Assets, Wealth and Spousal Violence

advertisement
Assets, Wealth and Spousal Violence:
Insights from Ecuador and Ghana
Abena D. Oduro, University of Ghana
Carmen Diana Deere, University of Florida
Zachary Catanzarite, University of Florida
Prepared for the
World Bank Workshop on Gender and Assets
June 14 2012
Introduction
• Several studies investigate factors that might
increase women’s bargaining power and reduce
the risk of abuse.
• Very few have considered the relationship
between women’s asset (i.e. land and
home)ownership and spousal violence
– Women’s homeownership deters physical and
psychological abuse (Panda and Agarwal 2005,
Bhattacharrya et al 2011)
– Evidence on association of spousal abuse and
women’s land ownership is mixed (Bhattacharyya et al
2011, Ezeh and Gage 2000, Panda and Agarwal, 2005,)
Introduction (contd.)
• This study adds to the growing literature on spousal
abuse in two ways:
– It considers ownership of a wider range of assets, i.e.
agricultural land, home ownership and ownership of other
real estate such as another residence, commercial building
and non-agricultural plot.
– It investigates women’s ownership of assets relative to
their partners.
• Places emphasis on relative value of women’s assets as a measure
of their fall back position
• Controls for the fact that different assets impact bargaining power
differently.
• Allow us to determine whether the preventive impact of women’s
share of wealth varies along the wealth distribution
Context
Ecuador
• Population: 14.7 million
• HDI rank: 83
• Law Against Domestic
Violence Towards
Women and the Family
(1995)
Ghana
• Population 25 million
• HDI rank: 135
• Domestic Violence Act
(2007)
The Data
•
•
•
•
Ecuador
EAFF-Ecuador Household
Asset Survey conducted in
2010
2,892 Households
Two-stage sampling
procedure
Sample size for this study:
1,938 couples –married
or in a consensual union
•
•
•
•
Ghana
GHAS-Ghana Household
Asset Survey conducted in
2010
2,170 Households
Two stage sampling
procedure
Sample size for this study:
886 couples – married or
in a consensual union
Survey Instrument
• Designed to be similar in several respects
• Domestic Violence Module- Respondents
were asked:
– How common domestic violence was in their
community or neighbourhood?
– Whether they had been abused physically,
verbally or psychologically
– Who the perpetrator of the abuse was
Incidence of Spousal Violence During Previous
12 months (Currently partnered women aged
18-49 years)
Type of Abuse
Ecuador
Ghana
N= 1,938
N = 886
Physical
3.3%
2.1%
Emotional
17.7%
11.2%
Any form of
abuse
18.1%
12.0%
Notes: *Categories are not mutually exclusive. The percentages are
weighted by the sample expansion factors.
Sources: EAFF (2010); GHAS (2010)
The Models
• The Dependent variables- Woman’s report of:
– Physical violence in past 12 months
– Emotional violence, i.e. verbal and psychological abuse, in
past 12 months
• Variable of Interest- Women’s asset ownership
measured as:
– Women’s ownership of any of the following real estate:
agricultural land, place of residence, other real estate .
Categorical variable that takes a value of 1 if owner, 0 if
not
– Women’s share of couple’s gross value of physical and
financial assets- continuous variable ranging from 0 to 1.
Other Explanatory Variables
• Characteristics of Woman
– Age, education and number of children aged 13 years and younger
• Characteristics of the Couple
– Age difference, difference in years of education, employment status
relative to spouse, relative spousal earnings
• Nature of the Relationship
– Type of union (i.e. married or in a consensual union), occurrence of
financial disagreements in past 12 months
• Household Context
– Socioeconomic status of household- gross value of assets, crowding,
location
• Community Context
– Woman’s perception of the frequency of domestic violence in the
community
Methodology
• Logistic regression
– Physical abuse
– Emotional Abuse
• Baseline model:
– Includes all explanatory variables except variable of interest.
• Model I:
– Includes woman’s ownership of asset variable in the baseline
• Model II:
– Includes woman’s share of couple wealth in the baseline
• Model III:
– Includes woman’s share of couple wealth and interaction of
woman’s share of couple wealth and household wealth tertiles
in the baseline.
