21
The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Embed Size (px)

Citation preview

Page 1: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

The Impact of Court Decentralization on Domestic Violence Against Women

Raúl AndradeJimena Montenegro

March 2009

Page 2: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

ObjectivesGeneral objective:

Evaluate if increasing access to formal justice plays a key role in decreasing rates of domestic violence against women

Specific objective:Measure the impact of a program to decentralize courts carried out in Peru between 1999 and 2002 on intimate partner violence against women

Questions to address: >> Is making courts available to poor localities a strategy that may help to enforce women rights? >> What is the effect of making judicial services available on intimate partner violence against women in Peru?>> Are effects different when condition by rural/urban areas?>> Are effects different for physical violence and psychological violence?

Page 3: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Motivation i) High levels of domestic violence against women:

41% of women in Peru suffered some type of physical aggression during their relationship. 25% of them reports to be humiliated by their intimate partners

ii) The third part of the Peruvian population lack of adequate access to justice: >> Lack of supply>> Unequal distribution of judicial services >> High cost of litigation>> Cultural barriers

iii) There is an important discussion regarding to what extent formal institutions may help to address the problem of domestic violence. But there is no empirical evidence

Page 4: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Set up of the study: Impact of the MBJ Program

Specific objectives of the program:

>> To improve availability of judicial services in regions far away from urban centers,

Increase the supply of judicial servicesMake the distribution of this supply more equal among poor

regions

>> Courts were built in places where because of geographical characteristics and lack of infrastructure, traveling to courts was very expensive. Mainly poor urban and rural areas

>> Courts are mainly focused on judicial problems at the basic levels, they have family judges, judges of peace and civil judges

>> Judges were made responsible only for jurisdictional chores, while administrative tasks were assigned to specialized staff

Page 5: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Methodology 1: Overview

>> Impact of living in a locality under the jurisdiction of an MBJ on:Whether a woman is a victim of intimate partner physical violenceWhether a woman is a victim of intimate partner psychological violence

>> Combination of matching and instrumental variables techniques

-- Propensity score matching helps to make treatment and control groups comparable in terms of observables characteristics. At the district level.Determinants of court location at the district level measured before the treatment are used to compute the propensity score. Matching one-to-one is used to find for each treated locality a similar control district

-- Instrumental variable techniques generates exogenous variability within these groups, that are already similar in observable characteristics. At the individual levelMBJ were implemented during Fujimori’s regime. We use electorate outcomes to instrument the final location of an MBJ. Done at the individual level (with clusters at the district level)

Page 6: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Methodology 2: Overview

Step 1: Propensity score matching at the district level

Comparable groups at the district level

Identifying assumption 1:

Districts are on average similar

Comparable groups at the woman level

Identifying assumption 2: Women are on average similar

Step 2: Instrumental variables techniques in

sample of women living in matched

districts

Exogenous variability within this

sample

Identifying assumption 3: - Political variable is

partially correlated allocation of MBJ

- Can be excluded from regression of interest

Page 7: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Methodology 3: Assumption 1

Propensity score distribution before and after the matching

Before After

01

23

45

kden

sity

psc

ore

0 .2 .4 .6 .8x

01

23

45

kden

sity

psc

ore

0 .1 .2 .3 .4 .5x

Page 8: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Methodology 4: Assumption 1Variables Treated

Locations Witness

Locations Difference Mean

comparison test1/

Population_1999 9,638.8790 9,249.4130 389.4667 0.7805 Poverty_ranking 925.1750 895.4125 29.7625 0.4436 Absolute_poverty_index 0.4563 0.4635 -0.0072 0.4761 Population_density 154.6534 97.9623 56.6911 0.2358 Km_provincial_capital 48.9875 49.0292 -0.0417 0.6131 Road1 0.2667 0.2458 0.0208 0.5363 Road3 0.3917 0.4208 -0.0292 0.5113 Road5 0.2625 0.2417 0.0208 0.5595 Road7 0.0458 0.0667 -0.0208 0.1660 Road9 0.0333 0.0250 0.0083 0.5648 Altitude_logarithm 7.4764 7.4789 -0.0025 0.9656 Malnutrition_rate 38.3852 39.2420 -0.8568 0.3029 Posts_deficit 5.4167 5.0667 0.3500 0.7288 Classrooms_deficit 4.3833 5.7375 -1.3542 0.3151 Population_without_water 41.2497 40.3518 0.8979 0.7529 Population_without_drain 77.0955 76.3525 0.7430 0.7573 Population_without_electricity 54.7787 60.3433 -5.5646 0.0460 * Proportion of children under 10 0.2926 0.2890 0.0036 0.3062 Proportion of young under 20 0.2198 0.2200 -0.0001 0.9551 Proportion of adults under 65 0.4238 0.4271 -0.0032 0.3073 Proportion of employees 0.0913 0.1004 -0.0091 0.1502 Proportion of unpaid family 0.1932 0.2046 -0.0114 0.2538 Proportion of household workers 0.0203 0.0212 -0.0009 0.7103 Proportion of independent 0.4664 0.4599 0.0065 0.6888 Proportion of employers 0.0157 0.0104 0.0053 0.1535

