adult equiv)

(5)

Notes: Survey weights used; robust standard errors in parentheses; marginal effects reported; outcome: dummy equal to 1 if household experienced any shock during 2007–2008; all covariates are from 2007; income sector dummies are not exclusive (no reference group); reference group for educational attainment: no education; constant and district dummies included but not reported. Significance levels: * p < 0.1. ** p < 0.05. *** p < 0.01.

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

A FEMINIZATION OF VULNERABILITY? FEMALE HEADSHIP, POVERTY, AND VULNERABILITY IN THAILAND AND VIETNAM9

(28% affected households in Thailand, 51% in Vietnam). This suggests that especially climatic volatility poses a serious threat to the wellbeing of households. Health shocks are also pronounced in the sample with 16.7% of Thai and 23.9% of

Vietnamese households being affected. The high incidence of health related hazards is mainly driven by illness of a household member. This result is broadly in line with Wagstaff and Lindelow (2010) who find that in neighboring Laos illnesses are the

Table 4. Correlates of shock severity Tobit: asset loss (ln)

(1)

(2) Female head

Female head

Thailand

Vietnam

1.079 (1.1619)

1.107* (0.5979)

De Facto FHH De Jure FHH

(3)

(4)

De Facto vs. De Jure Thailand

Vietnam

0.078 (2.2600) 1.421 (1.3638)

2.297* (1.1866) 0.739 (0.7018)

FHH, absent husband

Social shock HH size (adult equiv) Dep. ratio Head: Primary Edu Head: Middle Edu Head: Secondary Edu Head: Tertiary Edu Age of head Age sq. Housing index Ln (land) Non-farm sector Crops sector Livestock sector Observations Pseudo R2 Model Chi2 Headship joint signif. Headship prob > Chi2

Thailand

Vietnam

2.305* (1.1879) 0.638 (0.7831) 1.150 (1.2211) 0.046 (1.4849) 4.203*** (0.9732) 4.981*** (0.7191) 0.006 (0.3437) 0.493 (0.4186) 0.392 (0.7555) 0.411 (0.8448) 0.903 (1.0554) 2.844* (1.4683) 0.199 (0.1216) 0.002* (0.0012) 0.395 (0.5003) 0.150 (0.2916) 0.655 (0.5088) 2.202** (1.0803) 0.092 (0.8366) 1564 0.154 586.363 1.751 0.155

0.225 (0.8381) 4.112*** (1.1371) 7.541*** (0.9315) 0.555 (0.4409) 0.180 (0.6481) 1.174 (1.7479) 4.509 (2.7989) 3.023 (2.9569) 2.603 (3.9615) 0.195 (0.2121) 0.002 (0.0018) 0.438 (0.4486) 0.109 (0.4227) 1.608 (1.1291) 1.861 (1.4431) 1.185 (1.2087)

0.038 (1.4678) 4.157*** (0.9785) 5.023*** (0.7263) 0.002 (0.3450) 0.442 (0.4142) 0.303 (0.7510) 0.535 (0.8436) 0.789 (1.0615) 2.732* (1.4742) 0.185 (0.1246) 0.002 (0.0012) 0.380 (0.4997) 0.138 (0.2905) 0.605 (0.5098) 2.174** (1.0714) 0.090 (0.8345)

0.225 (0.8387) 4.106*** (1.1373) 7.529*** (0.9306) 0.576 (0.4422) 0.129 (0.6465) 1.112 (1.7435) 4.482 (2.7969) 2.987 (2.9549) 2.569 (3.9445) 0.210 (0.2165) 0.002 (0.0018) 0.422 (0.4475) 0.105 (0.4244) 1.550 (1.1264) 1.780 (1.4612) 1.230 (1.2173)

0.055 (1.4789) 4.205*** (0.9746) 4.972*** (0.7247) 0.001 (0.3446) 0.493 (0.4185) 0.384 (0.7597) 0.415 (0.8504) 0.901 (1.0579) 2.837* (1.4712) 0.199 (0.1214) 0.002* (0.0012) 0.398 (0.5010) 0.151 (0.2909) 0.654 (0.5088) 2.196** (1.0812) 0.103 (0.8337)

1290 0.100 234.472 0.863 0.353

1564 0.154 585.127 3.428 0.064

1290 0.100 234.792 0.543 0.581

1564 0.154 586.260 2.522 0.081

1290 0.100 235.948 0.586 0.624

FHH, single

Agricultural supply shock

(6) FHH subgroups

0.082 (2.2620) 0.916 (1.4505) 5.014 (4.2549) 0.190 (0.8392) 4.128*** (1.1344) 7.554*** (0.9286) 0.556 (0.4436) 0.139 (0.6473) 1.139 (1.7469) 4.529 (2.7694) 3.021 (2.9518) 2.419 (3.9734) 0.224 (0.2166) 0.002 (0.0018) 0.406 (0.4473) 0.083 (0.4219) 1.537 (1.1255) 1.605 (1.4480) 1.280 (1.2201)

FHH, widow

Market shock

(5)

Notes: Survey weights used; robust standard errors in parentheses; marginal effects reported; outcome: asset loss in ln(USD PPP) due to shocks experienced during 2007–2008; all covariates are from 2007 (except shock group dummies); income sector dummies are not exclusive (no reference group); reference group for shock groups: health shocks; reference group for educational attainment: no education; sample reduced to households with shock experience; constant and district dummies included but not reported. Significance levels: * p < 0.1. ** p < 0.05. *** p < 0.01.

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

10

WORLD DEVELOPMENT

most common type of household-specific shocks. Market shocks such as price shocks and job loss play a more important role in Thailand (incidence of 12.7%) than in Vietnam (2.6%). This finding may reflect that Thai village economies in the sample are economically more diversified than Vietnamese ones which explains why they have more non-agricultural income sources at risk. The share of households suffering from social shocks (including migration, a costly social obligation, a crime event, or damage to the house) is similar in both countries (7.5% in Thailand and 6.3% in Vietnam).

