What about the Women? Female Headship, Poverty and Vulnerability in Thailand and Vietnam Tobias Lechtenfeld with Stephan Klasen and Felix Povel 20-21 January 2011 OECD Conference, Paris
Thailand and Vietnam Study Area
Female-Headed Households Thailand Vietnam Population 63 Million 85 Million 25% of population in SE Asia Female Head 29.3% 17.4% in rural areas GDP Growth 7.4% 8.1% since mid-80s, annually Poverty 40% 74% headcount 1985 Reduction 12% 22% headcount 2008 Poverty reduction by 2/3 in only 2 decades How inclusive has growth been for typically poor and vulnerable groups: Female Headed Households, incl. Single Mothers, Widows?
Female-Headed Households 1. Theory Why could Female Headed Households be vulnerable? 2. Methods Poverty traps and Vulnerability 3. Data Households in Thailand and Vietnam 4. Results Shocks, Poverty traps and Income sources
Theory Why could Female Headed Households be most vulnerable? A. Differences between men and women (Gender-related Economic Gap) limited access to markets B. Differences between Male- and Female-Headed Households work burden lack of support
Theory A. Differences between men and women (Gender-related Economic Gap) access to land dominated by men, if female ownership, much smaller inheritance and land titling laws direct income effect (World Bank, 2007) formal credit markets (King et al., 2007) lack of collateral (Storey, 2004; Diagne et al., 2000) hard to start own business (King et al., 2007; Blackden Bhanu, 1999) insurance markets hardly functioning for men and women alike BUT: impact stronger for women lack of pension system lack of health insurance often access only through spouses (World Bank, 2001)
Theory A. Differences between men and women (cont) labour market different wages differential access to wage employment (Collier, 1994) less productive in employment girls receive less schooling (World Bank, 2001) less work experience (social stigma against labour) (King et al., 2007) less productive in farming adopt new production technology less likely (Chirwa, 2005; Asfaw and Admassie, 2004)
Theory B. Differences between Male- and Female-Headed Households double day burden handle domestic work and the role of main earner simultaneously (Moghadam, 1997). time and mobility constraints impacts negatively households income (Buvinic and Gupta, 1997). lack of support from social networks reduced family network to draw on (Bibars, 2001; Chant, 2008)
Empirical Evidence Poverty situation of female headed households Quisumbing et al.(2001): 10 country analysis, comparable methods Only in 2 of 10 countries FHH more poor Buvinic and Gupta (1997): review of 61 studies on FHH In 38 studies FHH more poor In 15 studies some FHH type more poor In 8 studies no evidence of increased poverty Situation largely country specific Lampietti and Stalker (2000), Ye (1998), and Haddad et al. (1996)
Data Sample Size: 4361 Households, 3 Panel Waves Country Female Male Thailand 451 1721 Vietnam 323 1866 De Jure De Facto Thailand 359 92 Vietnam 265 58 Widow Single Absent Husband Thailand 298 61 92 Vietnam 202 63 58
Analysis Six sets of analysis 1. Poverty 2. Exposure to shocks 3. Shock severity 4. Vulnerability to downside risk 5. Vulnerability to poverty 6. Consumption Smoothing
Analysis 1. Consumption Poverty Consumption Regression Outcome: Consumption per capita Adult equivalent scales: reflect different consumption needs Reduces bias in the poverty estimates Economies of Scale at household level Reduces bias in the poverty estimate