Descriptives
Ecuador
Ghana
N=1,938
N=886
Woman a Major Asset Owner (Percent)
54.5
21.8
Female share of Couple Wealth (Mean,
percent)
46.8
23.2
Woman’s Age (Years)
41.27
39.24
Spousal Age difference (Years)
4.09
7.95
Woman’s Years of Schooling
8.17
4.51
Spousal Schooling Difference (Years)
0.38
1.76
Consensual Union (Percent)
35.4
13.3
75.5
Monogamous Union (Percent)
Financial Disagreements (Percent)
15.1
13.3
Both Employed
58.2
85.6
Sources: EAFF (2010); GHAS (2010)
Logistic Regression Results for Physical Violence
Ecuador (N=1938)
Ghana (N=886)
Model
Variables
Coefficient
Standard Error
Coefficient
Standard Error
I
Woman Owns Real Estate
-0.177
0.295
-0.64
0.847
Likelihood Ratio Chi-Squared (df)
52.971 (18)
27.17(16)
Pseudo-R squared
II
0.200
Share of Couple Wealth
-2.766**
1.397
-3.91
4.282
Share of Couple Wealth Squared
2.210
1.415
5.63
5.26
Likelihood Ratio Chi-Squared (df)
57.096 (19)
28.13(17)
Pseudo-R squared
III
0.2075
Share of Couple Wealth
-2.293***
0.932
-7.498
6.692
Share of Wealth*Tertile 2
1.793
1.1868
5.288
7.590
Share of Wealth*Tertile 3
2.957**
1.354
10.982
6.912
Likelihood Ratio Chi-Squared (df)
59.775(20)
Pseudo-R squared
33.07(18)
0.243
The Odds of Physical Violence and Women’s Share of
Couple Wealth by Tertile, Ecuador and Ghana
Other Significant Explanatory Variables
Ecuador
• Financial Disagreements
(+ve)
• Report of Community
Violence(+ve)
• Employment: Man only
is employed
• Ghana
• Financial Disagreements
(+ve)
• Age of Woman (-ve)
• Years of education of
woman (-ve)
Logistic Regression Results for Emotional
Violence
Ecuador
Ghana
Model
Variables
Coefficient
Standard Error
Coefficient
Standard Error
I
Woman Owns Real Estate
-0.140
0.1753
-0.687*
0.379
Likelihood Ratio Chi-Squared (df)
137.939 (18)
104.98 (18)
Pseudo-R squared
II
0.197
Share of Couple Wealth
-0.451
0.899
Share of Couple Wealth Squared
1.051
0.846
Likelihood Ratio Chi-Squared (df)
143.269 (19)
0.668
105.24(18)
Pseudo-R squared
III
-1.261**
0.197
Share of Couple Wealth
1.200**
0.521
0.863
1.306
Share of Wealth*Tertile 2
-0.580
0.692
-4.570**
1.913
Share of Wealth*Tertile 3
-1.224*
0.745
-1.509
1.604
Likelihood Ratio Chi-Squared (df)
144.437 (20)
Pseudo-R squared
111.41 (19)
0.209
The Odds of Emotional Violence and Women’s Share of
Couple Wealth by Tertile, Ecuador and Ghana
Other Significant Explanatory Variables
•
•
•
•
Ecuador
Financial Disagreements
(+ve)
Perceptions of
community violence
(+ve)
Urban location (+ve)
Earnings: Woman earns
more than partner (+ve)
•
•
•
•
Ghana
Financial Disagreements
(+ve)
Perceptions of
community violence
(+ve)
Urban location (-ve)
Polygamous union (-ve)
Conclusion
• Asset variables behave differently across
models and between the two countries.
– Being an asset owner has a significant and
negative effect in Ghana
– In Ecuador woman’s share of couple wealth has a
significant negative effect on physical abuse.
– In Ghana woman’s share of couple wealth has a
significant deterrent effect for emotional abuse
only.
• Context Matters.
Conclusion contd.
• The deterrent effect of women’s share of wealth
depends on the socioeconomic status of the
household. Women in different socio-economic
strata face different risks.
• Ecuador:
– Woman in lowest third of household wealth with zero
share of couple wealth is predicted to be at risk from
physical abuse but is buffered from emotional abuse.
– However, when she increases her share of couple
wealth predicted likelihood of physical abuse declines
whilst likelihood of emotional abuse rises.
Conclusion contd.
• Predictors of both types of abuse:
– Both countries:
• Financial disagreements
• Perception of community violence
• Deterrents:
– Ecuador:
• Only male is employed, reduces likelihood of physical abuse
• Man’s years of schooling exceeds that of partner reduces
likelihood of emotional abuse
– Ghana:
• Age, Years of schooling of woman reduces physical violence
• Polygamous marriage reduces emotional violence
Conclusion
• Correlates of physical and emotional violence
are often different
• Common patterns across countries
• Context matters
• Impact of women’s share of couple’s wealth
on spousal violence is contingent on
household socioeconomic status.
Thank you for your attention
Download