In order to further scrutinize these insights probit regressions are run, where the dependent variable equals one if the observed household experiences any shock during 2007–2008 and zero otherwise (Table 3). All marginal effects of femaleheaded household dummies are positive suggesting that these types of households tend to be more exposed to shocks on average when it is controlled for other observables. However, none of the respective coefficients is significant. Also the tests for joint significance fail to reject the null hypothesis that the female headship dummies are jointly insignificant (bottom row

Table 5. Correlates of vulnerability to poverty OLS: vulnerability

(1)

(2) Female head

Female head

(3)

(4)

De Facto vs. De Jure

Thailand

Vietnam

0.003 (0.0058)

0.014 (0.0135)

De Facto FHH De Jure FHH

Thailand

Vietnam

0.003 (0.0144) 0.005 (0.0068)

0.002 (0.0265) 0.017 (0.0144)

FHH, absent husband

Head: Primary Edu Head: Middle Edu Head: Secondary Edu Head: Tertiary Edu Age of head Age square Housing index Ln (land) Non-farm sector Crops sector Livestock sector Constant Observations Adjusted R-squared Headship joint signif Headship prob > F

Thailand

Vietnam

0.003 (0.0265) 0.004 (0.0148) 0.066** (0.0311) 0.022*** (0.0055) 0.021*** (0.0058) 0.016 (0.0153) 0.039** (0.0147) 0.033** (0.0157) 0.055** (0.0245) 0.003 (0.0026) 0.000 (0.0000) 0.006 (0.0064) 0.005 (0.0079) 0.011 (0.0071) 0.006 (0.0285) 0.017 (0.0126) 0.015 (0.0831) 1030 0.196 1.487 0.222

0.004 (0.0032) 0.005 (0.0041) 0.026** (0.0121) 0.029* (0.0152) 0.044*** (0.0140) 0.038** (0.0151) 0.001 (0.0013) 0.000 (0.0000) 0.010*** (0.0030) 0.004 (0.0028) 0.002 (0.0055) 0.001 (0.0076) 0.003 (0.0067) 0.032 (0.0388)

0.022*** (0.0055) 0.021*** (0.0061) 0.015 (0.0153) 0.038*** (0.0145) 0.034** (0.0156) 0.055** (0.0240) 0.003 (0.0026) 0.000 (0.0000) 0.006 (0.0064) 0.005 (0.0080) 0.010 (0.0073) 0.005 (0.0283) 0.017 (0.0123) 0.011 (0.0842)

0.004 (0.0032) 0.005 (0.0041) 0.026** (0.0122) 0.029* (0.0152) 0.044*** (0.0141) 0.038** (0.0151) 0.000 (0.0013) 0.000 (0.0000) 0.010*** (0.0030) 0.004 (0.0028) 0.002 (0.0057) 0.001 (0.0075) 0.003 (0.0066) 0.029 (0.0396)

0.022*** (0.0055) 0.021*** (0.0059) 0.015 (0.0153) 0.037** (0.0145) 0.033** (0.0156) 0.054** (0.0244) 0.003 (0.0026) 0.000 (0.0000) 0.006 (0.0064) 0.005 (0.0079) 0.011 (0.0073) 0.005 (0.0286) 0.017 (0.0124) 0.013 (0.0833)

842 0.078 0.356 0.552

1030 0.192 1.115 0.293

842 0.078 0.302 0.740

1030 0.191 0.714 0.492

842 0.077 1.826 0.147

FHH, single

Dep. ratio

(6) FHH subgroups

0.003 (0.0144) 0.002 (0.0077) 0.016** (0.0070) 0.004 (0.0033) 0.005 (0.0041) 0.026** (0.0122) 0.029* (0.0153) 0.044*** (0.0142) 0.037** (0.0146) 0.000 (0.0013) 0.000 (0.0000) 0.010*** (0.0030) 0.003 (0.0028) 0.002 (0.0057) 0.002 (0.0076) 0.003 (0.0066) 0.028 (0.0396)

FHH, widow

HH size (adult equiv)

(5)

Notes: Survey weights used; robust standard errors in parentheses; outcome: vulnerability to poverty as proposed by Calvo and Dercon (2005) in 2008; all covariates are from 2007; income sector dummies are not exclusive (no reference group); reference group for educational attainment: no education; village dummies included but not reported. Significance levels: * p < 0.1. ** p < 0.05. *** p < 0.01.

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

A FEMINIZATION OF VULNERABILITY? FEMALE HEADSHIP, POVERTY, AND VULNERABILITY IN THAILAND AND VIETNAM11

of Table 3). Coefficients of the other covariates are largely as expected. Logged land holdings are a significant and positive correlate of the dependent shock dummy in both countries. That is, land seems to be a proxy for the amount of crops planted, i.e., at risk, as opposed to wealth. Many planted crops may render households more susceptible to events such as crop pests and bad weather. This would also help to explain the positive association between engagement in the crop sector and shock exposure in Vietnam. In both countries the correlations between shock exposure and female headship dummies remain statistically insignificant irrespective of the inclusion or omission of additional controls (Table 11 in the appendix). Moreover, we run regressions similar to the ones in Table 3 with modified dependent variables capturing only income, market, agricultural supply, health, and social shocks, respectively. The coefficients of female headship dummies are largely insignificant in these regressions (results not reported). In other words, we do not find statistical evidence that female-headed households in general, or particular types of these households, are more prone to shocks than male-headed ones in our rural sample provinces in Thailand and Vietnam. Next we turn to the severity of adverse events of those households that experienced a shock. Table 4 shows marginal effects from a Tobit regression of logged asset loss in USD PPP due to shocks during 2007–2008 on shock group dummies, with health shocks being the reference group, and additional control variables from 2007. In Thailand losses of female-headed households are not significantly different from losses of male-headed households. By contrast, Vietnam households led by women lose more on average (significant at 10% level; column 2). This is driven by de facto femaleheaded households (columns 4 and 6), which contribute to the joint significance of female headship dummies in two out of three specifications (columns 2 and 4). Dependence on an uncertain remittance income from an absent male is presumably the main reason for this effect. 13 With respect to the other covariates we find that agricultural supply and social shock dummies are significant and positive across all specifications. Vietnamese households engaged in the crop sector lose significantly fewer assets due to a shock most likely because they have less to lose than richer non-farm households. 14 Also, we run Tobit regressions of the same type with logged income loss and additional expenditures in USD PPP as dependent variables (results not reported). While de facto female-headed households in Vietnam lose significantly more assets due to shocks, their Thai counterparts lose significantly more income. In terms of extra expenditures households headed by (different types of) women are not significantly worse off than male-headed households. Overall, these results suggest that shocks experienced by de facto female-headed households in both countries are relatively severe. Given that in both countries de facto femaleheaded households are on average richer than male-headed households (Tables 8 and 9 in the appendix) this finding may be driven by the fact that the former have more wealth to lose than the latter. In order to scrutinize this explanation an asset index is added as covariate to the regression with asset loss as dependent variable and a variable capturing the income decile to which households belong as covariate to the regression with income loss as dependent variable (results not shown). The coefficients of female headship dummies are found to be robust to the inclusion of these variables. Finally, we analyze the determinants of vulnerability to poverty in 2008 using lagged covariates from 2007. Neither in Thailand nor in Vietnam female-headed households are statistically different from male-headed ones (Table 5, columns 1