1. Consumption Poverty (2007) (1) (2) (3) (4) (5) (6) Shock GENDER FACTO/JURE TYPE Exposure Thailand Vietnam Thailand Vietnam Thailand Vietnam Female Head 0.0936*** -0.0289 (0.0297) (0.0350) De Facto FHH 0.289*** 0.183** (0.0482) (0.0697) De Jure FHH 0.0380-0.0854** (0.0335) (0.0372) FHH, absent husband 0.293*** 0.178** (0.0483) (0.0697) FHH, widow 0.0190-0.0459 (0.0348) (0.0377) FHH, single 0.122-0.210*** (0.0744) (0.0736) Observations 2,169 2,180 2,169 2,180 2,169 2,180 Adj. R-squared 0.215 0.386 0.221 0.391 0.221 0.393 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in parentheses, with village dummies Regressions include controls: dep. ratio, edu level, age, inc sources, land
1. Consumption Poverty (2008) (1) (2) (3) (4) (5) (6) Shock GENDER FACTO/JURE TYPE Exposure Thailand Vietnam Thailand Vietnam Thailand Vietnam Female Head 0.0777** -0.0350 (0.0345) (0.0323) De Facto FHH 0.188*** 0.203*** (0.0667) (0.0680) De Jure FHH 0.0495-0.101*** (0.0384) (0.0353) FHH, absent husband 0.187*** 0.196*** (0.0667) (0.0678) FHH, widow 0.0598-0.0504 (0.0429) (0.0390) FHH, single -0.00195-0.253*** (0.0826) (0.0638) Observations 2,121 2,144 2,121 2,144 2,121 2,144 Adj. R-squared 0.242 0.342 0.243 0.349 0.243 0.352 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in parentheses, with village dummies Regressions include controls: dep. ratio, edu level, age, inc sources, land
Analysis 2. Exposure to shocks Probit Regression - Outcome: Binary Shock Aggregate Income Shock Health Shock Social Shock Credit Problem Birth Social Obligation Price Shock Illness Migrated Hh Member Job / Business Loss Accident Crime / Law / Jail Remittance Drop Death House Damage Livestock Disease Crop Pest Storm / Rain / Cold Drought
Credit Problem Price Shock Job / Business Loss Remittance Drop Livestock Disease Crop Pest Storm / Rain / Cold Drought Birth Illness Accident Death Social Obligation Migrated Hh Member Crime / Law / Jail House Damage 2. Exposure to shocks: Shock Incidence (2007) Any Shock No Shock Thailand 32.32 67.68 Vietnam 59.95 40.05 Income Shock Health Shock Social Shock Thailand 21.61 9.84 4.14 Vietnam 43.29 23.24 3.97 Market Shock Agricultural Supply Shock Thailand 6.07 17.15 Vietnam 2.74 41.69 Thailand 2.9 1.75 1.66 0.28 0.32 2.34 5.43 9.93 0.23 6.39 0.32 3.08 1.33 0.41 1.24 1.38 Vietnam 0.37 1.83 0.55 0.05 8.4 9.54 22.97 8.17 1.83 18.31 2.01 1.96 0.73 1.14 1.64 0.5
Credit Problem Price Shock Job / Business Loss Remittance Drop Livestock Disease Crop Pest Storm / Rain / Cold Drought Birth Illness Accident Death Social Obligation Migrated Hh Member Crime / Law / Jail House Damage 2. Exposure to shocks: Shock Incidence (2008) Any Shock No Shock Thailand 60.96 39.04 Vietnam 73.09 26.91 Income Shock Health Shock Social Shock Thailand 46.53 23.53 10.89 Vietnam 61.38 24.63 8.68 Market Shock Agricultural Supply Shock Thailand 19.38 38.76 Vietnam 2.43 60.31 Thailand 2.26 14.29 3.25 1.08 1.27 11.27 13.67 23.57 1.37 17.44 4.53 2.07 4.48 0.61 4.76 1.79 Vietnam 0.09 1.63 0.75 0.05 12.83 11.38 47.62 5.74 2.15 19.5 2.57 2.19 3.92 1.31 2.29 1.82
2. Exposure to shocks (2007) (1) (2) (3) (4) (5) (6) Any Shock GENDER FACTO/JURE TYPE Exposure Thailand Vietnam Thailand Vietnam Thailand Vietnam Female Head 0.00881 0.0790** (0.0267) (0.0316) De Facto FHH -0.0532 0.0907 (0.0595) (0.0748) De Jure FHH 0.0253 0.0757** (0.0283) (0.0376) FHH, absent husband -0.0505 0.0907 (0.0597) (0.0746) FHH, widow 0.0109 0.0756* (0.0295) (0.0407) FHH, single 0.0893 0.0760 (0.0658) (0.0746) Observations 2,172 2,189 2,172 2,189 2,172 2,189 Wald Chi2 159.091 253.140 160.654 253.142 161.985 253.456 Prob > Chi2 0.000 0.000 0.000 0.000 0.000 0.000 Pseudo R2 0.058 0.066 0.059 0.066 0.059 0.066 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in parentheses, with district dummies Regressions include controls: dep. ratio, edu level, age, remittances, inc sources, land
