Do improved property rights decrease violence against ...dse.univr.it/ssdev/documents/material2013/slidesAmaralSofia.pdf · Introduction Background Reprtedo violence IPV Conclusion

Introduction Background Reported violence IPV Conclusion and Discussion

Do improved property rights decrease violence

against women in India?

So�a Amaral

University of Birmingham, UK

5th Summer School in Development Economics, Alba di Canazei

July, 2013

1/21

Introduction Background Reported violence IPV Conclusion and Discussion

Motivation

Women's rights and conditions in India[Du�o, 2011; Doepke et al, 2011]

Many arguments in favour of gender-speci�c targeting programs andtheir impact on empowerment but, does targeting transfers towomen improve their welfare?[Bobonis et al, (2012); Eswaran and Malhotra, (2011); Block andRao, (2000); Ferrari and Iyengar, (2012)]

Objectives:

Identify the causal e�ect of a legislative change in inheritance rightson reported crime rates and intimate partner violence

2/21

Introduction Background Reported violence IPV Conclusion and Discussion

Preview of results

Overall increase in household economic conditions decreases theequilibrium level of in�icted violence

Decrease in reported-crimes not robust to the inclusion ofstate-speci�c trends

Decrease in IPV for the younger cohorts who are wives of the HH

3/21

Introduction Background Reported violence IPV Conclusion and Discussion

Channels of TransmissionE�ects of a change in the wife's wealth:

1 Intrahousehold welfare

If the threat point is divorce this could lead to a decrease in violence[Tauchen et al, (1991); Farmer at al, (1997)]

If the threat point is not divorce the husband may use violence toregain control of household distribution of resources: ambiguos[Block and Rao, (2000); Bobonis et al, (2012); Eswaran andMalhotra, (2012); Chin, (2012)]

2 Marriage Market: could improve the quality of the match thoughthis is not necessarily true

3 Reporting behaviour: Tolerance of violence, incidence and reportingIyengar, (2009)

4/21

Introduction Background Reported violence IPV Conclusion and Discussion

Hindu Succession Act

HSA, (1956) is the major inheritance law in India.

Rules property law/inheritance laws for individuals of Hindu-type

religions

Applies to Hindus, Sikhs, Buddhists and Jains

Does not apply in Jammu and Kashmir (Arunachal Pradesh,Manipur, Nagaland, Mizoram and Meghalaya)

Joint Family Property: ancestral property

Separate Property: self- acquired property

Male coparcener: Hindu family male members

Daughters had rights to separate property but not ancestral property

HSA

5/21

Introduction Background Reported violence IPV Conclusion and Discussion

Hindu Succession Act

Amendments in 5 states :

1 Kerala → 1976 (abolished the joint family property)2 Andhra Pradesh→ 19863 Tamil Nadu→19894 Maharashtra and Karnataka→1994

Central Amendments: 2005 (abolished marital status requirement)

Daughters (women) have now the same rights and liabilities as if theywere sons (men)

6/21

Introduction Background Reported violence IPV Conclusion and Discussion

Problem and Strategy

Reverse Causality: Improved inheritance rights are a vehicle toimprove women's economic autonomy. But, violence can a�ectautonomy and autonomy may itself impinge on violence againstwomen: Basu, EJ (2006) and Eswaran et al, CJE (2011)

Use of the exogenous state and year variation caused by the legalamendments to identify the causal impact of female inheritancerights on violence committed against them

Identifying assumption for DiD: States that pass the amendmentswoud have had the same linear trend in crime rates as those who didnot in the absence of the amendments

7/21

Introduction Background Reported violence IPV Conclusion and Discussion

Empirical Estimation

yst = β0 + λt + αs + δTst + β′X+ εst

δ is the di�erence-in-di�erence estimator given by:δ = E [yR

1− yR

0]− E [yNR

1− yNR

0]

yst measured as the (ln) rate per 1000 female population

yst = β0 + λt + αs + δTst + φTst

∑k

Post-law yearsk + β′X+ εst

Coe�cient of interest is φ which captures the dynamic e�ect of theamendments on yst

Grouped years where k = [1, 2]; [3, 4]; [5, > 5]

8/21

Introduction Background Reported violence IPV Conclusion and Discussion

Data

Main dependent variables:Total, Family and Non-Family crimes

State- level analysis for the 16 major states of India spanning theyears 1975-2004

Crime, accidental deaths, suicides data collected from yearlypublications of the National Crime Records Bureau

Controls collected from the Census and Reserve Bank of India;political data collected from several reports from the ElectionCommission

Controls: literacy rate, rural population, growth rate of per capitaincome, female-male ratio, police force per capita, gender of theCM, % seats held by women in VS

Desc. stat.