and 2). This insight does not change if female headship is disaggregated into de jure and de facto (columns 3 and 4). However, disaggregating the headship dummies even further, we find that single-headed households are significantly less vulnerable to poverty in Thailand but significantly more in Vietnam (columns 5 and 6; in both cases at 5% level). When examining the results for this group of households in Tables 2–5, we see that the higher vulnerability to poverty of single female-headed household in Vietnam is primarily due to their greater average consumption poverty, rather than their higher exposure to shocks. Since they are poor in most states of the world, the vulnerability measure used here will classify them as particularly vulnerable. 15 For Thailand, the comparison across tables suggests that the reason for the lower vulnerability of single female-headed households there is related to their lower consumption poverty as well as their lower severity of shocks. The other covariates in Table 5 exhibit plausible correlations with the dependent variable. In Thailand the significant and negative correlation between vulnerability to poverty and single-headed households is robust to the omission and inclusion of different controls, suggesting that this group has rather stable income sources. Overall, shedding light on the vulnerability to poverty of different types of female-headed households adds value to our analysis. Certainly, in Vietnam single-headed households are not just more vulnerable to poverty but also consumption poorer (Table 2). But in Thailand none of the previous findings pointed at single-headed households being less vulnerable to poverty than male-headed ones. 6. CONCLUSION Female-headed households are sometimes thought to be disadvantaged regarding their level of consumption, shock exposure, and vulnerability. Despite such assertions research regarding the relative poverty status of female-headed households is inconclusive and there is little empirical evidence regarding their vulnerability. The study aims to contribute to the discussion by scrutinizing whether female-headed households from rural Thailand and Vietnam (i) consume less, (ii) are more exposed to shocks, (iii) suffer from more severe shocks, (iv) and are more vulnerable to poverty than male-headed households. In order to study the heterogeneity of households headed by women the analysis differentiates between de jure and de facto female-headed households. By distinguishing between households headed by singles, widows, as well as women whose husband is absent female-headed households are disaggregated even further. Thus, it is possible to reveal systematic differences within this group of households. All in all, we do not find compelling evidence that femaleheaded households as a group are generally worse off than male-headed ones, but the heterogeneity of female-headed households matters: In comparison with male headed households de facto female-headed ones are in fact consumption richer in Thailand but tend to be prone to more severe shocks in both countries (in terms of income loss in Thailand and in terms of asset loss in Vietnam). With respect to de jure female-headed households our results suggest that in Vietnam only single-headed households are particularly consumption poor and vulnerable to poverty. In contrast, in Thailand they are less vulnerable to poverty than households headed by men. In other words, different types of female-headed households vary substantially from each other in terms of poverty and vulnerability, except for shock exposure. This holds true for both Thailand and Vietnam. These results are descriptive in the

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

12

WORLD DEVELOPMENT

sense that endogeneity of the female-headship might account for the results: for example, de facto female headed households in Thailand might consist of a selected group of households that are better off in unobservable characteristics. But we argue that even these descriptive results matter. Household headship is often used in many developing countries for targeting cash transfers, social health insurance, agricultural training, and programs to provide increased access to productive resources (e.g., Quisumbing & Pandolfelli, 2010). However, our results do not suggest a general need for targeting all types of female-headed households in rural Thailand and Vietnam. Instead, primarily households headed by singles

in Vietnam seem to deserve policy attention; additionally, de facto female-headed households might need particular support to deal with the substantial shocks and vulnerabilities they suffer from. Future research on poverty and vulnerability of femaleheaded households should account for different types of headship. From our point of view it would be of great interest to apply our analysis to other contexts where larger gender differences exist than in Southeast Asia, such as South Asia or selected SubSaharan African countries. As this would require detailed data on shock exposure, probability, and severity, such an extension would require new survey data in many cases.

NOTES 1. An exception is Glewwe and Hall (1998) who scrutinize female-headed households’ vulnerability to macroeconomic shocks in Peru and do not find that this group of households is disproportionally vulnerable. 2. For evidence on gender inequalities in access to land, see World Bank (2011) as well as the OECD’s Social Institutions and Gender Index (e.g. Branisa, Klasen, & Ziegler, 2013). 3. A prominent dimension other than income and consumption analyzed in the context of gender research is time. Women are particularly vulnerable to time poverty since they do domestic work in addition to income generating activities. Especially women with a double day burden such as single mothers suffer from time poverty even if they are not deprived in terms of income or consumption (World Bank, 2011; UNDP, 1995). 4. Other approaches include vulnerability as low expected utility (Ligon & Schechter, 2003) and vulnerability as uninsured exposure to shocks (Townsend, 1994). See Klasen and Povel (2013) for a survey. 5. See Hardeweg, Klasen, and Waibel (2013) and Klasen, Lechtenfeld, and Povel (2013) for more detailed information on sampling, field work, and approaches to capture risks and shocks. 6. It is not further disaggregated because there are only very few male single- or widower-headed households or male-headed households with an absent spouse in the data. That is, we can only compare different types of female-headed households with male-headed households in general but not with their male-headed counterparts in particular. 7. Although we regress outcome variables from 2008 on headship dummies from 2007, we cannot discard the possibility that the femaleheaded households in our sample were formed endogenously.