2. Exposure to shocks (2008) (1) (2) (3) (4) (5) (6) Any Shock GENDER FACTO/JURE TYPE Exposure Thailand Vietnam Thailand Vietnam Thailand Vietnam Female Head 0.0279 0.00648 (0.0289) (0.0341) De Facto FHH -0.0199-0.0628 (0.0556) (0.0509) De Jure FHH 0.0403 0.0258 (0.0321) (0.0375) FHH, absent husband -0.0197-0.0629 (0.0557) (0.0507) FHH, widow 0.0378 0.0263 (0.0340) (0.0425) FHH, single 0.0521 0.0243 (0.0780) (0.0649) Observations 2,121 2,127 2,121 2,127 2,121 2,127 Wald Chi2 139.680 497.003 140.504 500.085 140.538 500.148 Prob > Chi2 0.000 0.000 0.000 0.000 0.000 0.000 Pseudo R2 0.049 0.210 0.050 0.211 0.050 0.211 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in parentheses, with district dummies Regressions include controls: dep. ratio, edu level, age, remittances, inc sources, land
Analysis 3. Shock Severity: Asset Loss Tobit Regression Outcome: Asset Loss from Shock, in ln(usd PPP) Intuition Model - Shocks can directly destroy assets (floods, fire, theft, etc) - Shock causes income loss and increased expenditure, making it necessary to sell assets
3. Shock Severity: Tobit, Asset Loss (2008) (1) (2) (3) (4) (5) (6) Downward GENDER FACTO/JURE TYPE Risk Thailand Vietnam Thailand Vietnam Thailand Vietnam Female Head -0.830 1.206* (1.234) (0.643) De Facto FHH 0.841 2.500** (2.346) (1.132) De Jure FHH -1.336 0.759 (1.464) (0.809) FHH, absent husband 0.833 2.513** (2.349) (1.137) FHH, widow -0.805 0.653 (1.564) (0.933) FHH, single -5.054 1.181 (4.390) (1.281) Observations 1,290 1,564 1,290 1,564 1,290 1,564 Pseudo R2 0.075 0.139 0.075 0.140 0.076 0.140 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in parentheses, with village dummies Regressions include controls: dep. ratio, edu level, age, remittances, inc sources, land
Analysis 4. Vulnerability to Downside Risk (Povel 2009) OLS Regression Outcome: Vulnerability measure based on future expectation Intuition Sum of expected values of all possible deprivations of future states of the world of household with probability of occurrence and a degree of risk aversion Vulnerability Index: Downside Risk
4. Vulnerability to Downside Risk (2008) (1) (2) (3) (4) (5) (6) Downward GENDER FACTO/JURE TYPE Risk Thailand Vietnam Thailand Vietnam Thailand Vietnam Female Head 0.00276-0.00125 (0.00285) (0.00149) De Facto FHH 0.00757-0.00494* (0.00580) (0.00294) De Jure FHH 0.00152-0.000225 (0.00308) (0.00191) FHH, absent husband 0.00744-0.00493* (0.00581) (0.00296) FHH, widow 0.00301-0.000315 (0.00351) (0.00215) FHH, single -0.00590 4.58e-05 (0.00435) (0.00292) Observations 2,121 2,144 2,121 2,144 2,121 2,144 Adj. R-squared 0.013 0.082 0.013 0.082 0.014 0.082 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in parentheses, with village dummies Regressions include controls: dep. ratio, edu level, age, remittances, inc sources, land
Analysis 5. Vulnerability to Poverty (Calvo & Dercon 2005) OLS Regression Outcome: Vulnerability measure based on past experiences Intuition Sum of expected values of all possible deprivations defined as, and censored at the poverty line with income, of future states of the world of household with probability of occurrence and a degree of risk aversion Vulnerability Index