9/21

Introduction Background Reported violence IPV Conclusion and Discussion

Data: Total crimes against women 1975-2004

Disag. Map

10/21

Introduction Background Reported violence IPV Conclusion and Discussion

Results

Results do not vary in direction and signi�cance when separating betweenNon-Family and Family crimes

Non-Family Family11/21

Introduction Background Reported violence IPV Conclusion and Discussion

Falsi�cation exercises

12/21

Introduction Background Reported violence IPV Conclusion and Discussion

Falsi�cation exercises

13/21

Introduction Background Reported violence IPV Conclusion and Discussion

Falsi�cation exercises

No e�ect found on non-female crimes: robbery and riots

Testing a comparative advantage: Is this e�ect driven by women inrice crop areas in comparison to wheat crop areas? → results seemfavourable

14/21

Introduction Background Reported violence IPV Conclusion and Discussion

Empirical Estimation: Intimate partner violence

yisky = β0 + αs + ζk + λy + β′Xisky + δ1Tisk,(y≥r ′−17,m>r ′) + εisky

yisky dummy variable and it is a measure of the stock level ofviolence if the wife has ever been beaten by the husband

δ1 captures the e�ect of the HSAA for the younger cohorts, i.e. wereless than or 17 at the the reform year r ′ in reform states andunmarried: Tisk,(y≥r ′−17,m>r ′)

Inclusion of a control group term, the older cohort:Tisk,(y<r ′−24,m<r ′)

15/21

Introduction Background Reported violence IPV Conclusion and Discussion

Data

NFHS waves 1998-99 and 2005-06: repeated- cross section acrossbirth cohorts

Surveys have a representative sample of all-India and at state-levelwith response rates above 90%

Household level questionnaires for eligible women i.e. ever-marriedwomen aged between 15-49

Sample is restricted to women who are the wife of the HH andwhere interviewed for the DV section without interruptions

Women in the sample were born between [1949-1991]

Sample restricted to 16 states of India

Control variables include age gap and son preference, number of HHmembers, hindu-type religion, ownership of television and educationand husband's employment status

Desc. stat.

16/21

Introduction Background Reported violence IPV Conclusion and Discussion

Results

Results do not change when including the older women as control groupFalsi�cation

17/21

Introduction Background Reported violence IPV Conclusion and Discussion

Conclusion and Discussion

I �nd that improved property rights increased welfare conditions ofwomen i.e. decreased violence committed against them:

Results of reported crimes suggest a decrease in reported violence

Results are marginally not robust to the inclusion of state-lineartrends

Decrease of domestic violence at the household level persist toseveral tests

18/21

Introduction Background Reported violence IPV Conclusion and Discussion

Reconciliation with the literature

Roy, (2002,2013); Rosenblum, (2013); Anderson and Genicot,(2012); Brule, (2012);

Empowerment e�ect?

Are these the selected women thus, the empowerment e�ect has ane�ect?

No empowerment e�ect on women but on men?

Work- in-progress:

DDD

The e�ect on daughters-in-law?

19/21

Introduction Background Reported violence IPV Conclusion and Discussion

Motherland or Fatherland

Is this result due to empowerment e�ect or a shared control?

20/21

Introduction Background Reported violence IPV Conclusion and Discussion

Conclusion and Discussion

Thank you!Comments are welcome!

Contact: amaral.so�[email protected]

21/21

Results

Old/Young

22/21

Literature

E�ects of changes in surplus within the marriage and in outside options:

Theory: intrahousehold distribution of resources models → mixede�ects→ Tauchen et al, NBER (1991); Bloch and Rao, AER (2000)

Empirics: e�ect of improvements in economic conditions of womenand welfare� ambiguos results between developed and developingcountries→ Bobonis et al, AER( 2012)

Violence against women in India: Iyer et al, AER, (2012); Andersonand Genicot, (2012); Eswaran and Malhotra, CJE (2011); Jensen atal, QJE (2009); Panda and Agarwal, WD(2005)

Hindu Sucession Act: Roy, (2002,2013); Anderson and Genicot,(2012); Brule, (2012); Rosenblum, (2013);

23/21

Hindu Sucession Act

Here

24/21

DataCruelty by Husband and Relatives 1990-2004 FamilyDowry 1988-2004 FamilyDowry Deaths 1990-2004 FamilyImmoral Tra�c 1975-2004 Non - FamilyImportation of Girls 1994-2004 Non - FamilyIndecent Repr. of Women 1994-2004 Non - FamilyMolestation 1990-2004 Non - FamilyKidnapping of Women 1988-2004 Non - FamilyRape 1975-2004 Non - FamilySexual Harassment 1990-2004 Non - Family

Suicides- Property disputes 1988-2004Suicides- Family 1975-2004Suicides- Love a�airs 1975-2004

Accidental Deaths- Other causes (F+M) 1975-2004Accidental Deaths-Natural (F+M) 1975-2004

25/21

Data: Total crimes against women 1975-2004

Disagregated

26/21

Data: Total crimes against women 1975-2004

Map

27/21

Results

Family

28/21

Results

Non-Family

29/21

Further Falsi�cation exercises

1 Comparative advantage?:

1 Pardhan, (1974); Qian, (2005); Mbiti,(2008); Mahajan, (2012);Carranza, (2012): Women's productivity in some agricultural crops ishigher than men;

2 Test whether women in rice crop areas would be more likely tobene�t from the reform

3 Several issues with testing this hypothesis → results seem favourable.

2 Reporting vs. incidence e�ect?

Anderson and Genicot, (2012) show that there was an increase inmale and female suicides due to the HSAA → I con�rm this resultbut not for suicides due to property disputesThe e�ect points to a decrease without associated change inreporting behaviour → Disentangling a decrease in reporting from anon-decrease in incidence does not seem plausible

30/21

Descriptive Statistics

31/21

Descriptive Statistics

32/21