9. For example, to what extent is a drought or flood preventable by adequate irrigation or flood-control measures? 10. The respective amounts were elicited by the following questions: What is (i) the estimated total loss of income due to the event in the reference period?; (ii) the estimated loss of assets due to the event in the reference period?; (iii) the estimated total extra expenditure due to the event in the reference period? 11. The shock type dummies include market, agricultural supply, health and social shocks. 12. Since we only control for remittances rather crudely using this dummy, remittance amounts or the ability to receive remittances when needed might still play a role in the improved well-being of these households. 13. At the same time, it is possible that high risk caused migration of the male household head. We try to control for this to the extent possible using covariates and lagged information but cannot fully rule this out. 14. Recall that results from Table 2 suggest that Vietnamese households engaged in the crop sector are significantly consumption poorer than the ones not engaged in this sector. 15. One may argue that this constitutes a problem of the vulnerability measure used. Singe female-headed households appear to be rather chronically poor than suffering from shocks and risks. In that sense, the measure of downside risk proposed by Povel in this special issue does not suffer from this problem since it focuses on downside risks and not on the likelihood to be below the poverty line as the Calvo and Dercon measure we use here does.

8. When investigating the correlates of shock severity shock-type dummies are additionally included.

REFERENCES Behrman, J. R., & Tinakorn, P. (1999). The impact of the Thai financial crisis on Thai labor. Bangkok: Thailand Development Research Institute. Bibars, I. (2001). Victims and heroines: Women, welfare and the Egyptian state. London: Zed. Branisa, B., Klasen, S., & Ziegler, M. (2013). Gender inequality in social institutions and gendered development outcomes. World Development, 45, 252–268. Budowski, M., Tillman, R., & Bergman, M. M. (2002). Poverty, stratification, and gender. Swiss Journal of Sociology, 28(2), 297–317.

Buvinic, M., & Gupta, G. R. (1997). Female-headed households and female-maintained families: Are they worth targeting to reduce poverty in developing countries?. Economic Development and Cultural Change, 45(2), 259–280. Calvo, C., & Dercon, S. (2005). Measuring individual vulnerability: Department of Economics discussion paper series 229. Oxford University. Chant, S. (1997). Women-headed households: Poorest of the poor? Perspectives from Mexico, Costa Rica and the Philippines. IDS Bulletin, 28(3), 26–48.

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

A FEMINIZATION OF VULNERABILITY? FEMALE HEADSHIP, POVERTY, AND VULNERABILITY IN THAILAND AND VIETNAM13 Chant, S. (Ed.) (2010). The international handbook of gender and poverty: Concepts, research, policy. Cheltenham, UK: Edward Elgar. Chant, S. (2008). Dangerous equations? How female-headed households became the poorest of the poor: Causes, consequences and cautions. In J. Momsen (Ed.), Gender and development: Critical concepts in development studies. Theory and classics (Vol. I). London, UK: Routledge. Chaudhuri, S., Jalan, J., & Suryahadi, A. (2002). Assessing household vulnerability to poverty from cross-sectional data: A methodology and estimates from Indonesia: Discussion paper 0102-52. Columbia University. Chen, M., & Dre`ze, J. (1995). Recent research on widows in India: Workshop and conference report. Economic and Political Weekly, 30(39), 2435–2450. Christiaensen, L., & Subbarao, K. (2004). Toward an understanding of household vulnerability in rural Kenya: World Bank policy research working paper 3326. Washington DC: The World Bank. Deere, C. D., & Leon, M. (2003). The gender asset gap: Land in Latin America. World Development, 31(6), 925–947. Deolalikar, A. B. (2002). Poverty, growth, and inequality in Thailand: Economics and research department working paper 8. Manila: Asian Development Bank. Dercon, S. (2005). Vulnerability: A micro perspective. Paper presented at the Annual Conference on Development Economics, May 23–24, Amsterdam, The Netherlands. Dre`ze, J., & Srinivasan, P. V. (1997). Widowhood and poverty in rural India: Some inferences from household survey data. Journal of Development Economics, 54(2), 217–234. Duflo, E. (2012). Women’s empowerment and economic development: CEPR discussion paper 8734. Fafchamps, M. (2000). Ethnicity and credit in African manufacturing. Journal of Development Economics, 61(1), 205–235. Fafchamps, M., Udry, C., & Czukas, K. (1998). Drought and savings in West Africa: Are livestock a buffer stock?. Journal of Development Economics, 55(2), 273–305. Glewwe, P., Gragnolati, M., & Zaman, H. (2002). Who gained from Vietnam’s boom in the 1990’s? An analysis of poverty and inequality trends. Economic Development and Cultural Change, 50(4), 773–792. Glewwe, P., & Hall, G. (1998). Are some groups more vulnerable to macroeconomic shocks than others? Hypothesis tests based on panel data from Peru. Journal of Development Economics, 56(1), 181–206. Goldin, C. (1994). The U-shaped female labor force function in economic development and economic history: Working paper 4707. National Bureau of Economic Research. Hardeweg, B., Klasen, S., & Waibel, H. (2013). Establishing a database for vulnerability research. In S. Klasen, & H. Waibel (Eds.), Vulnerability to poverty. London: Palgrave. Horrace, W. C., & Oaxaca, R. L. (2001). Inter-industry wage differentials and the gender wage gap: An identification problem. Industrial and Labour Relations Review, 54(3), 611–618. Jacoby, H. G., & Skoufias, E. (1997). Risk, financial markets, and human capital in a developing country. Review of Economic Studies, 64(3), 311–335. Kamanou, G., & Morduch, J. (2004). Measuring vulnerability to poverty. In S. Dercon (Ed.), Insurance against poverty. New York: Oxford University Press. King, E., Klasen, S., & Porter, S. (2009). Women and development. In B. Lomborg (Ed.), Global crises, global solutions (2nd ed.). Cambridge University Press. Klasen, S. (2006). Pro-poor growth and gender inequality. In L. Menkhoff (Ed.), Pro-poor growth: Policy and evidence. Berlin: Duncker&Humblot. Klasen, S. (2007). Gender-related indicators of well-being. In M. McGillivray (Ed.), Human well-being: Concept and measurement (pp. 167–192). London: Palgrave. Klasen, S., & Gaddis, I. (2013). Structural change, economic development, and women’s labor force participation. Journal of Population Economics. http://dx.doi.org/10.1007/s00148-013-0488-2.