5. Vulnerability to Poverty: Calvo Dercon (2007) (1) (2) (3) (4) (5) (6) Downward GENDER FACTO/JURE TYPE Risk Thailand Vietnam Thailand Vietnam Thailand Vietnam Female Head -0.00701* -0.000700 (0.00378) (0.00928) De Facto FHH -0.0139*** -0.0289*** (0.00465) (0.00944) De Jure FHH -0.00506 0.00680 (0.00429) (0.0108) FHH, absent husband -0.0140*** -0.0285*** (0.00466) (0.00947) FHH, widow -0.00444 0.00271 (0.00456) (0.0115) FHH, single -0.00785 0.0196 (0.00747) (0.0185) Observations 2,172 2,189 2,172 2,189 2,172 2,189 Adj. R-squared 0.049 0.151 0.049 0.153 0.049 0.153 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in parentheses, with village dummies Regressions include controls: dep. ratio, edu level, age, remittances, inc sources, land
5. Vulnerability to Poverty: Calvo Dercon (2008) (1) (2) (3) (4) (5) (6) Downward GENDER FACTO/JURE TYPE Risk Thailand Vietnam Thailand Vietnam Thailand Vietnam Female Head 0.00175 0.00417 (0.00313) (0.00498) De Facto FHH -0.00461-0.0207*** (0.00295) (0.00735) De Jure FHH 0.00338 0.0110* (0.00379) (0.00576) FHH, absent husband -0.00473-0.0206*** (0.00295) (0.00738) FHH, widow 0.00469 0.0106* (0.00422) (0.00632) FHH, single -0.00314 0.0125 (0.00763) (0.0108) Observations 2,121 2,144 2,121 2,144 2,121 2,144 Adj. R-squared 0.034 0.144 0.034 0.147 0.034 0.147 *** p<0.01, ** p<0.05, * p<0.1 Robust standard errors in parentheses, with village dummies Regressions include controls: dep. ratio, edu level, age, remittances, inc sources, land
Analysis 6. Consumption Smoothing (Townsend 1994) OLS Regression Outcome Changes in consumption over time Determinants - Income changes over time - Gender of headship Intuition Coefficients show the degree of uninsured exposure to risk
6. Consumption Smoothing (2007-2008) (1) (2) (3) (4) (5) (6) Female Head De Facto vs. De Jure FHH Subgroups OLS: Consumption Change Thailand Vietnam Thailand Vietnam Thailand Vietnam Income Change 0.00567*** 0.00117 0.00569*** 0.00117 0.00566*** 0.00118 Female Head -0.0501 0.0402 Female Head * Income Change 0.222*** 0.00312* De Facto FHH 0.107 0.143 De Jure FHH -0.0721 0.0154 De Facto FHH * Income Change 0.425** 0.000988 De Jure FHH * Income Change 0.204** 0.00320* FHH, absent husband 0.103 0.144 FHH, widow -0.116 0.0152 FHH, single 0.121-0.0189 FHH, absent husb * Inc Change 0.423** 0.00111 FHH, widow * Income Change 0.181* 0.00307* FHH, single * Income Change 0.344* 0.133 Observations 781 982 781 982 781 982 Adj R2 0.047 0.064 0.046 0.063 0.047 0.062
Summary of Results Widows, Singles (de jure female headed households) more exposed to shocks less able to insure consumption against income shocks Absent Husbands (de facto female headed households) better off through remittances asset losses from shocks more severe but: counterfactual spurious: unclear why do husbands leave? Perceived vs experienced risk some difference between past shocks and future risk apparently gender related differences in risk perception
Conclusions 1. Welfare Generally female heads are not poorer Gender Analysis only useful when broken down in subgroups 2. Policy Headship is a useful concept for targeting Widows / Singles in need of social policy 3. Further Research Panel Analysis using 3 waves Selection process of migration decision of husband Are families able to weather through macro crises? Transmission of poverty to children (education, health)
Thank you! tlechtenfeld@uni-goettingen.de Göttingen University, Germany Development Economics Research Group