Klasen, S., Lechtenfeld, T., & Povel, F. (2013). Using household surveys to capture vulnerability. In S. Klasen, & H. Waibel (Eds.), Vulnerability to poverty. London: Palgrave. Klasen, S., & Povel, F. (2013). Defining and measuring vulnerability: State of the art and new proposals. In S. Klasen, & H. Waibel (Eds.), Vulnerability to poverty. London: Palgrave. Klasen, S., & Waibel, H. (2013). Vulnerability to poverty. London: Palgrave. Kochar, A. (1995). Explaining household vulnerability to idiosyncratic income shocks. The American Economic Review, 85(2), 159–164. Lambert, S. (1994). La migration comme instrument de diversification des risquesdans la familleivoirienne. Revue d’economie du developpement, 2(1), 30–38. Lanjouw, P., & Ravallion, M. (1995). Poverty and household size. Economic Journal, 105(433), 1415–1434. Ligon, E., & Schechter, S. (2003). Measuring vulnerability. Economic Journal, 113(486), 95–102. Lipton, M., & Ravallion, M. (1995). Poverty and policy. In J. Behrman, & T. N. Srinivasan (Eds.). Handbook of development economics (Vol. 3). Amsterdam: North Holland. Marcoux, A. (1998). The feminization of poverty: Claims, facts, and data needs. Population and Development Review, 24(1), 131–139. Moghadam, V. M. (2005). Globalizing women: Transnational feminist networks. Baltimore: Johns Hopkins University Press. Morduch, J. (2004). Consumption smoothing across space. In S. Dercon (Ed.), Insurance against poverty. Oxford University Press. Nguyen, B., Albrecht, J. W., Vroman, S. B., & Westbrook, M. D. (2007). A quantile regression decomposition of urban–rural inequality in Vietnam. Journal of Development Economics, 83(2), 466–490. Oostendorp, R. (2010). Globalization and the gender wage gap. World Bank Economic Review, 23(1). Pritchett, L., Suryahadi, A., & Sumarto, S. (2000). Quantifying vulnerability to poverty – A proposed measure, with application to Indonesia: Development economics working papers 83. East Asian Bureau of Economic Research. Quisumbing, A. R., Haddad, L., & Pena, C. (2001). Are women overrepresented among the poor? An analysis of poverty in 10 developing countries. Journal of Development Economics, 66(1), 225–269. Quisumbing, A. R., & Pandolfelli, L. (2010). Promising approaches to address the needs of poor female farmers: Resources. Constraints, and Interventions, World Development, 38(4), 581–592. Seguino, S. (2000). Gender inequality and economic growth: A crosscountry analysis. World Development, 28, 1211–1230. Storey, D. J. (2004). Racial and gender discrimination in the micro firms credit market? Evidence from Trinidad and Tobago. Small Business Economics, 23(5), 401–422. Townsend, R. (1994). Risk and insurance in village India. Econometrica, 62(3), 539–591. Udry, C. (1995). Risk and saving in Northern Nigeria. American Economic Review, 85(5), 1287–1300. UNDP. (1995, 2010). Human development report. New York: Oxford University Press. Wagstaff, A., & Lindelow, M. (2010). Are health shocks different? Evidence from a multi-shock survey in Laos: Development research group, WPS 5335. Washington DC: The World Bank. WEF (2008). Global gender gap report 2008. Geneva: World Economic Forum. World Bank (2001). World bank policy research report 2001: Engendering development: through gender equality in rights, resources, and voice. New York: Oxford University Press. World Bank (2007). Global monitoring report: Millennium development goals, confronting the challenges of gender equality and fragile states. Washington, DC: World Bank. World Bank (2011). World development report 2012: Gender equality and development. Washington, DC: The World Bank.

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

14

Table 6. Headship and sample size 2008 (wave 2) Country

Female

Male

Thailand

420 19.8% 334 15.8%

1701 80.2% 1810 84.2%

Vietnam

Thailand Vietnam

Thailand Vietnam

De Jure

De Facto

342 81.4% 267 80.3%

78 18.6% 67 19.7%

Widow

Single

Absent Husband

289 84.5% 204 75.8%

53 15.5% 63 24.2%

78 100.0% 67 100.0%

Note: Percent by subgroup using survey weights.

Table 7. Prevalence of shock exposure by country, in percent Any shock

No shock

46.64 66.52

53.36 33.48

Thailand Vietnam Income shock

Health shock

Social shock

Thailand

34.07

16.69

7.52

Vietnam

52.34

23.94

6.33

Market Shock

Agricultural Supply Shock

Thailand

12.73

27.96

Vietnam

2.59

51.00

Credit Price Job/business Remittance Livestock Crop Storm/rain/ Drought Birth Illness Accident Death Social Migrated Crime/law/ House problem shock loss drop disease pest cold obligation hh member jail damage Thailand Vietnam

2.58 0.23

8.02 1.73

2.46 0.65

0.68 0.05

0.80 10.62

6.81 10.46

9.55 35.30

Notes: Survey weights used; mean prevalence of shock exposure during 2007 and 2008.

16.75 6.96

0.80 1.99

11.92 18.91

2.43 2.29

2.58 2.08

2.91 2.33

0.51 1.23

3.00 1.97

1.59 1.16

WORLD DEVELOPMENT

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

APPENDIX

Variable

Unit

Male-headed

Female-headed

De Jure FHH

De Facto FHH

Widow

Single

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

ln (cons per capita) ln (cons per adult)

USD PPP USD PPP

7.085 7.550

6.773 7.270

7.109 7.597

6.771 7.232

7.058 7.542

6.726 7.173

7.310 7.813

6.968 7.485

7.017 7.512

6.734 7.180

7.260 7.688

6.698 7.151

HH composition HH size HH size (adult equiv) Dependency Ratio Children aged upto 1 Children aged upto 2 Children aged upto 3 Children aged upto 4 Children aged upto 5

Members Members Ratio Members Members Members Members Members

4.068 2.470 1.558 0.089 0.161 0.223 0.283 0.353

4.516 2.676 1.666 0.089 0.165 0.255 0.335 0.414

3.554 2.095 1.724 0.093 0.164 0.226 0.281 0.335

3.098 1.878 1.657 0.057 0.112 0.156 0.220 0.271

3.643 2.144 1.614 0.089 0.142 0.195 0.248 0.295

3.036 1.856 1.479 0.050 0.084 0.123 0.184 0.231

3.206 1.906 2.152 0.108 0.250 0.347 0.412 0.488

3.370 1.975 2.435 0.091 0.232 0.303 0.374 0.446

3.805 2.219 1.638 0.101 0.158 0.221 0.278 0.325

3.092 1.888 1.445 0.054 0.095 0.139 0.194 0.249

2.852 1.775 1.497 0.033 0.066 0.066 0.098 0.148

2.858 1.756 1.588 0.037 0.051 0.071 0.153 0.171

Education No education of HH head Primary education Middle school edu Secondary education Tertiary education HH headcanread Any schooling

% % % % % % %

8.1% 80.0% 5.1% 4.2% 2.7% 92.9% 95.5%

10.4% 22.7% 45.2% 16.7% 4.9% 91.1% 91.2%

21.9% 72.1% 3.1% 1.3% 1.6% 79.4% 87.1%

30.7% 24.5% 35.4% 7.3% 2.1% 73.8% 71.4%

25.0% 70.8% 1.7% 1.4% 1.1% 76.1% 85.0%

37.0% 25.2% 29.6% 6.5% 1.7% 68.6% 65.6%

9.8% 77.2% 8.7% 1.1% 3.3% 92.4% 95.6%

3.4% 21.1% 60.7% 11.0% 3.8% 96.6% 96.8%

26.5% 70.8% 1.7% 0.7% 0.3% 74.5% 82.9%

38.2% 26.5% 27.1% 6.0% 2.3% 66.9% 65.0%

18.0% 70.6% 1.6% 4.9% 4.9% 83.7% 95.1%

33.1% 21.3% 37.5% 8.2% 0.0% 74.2% 67.3%

Age of HH head Age of HH head, squared

Years Yearssq

53 2999

47 2393

59 3718

54 3146

64 4219

58 3514

41 1765

38 1538

66 4479

61 3842

53 2946

48 2472

Income and remittances Ln (land size) Noof income sources Remittance received Remittance sent Remittance net recipient Net remittances per capita

Ln (hectar) Amount % % % USD PPP

0.570 3.71 9.4% 2.8% 9.2% 28.09

0.842 3.16 4.5% 6.8% 4.2% 11.17

0.035 3.17 8.4% 2.0% 8.4% 18.98

1.588 2.66 4.7% 3.6% 4.3% 19.07

0.103 3.24 10.0% 1.9% 10.0% 22.43

1.647 2.70 5.4% 4.3% 4.9% 11.50

0.229 2.89 2.2% 2.2% 2.2% 5.51

1.328 2.47 1.8% 0.2% 1.8% 52.19

0.206 3.29 10.7% 2.3% 10.7% 19.88

1.551 2.80 6.4% 4.0% 5.8% 17.21

0.403 2.98 6.5% 0.0% 6.5% 34.93

1.953 2.38 2.0% 5.2% 2.0% 6.61

% % % % %

95.5% 85.4% 78.1% 40.4% 71.2%

94.8% 90.2% 77.8% 40.9% 29.0%

91.8% 72.2% 63.4% 30.8% 68.5%

86.8% 79.5% 68.9% 35.1% 17.5%

91.6% 71.3% 64.6% 30.9% 69.1%

86.5% 79.4% 66.9% 36.7% 18.1%

92.4% 76.0% 58.7% 30.4% 66.2%

88.0% 79.9% 77.4% 28.4% 14.7%

92.6% 73.1% 67.1% 31.5% 69.8%

87.3% 82.2% 72.7% 39.4% 16.4%

86.9% 62.3% 52.5% 27.9% 65.6%

84.0% 70.6% 48.7% 28.0% 23.3%

% % %

72.7% 56.3% 30.0%

64.0% 49.4% 23.9%

62.6% 49.5% 24.0%

53.9% 40.9% 19.5%

66.1% 53.8% 24.2%

57.5% 43.8% 20.1%

48.9% 32.6% 22.8%

38.3% 28.3% 16.9%

65.8% 53.1% 23.9%

55.9% 43.2% 20.0%

67.2% 57.3% 26.2%

62.7% 45.6% 20.3%

Shocks Income shock Market shock Supply shock Health shock Social shock

% % % % %

22.3% 6.2% 17.7% 9.4% 3.8%

39.7% 3.3% 37.8% 23.1% 3.9%

19.1% 5.5% 15.1% 11.8% 5.3%

35.3% 1.7% 34.0% 27.4% 4.1%

19.8% 5.0% 15.6% 13.4% 5.0%

37.3% 0.8% 36.9% 27.9% 3.8%

16.3% 7.6% 13.0% 5.4% 6.5%

26.6% 5.4% 21.3% 25.2% 5.3%

20.2% 3.7% 17.2% 13.1% 4.7%

38.6% 0.6% 38.0% 29.5% 2.7%

18.0% 11.5% 8.2% 14.7% 6.5%

33.3% 1.7% 33.3% 22.7% 7.5%

Households

N

1724

1867

451

323

359

265

92

58

298

202

61

63

Income sector Busy in agriculture Crop sector Livestock sector Livestock products Fishing Non-farm sector Offfarm employment Self employment

Note: Values are population weighted.

A FEMINIZATION OF VULNERABILITY? FEMALE HEADSHIP, POVERTY, AND VULNERABILITY IN THAILAND AND VIETNAM15

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

Table 8. Summary statistics—2007

Unit

Male-headed

Female-headed

De Jure FHH

16

De Facto FHH

Widow

Single

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

ln (cons per capita) ln (cons per adult)

USD PPP USD PPP

7.278 7.748

6.983 7.477

7.260 7.760

6.988 7.449

7.209 7.709

6.946 7.397

7.483 7.984

7.157 7.659

7.191 7.700

6.968 7.417

7.306 7.759

6.878 7.333

HH composition HH Size HH Size (adult equiv) Dependency ratio Children aged upto 1 Children aged upto 2 Children aged upto 3 Children aged upto 4 Children aged upto 5