Covariates: Thailand (2007) Variable Male Headed Female Headed Absent Husband Widow Single Unit Consumption 7.083 7.109 7.058 7.309 7.016 7.260 ln(usd PPP per adult) HH Size 4.068 3.554 3.643 3.207 3.805 2.852 members Dependency Ratio 1.558 1.724 1.614 2.153 1.638 1.497 ratio Children age<6 0.353 0.335 0.295 0.489 0.326 0.148 members Read 92.9% 79.4% 76.0% 92.4% 74.5% 83.6% % Schooling 95.5% 87.1% 85.0% 95.7% 82.9% 95.1% % Age 53 59 64 41 66 53 years Land Size 0.572 0.035 0.102-0.228 0.206-0.406 ln(hectar) Income Sources 3.71 3.17 3.24 2.89 3.29 2.98 number Net remittances 28.16 18.97 22.43 5.49 19.89 34.81 USD PPP per cap income shock 22.3% 19.1% 19.8% 16.3% 20.1% 18.0% % market shock 6.2% 5.5% 5.0% 7.6% 3.7% 11.5% % supply shock 17.7% 15.1% 15.6% 13.0% 17.1% 8.2% % health shock 9.3% 11.8% 13.4% 5.4% 13.1% 14.8% % social shock 3.8% 5.3% 5.0% 6.5% 4.7% 6.6% % Households 1724 451 359 92 298 61 N
Covariates: Thailand (2008) Variable Male Headed Female Headed Absent Husband Widow Single Unit Consumption 7.277 7.259 7.208 7.483 7.190 7.307 ln(usd PPP per adult) HH Size 4.075 3.686 3.778 3.282 3.913 3.038 members Dependency Ratio 1.554 1.754 1.693 2.022 1.730 1.493 ratio Children age<6 0.340 0.360 0.336 0.462 0.367 0.170 members Read 93.1% 80.7% 77.8% 93.6% 76.8% 83.0% % Schooling 95.4% 88.1% 86.0% 97.4% 84.8% 92.5% % Age 54 60 64 43 66 54 years Land Size 0.668 0.220 0.212 0.255 0.264-0.070 ln(hectar) Income Sources 3.79 3.44 3.49 3.22 3.51 3.42 number Net remittances 0.70 0.06-0.05 0.54 0.00-0.31 USD PPP per cap income shock 47.8% 41.4% 42.4% 37.2% 42.9% 39.6% % market shock 20.1% 16.4% 16.7% 15.4% 17.3% 13.2% % supply shock 39.9% 34.3% 34.8% 32.1% 35.6% 30.2% % health shock 22.9% 26.2% 27.2% 21.8% 27.7% 24.5% % social shock 10.6% 11.9% 11.4% 14.1% 11.1% 13.2% % Households 1701 420 342 78 289 53 N
Covariates: Vietnam (2007) Variable Male Headed Female Headed Absent Husband Widow Single Unit Consumption 6.740 6.744 6.699 6.948 6.712 6.658 ln(usd PPP per adult) HH Size 4.552 3.139 3.091 3.362 3.129 2.968 members Dependency Ratio 1.677 1.638 1.465 2.428 1.435 1.559 ratio Children age<6 0.430 0.260 0.223 0.431 0.233 0.190 members Read 89.6% 70.0% 64.5% 94.8% 62.9% 69.8% % Schooling 90.3% 68.7% 62.6% 96.6% 61.4% 66.7% % Age 47 54 57 38 60 48 years Land Size -0.805-1.543-1.582-1.366-1.486-1.887 ln(hectar) Income Sources 3.20 2.76 2.80 2.55 2.90 2.51 number Net remittances -13.16 25.77 23.20 37.48 31.75-4.21 USD PPP per cap income shock 44.0% 39.0% 41.1% 29.3% 43.1% 34.9% % market shock 2.9% 1.5% 0.8% 5.2% 0.5% 1.6% % supply shock 42.4% 37.8% 40.8% 24.1% 42.6% 34.9% % health shock 22.6% 26.9% 27.5% 24.1% 29.2% 22.2% % social shock 3.9% 4.3% 4.2% 5.2% 3.0% 7.9% % Households 1867 323 265 58 202 63 N
Covariates: Vietnam (2008) Variable Male Headed Female Headed Absent Husband Widow Single Unit Consumption 6.950 6.955 6.915 7.118 6.938 6.840 ln(usd PPP per adult) HH Size 4.559 3.174 3.105 3.448 3.137 3.000 members Dependency Ratio 1.628 1.594 1.412 2.322 1.334 1.663 ratio Children age<6 0.383 0.243 0.213 0.358 0.206 0.238 members Read 88.0% 70.4% 66.7% 85.1% 63.2% 77.8% % Schooling 88.3% 70.1% 65.5% 88.1% 61.8% 77.8% % Age 48 55 59 39 62 48 years Land Size -0.705-1.485-1.519-1.347-1.382-1.962 ln(hectar) Income Sources 3.87 3.47 3.53 3.24 3.59 3.32 number Net remittances -5.69 0.87 1.63-2.13 0.41 5.57 USD PPP per cap income shock 62.4% 56.0% 55.8% 56.7% 57.4% 50.8% % market shock 2.8% 0.6% 0.7% 0.0% 1.0% 0.0% % supply shock 61.1% 56.0% 55.8% 56.7% 57.4% 50.8% % health shock 24.1% 27.2% 29.6% 17.9% 30.4% 27.0% % social shock 8.6% 9.0% 9.0% 9.0% 9.8% 6.3% % Households 1810 334 267 67 204 63 N
Analysis Axioms of Vulnerability Measurement (i) (ii) (iii) (iv) (v) (vi) (vii) symmetry over states continuity and differentiability scale invariance normalization probability-dependent effect of outcomes probability transfer constant relative or absolute risk sensitivity In-depth discussion available in Calvo and Dercon (2005)