Members Members Ratio Members Members Members Members Members

4.076 2.468 1.554 0.075 0.129 0.203 0.276 0.340

4.538 2.697 1.618 0.079 0.136 0.211 0.297 0.376

3.685 2.150 1.754 0.081 0.147 0.219 0.288 0.359

3.101 1.874 1.603 0.076 0.086 0.133 0.173 0.240

3.777 2.197 1.693 0.076 0.146 0.213 0.269 0.336

3.020 1.836 1.429 0.074 0.083 0.117 0.157 0.219

3.283 1.941 2.021 0.102 0.153 0.243 0.370 0.461

3.429 2.029 2.310 0.081 0.098 0.197 0.237 0.325

3.913 2.257 1.730 0.083 0.155 0.232 0.294 0.367

3.039 1.851 1.348 0.065 0.076 0.106 0.157 0.218

3.038 1.870 1.493 0.038 0.095 0.114 0.133 0.170

2.962 1.791 1.683 0.102 0.105 0.154 0.156 0.222

Education No education of HH head Primary education Middle school edu Secondary education Tertiary education HH head can read Edu_school

% % % % % % %

8.0% 79.4% 5.2% 4.8% 2.5% 93.1% 95.4%

12.3% 22.8% 43.5% 15.7% 5.7% 89.3% 89.2%

19.9% 74.6% 2.4% 1.9% 1.2% 80.8% 88.1%

27.3% 26.7% 35.6% 7.6% 2.7% 74.4% 73.1%

22.7% 73.2% 1.5% 1.8% 0.9% 77.8% 86.0%

31.6% 28.1% 31.3% 6.5% 2.5% 70.6% 68.5%

7.7% 80.8% 6.4% 2.6% 2.6% 93.6% 97.4%

9.9% 20.9% 53.4% 12.3% 3.5% 90.1% 91.8%

23.5% 74.1% 1.0% 1.0% 0.3% 76.9% 84.8%

35.0% 30.2% 26.1% 5.9% 2.8% 67.3% 65.1%

18.8% 68.0% 3.8% 5.7% 3.7% 83.1% 92.5%

21.1% 21.5% 47.5% 8.2% 1.7% 80.7% 79.2%

Age of HH head Age of HH head, squared

Years Yearssq

54 3092

48 2507

60 3850

55 3243

64 4286

59 3658

43 1939

39 1551

66 4514

62 4035

54 3043

48 2476

Income and remittances Ln (land size) No of income sources

Ln (hectar) Amount

0.669 3.79

0.749 3.89

0.220 3.44

1.556 3.49

0.212 3.49

1.578 3.55

0.256 3.22

1.464 3.26

0.264 3.51

1.439 3.61

0.069 3.41

2.013 3.37

Remittance received Remittance sent Remittance net recipient Net remittances per capita

% % % USD PPP

9.6% 4.0% 8.7% 0.70

7.8% 13.2% 5.4% 6.30

9.5% 4.8% 9.3% 0.06

9.7% 7.1% 7.0% 0.87

10.2% 5.0% 9.9% 0.05

10.1% 6.3% 6.7% 1.84

6.4% 3.8% 6.4% 0.54

8.0% 10.3% 8.0% 3.06

11.0% 5.2% 10.7% 0.00

10.0% 6.6% 6.6% 0.42

5.7% 3.7% 5.7% 0.31

10.4% 5.4% 7.0% 6.29

% % % % %

97.5% 87.4% 84.6% 31.3% 80.6%

97.3% 92.0% 85.9% 55.9% 63.7%

93.8% 78.5% 78.6% 24.6% 78.1%

93.2% 82.8% 82.8% 48.8% 61.3%

93.8% 77.1% 79.8% 24.6% 78.4%

93.1% 83.5% 82.8% 48.5% 62.6%

93.6% 84.6% 73.1% 24.4% 76.9%

93.6% 79.8% 82.6% 49.9% 56.2%

94.1% 77.1% 80.3% 25.3% 79.6%

94.6% 86.7% 86.5% 49.9% 61.8%

92.4% 77.3% 77.3% 20.7% 71.7%

88.5% 73.5% 71.1% 44.1% 65.1%

% % %

71.7% 56.5% 28.8%

70.8% 57.5% 26.3%

63.6% 51.7% 23.1%

57.4% 39.7% 24.1%

66.7% 54.1% 24.9%

59.7% 44.5% 23.0%

50.1% 41.1% 15.4%

47.7% 20.5% 28.7%

65.1% 54.0% 23.2%

57.9% 43.5% 22.4%

75.4% 54.7% 34.0%

65.6% 47.4% 24.9%

Shocks Income shock Market shock Supply shock Health shock Social shock

% % % % %

47.8% 20.1% 39.9% 22.9% 10.7%

61.5% 2.6% 60.3% 24.3% 8.6%

41.4% 16.4% 34.3% 26.2% 11.9%

55.2% 0.7% 55.2% 26.6% 8.8%

42.4% 16.7% 34.8% 27.2% 11.4%

54.7% 0.8% 54.7% 28.2% 8.5%

37.2% 15.4% 32.0% 21.9% 14.1%

57.4% 0.0% 57.4% 20.1% 10.0%

42.9% 17.3% 35.6% 27.7% 11.0%

56.1% 1.1% 56.1% 29.2% 9.5%

39.6% 13.2% 30.1% 24.6% 13.2%

50.4% 0.0% 50.4% 25.3% 5.4%

Households

N

1701

1810

420

334

342

267

78

67

289

204

53

63

Income sector Busy in agriculture Crop sector Livestock sector Livestock products Fishing Non-farm Sector Offfarm employment Self employment

Note: Values are population weighted.

WORLD DEVELOPMENT

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

Table 9. Summary statistics—2008 Variable

A FEMINIZATION OF VULNERABILITY? FEMALE HEADSHIP, POVERTY, AND VULNERABILITY IN THAILAND AND VIETNAM17 Table 10. Correlates of consumption OLS: ln (consumption)

(1)

(2)

FHH subgroups

FHH, absent husband FHH, widow FHH, single

(3)

(4)

Demography

(5)

(6)

HH head

(7)

(8) Wealth

Vietnam

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

0.129** (0.0621) 0.087** (0.0409) 0.131 (0.0830)

0.074 (0.0644) 0.048 (0.0341) 0.217*** (0.0802)

0.147** (0.0662) 0.093** (0.0402) 0.090 (0.0836) 0.067*** (0.0179) 0.085*** (0.0176)

0.101 (0.0616) 0.118*** (0.0365) 0.280*** (0.0812) 0.068*** (0.0156) 0.080*** (0.0167)

0.121* (0.0671) 0.024 (0.0419) 0.120 (0.0789)

0.105 (0.0655) 0.035 (0.0361) 0.135* (0.0801)

0.110* (0.0602) 0.046 (0.0381) 0.177** (0.0741)

0.078 (0.0650) 0.030 (0.0352) 0.184** (0.0816)

0.119* (0.0617) 0.094** (0.0418) 0.108 (0.0867)

0.070 (0.0645) 0.057* (0.0340) 0.242*** (0.0810)

0.152*** (0.0459) 0.360*** (0.0703) 0.538*** (0.0851) 0.896*** (0.1072) 0.018** (0.0083) 0.000** (0.0001)

0.106*** (0.0368) 0.250*** (0.0360) 0.379*** (0.0508) 0.635*** (0.0536) 0.017*** (0.0058) 0.000** (0.0001) 0.179*** (0.0136) 0.073*** (0.0105)

0.158*** (0.0271) 0.028** (0.0119) 0.051* (0.0272) 0.121** (0.0489) 0.041 (0.0375) 7.698*** (0.0428) 2130 0.241 4.087 0.009

Dep. ratio Head: Primary Edu Head: Middle Edu Head: Secondary Edu Head: Tertiary Edu Age of head Age square Housing index ln (land)

7.762*** (0.0065)

7.612*** (0.0051)

8.016*** (0.0439)

7.913*** (0.0439)

7.220*** (0.2229)

6.927*** (0.1410)

7.655*** (0.0121)

7.620*** (0.0221)

0.061** (0.0258) 0.036 (0.0414) 0.136*** (0.0353) 7.774*** (0.0310)

2113 0.072 5.104 0.002

2130 0.235 3.521 0.018

2113 0.094 5.264 0.002

2130 0.261 7.897 0.000

2113 0.137 1.782 0.155

2130 0.299 2.370 0.075

2113 0.172 4.576 0.005

2130 0.262 2.304 0.081

2113 0.080 4.672 0.004

Non-farm sector Crops sector Livestock sector

Observations Adjusted R-squared Headship joint signif. Headship prob > F

(10)

Thailand

HH size (adult equiv)

Constant

(9)

Economic activity

Notes: Survey weights used; robust standard errors in parentheses; outcome: log of consumption in 2008 in USD PPP per day per adult equivalent (World Bank scale) with economies of scale (0.8); all covariates are from 2007; income sector dummies are not exclusive (no reference group); reference group for educational attainment: no education; village dummies included but not reported. Significance levels: * p < 0.1. ** p < 0.05. *** p < 0.01.

Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003

18

WORLD DEVELOPMENT Table 11. Correlates of shock exposure (1) Probit: any shock

FHH, absent husband FHH, widow FHH, single

(2)

FHH subgroups

(3)

(4)

Demography

(5)

(6)

(7)

HH head

(8) Wealth

(9)

(10)

Economic activity

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

Thailand

Vietnam

0.031 (0.0550) 0.017 (0.0316) 0.024 (0.0659)

0.018 (0.0631) 0.019 (0.0336) 0.058 (0.0550)

0.002 (0.0568) 0.003 (0.0320) 0.012 (0.0667) 0.053*** (0.0147) 0.009 (0.0150)

0.009 (0.0644) 0.011 (0.0348) 0.026 (0.0560) 0.035*** (0.0126) 0.016 (0.0132)

0.027 (0.0568) 0.017 (0.0339) 0.016 (0.0665)

0.021 (0.0640) 0.033 (0.0361) 0.078 (0.0555)

0.005 (0.0557) 0.008 (0.0318) 0.014 (0.0670)

0.004 (0.0647) 0.014 (0.0340) 0.008 (0.0563)

0.014 (0.0556) 0.001 (0.0321) 0.012 (0.0666)

0.020 (0.0662) 0.004 (0.0337) 0.013 (0.0566)

0.021 (0.0366) 0.031 (0.0636) 0.072 (0.0681) 0.244*** (0.0791) 0.008 (0.0059) 0.000 (0.0001)

0.010 (0.0319) 0.052* (0.0314) 0.061 (0.0379) 0.190*** (0.0511) 0.004 (0.0046) 0.000 (0.0000) 0.020* (0.0112) 0.038*** (0.0078)

0.054*** (0.0171) 0.062*** (0.0078) 0.015 (0.0242) 0.130*** (0.0305) 0.014 (0.0266)

0.025 (0.0214) 0.253*** (0.0324) 0.040* (0.0244)

HH size (adult equiv) Dep. ratio Head: Primary Edu Head: Middle Edu Head: Secondary Edu Head: Tertiary Edu Age of head Age square Housing index ln (land) Non-farm sector Crops sector Livestock sector Constant

Observations Pseudo R2 Model Wald Chi2 Model p-val FHHjoint Chi2 FHH p-val

2116 0.026

2123 0.159

2116 0.031

2122 0.163

2116 0.033

2122 0.165

2116 0.036

2123 0.188

2116 0.034

2123 0.194

74.102 0.007 0.656 0.884 0.031 (0.0550)

393.584 0.000 1.417 0.702 0.018 (0.0631)

87.480 0.001 0.052 0.997 0.002 (0.0568)

405.205 0.000 0.363 0.948 0.009 (0.0644)

93.598 0.000 0.495 0.920 0.027 (0.0568)

409.734 0.000 2.668 0.446 0.021 (0.0640)

100.510 0.000 0.123 0.989 0.005 (0.0557)

466.643 0.000 0.211 0.976 0.004 (0.0647)

95.013 0.000 0.102 0.992 0.014 (0.0556)

480.808 0.000 0.163 0.983 0.020 (0.0662)

Notes: Survey weights used; robust standard errors in parentheses; marginal effects reported; outcome: dummy equal to 1 if household experienced any shock during 2007–2008; all covariates are from 2007; income sector dummies are not exclusive (no reference group); reference group for educational attainment: no education; constant and district dummies included but not reported. Significance levels: * p < 0.1. ** p < 0.05. *** p < 0.01.

Available online at www.sciencedirect.com

ScienceDirect Please cite this article in press as: Klasen, S. et al. A Feminization of Vulnerability? Female Headship, Poverty, and Vulnerability in Thailand and Vietnam, World Development (2014), http://dx.doi.org/10.1016/j.worlddev.2013.11.003