Page 9: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Methodology 5: Assumption 2

Women living in Treated

Locations

Women living in Witness

Locations

Difference

Mean

comparison

test1/Language: Spanish 0.8306 0.8735 0.0429 0.0000 *Childhood place of residence: Capital/large city 0.3347 0.3083 -0.0264 0.0000 *Age 29.7178 29.9978 0.2799 0.0411 *Ever pregnant 0.6690 0.6963 0.0273 0.0000 *Number of children under 5 0.6964 0.6765 -0.0200 0.0848 *

Women living in Treated

Locations

Women living in Witness

Locations

Difference

Mean

comparison

test1/Language: Spanish 0.7709 0.8897 0.1188 0.0000 *Childhood place of residence: Capital/large city 0.2040 0.3072 0.1033 0.0000 *Age 29.9384 30.0891 0.1507 0.4694Ever pregnant 0.6984 0.6892 -0.0091 0.3441Number of children under 5 0.7374 0.7175 -0.0199 0.2841

Mean differences in women's characteristics

General sample

Matched sample

Page 10: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Political variables as a determinant of final MBJs location:

>> It is known that Fujimori used to allocate public resources and social services (schools, health facilities, roads, among others) according to political criteria (Schady 1999, 2002; Paxson, C. and N. R. Schady (2002)

>> According to Alonso (2003) and interviewed authorities this happened as well in this program. The initial number of planned courts were 83. Because of budget cuts only 43 were built. The final selection of the localities where the courts were built was based on political interests

>> First stage regression results (to be shown) support this idea

Exclusion restriction:

>> There is no reason to think that responses from families to political strategies affect domestic violence, once controlling for a variety of factors, such as level of income, education, alcohol consumption, history of family violence, native language, etc.

Methodology 6: Assumption 3

Page 11: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Data

>> Data from the Demographic and Health Surveys (ENDES Continua 2004-2006), prepared by the National Institute of Statistics and Informatics (INEI)

>> In recent years, ENDES included a module on domestic violence in order to study the linkages between violence, health, and demographic outcomes

>> In Peru, the violence module has been included since the year 2004. The questionnaire is answered by all women between 15 and 49 years old present in the household. We use data for 2004; 2005 and 2006

Page 12: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Results 1: Physical violenceTotal Rural Urban Total Rural Urban

OLS 0.244 0.110 -0.500 0.0357 0.033 -0.025(0.558) (0.549) (0.324) (0.0159) (0.024) (0.0214)

Obs. 17760 6105 11655 4352 1794 2558

IV: Prop. of votes for Fujimori 2000 1.362 0.548 1.357 -0,347* -0,195* 0,26*(1.720) (5.250) (0.960) (0.1174) (0.074) (0.367)

Obs. 17732 6105 11655 4352 1794 2558

First stage -0,446* -0,1459* -0.333 -0,768* -1,48* 0,477*(0.184) (0.041) (0.005) (0.082) (0.102) (0.133)

Obs. 17732 6105 11655 4352 1794 2558

IV: Prop of votes for non-traditional party 1998 -2.063 1,464* 18.46 -0,442* -0,209* -0,193(1.880) (0.537) (33.88) (0.136) (0.218) (0.178)

Obs. 17635 6105 11530 4298 1794 2504

First stage -0,3866* -1,147* -0.0189 0,831* -0,968* 1,613*(0.016) (0.0375) (0.015) (0.099) (0.197) (0.111)

Obs. 17635 6105 11530 4298 1794 2504

General Sample Matched Sample - IDEN

Page 13: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Results 2: Psychological violence

Total Rural Urban Total Rural UrbanOLS 0.258 0.147 -0.749 0.0393 0.111* -0.0141

(0.532) (0.526) (0.309) (0.0150) (0.023) (0.0206) Obs. 17732 6105 11655 4352 1794 2558

IV: Prop. of votes for Fujimori 2000 1.44 -0.05 22.27 -0,277* -0.233* 0.1488(1.647) (5.019) (27.850) (0.1110) (0.073) (0.288)

Obs. 17732 6105 11627 4352 1794 2558

First stage -0,446* -0,1459* -0.333 -0,768* -1,48* 0,477*(0.184) (0.041) (0.005) (0.082) (0.102) (0.133)

Obs. 17732 6105 11655 4352 1794 2558

IV: Prop of votes for non-traditional party 1998 -3.42 1.57 37.85 -0.331* 0.178 -0.167*(1.810) (0.513) (42.35) (0.127) (0.192) (0.075)

Obs. 17635 6105 11530 4298 1794 2504

First stage -0,3866* -1,147* -0.0189 0,831* -0,968* 1.613*(0.016) (0.0375) (0.015) (0.099) (0.197) (0.111)

Obs. 17635 6105 11530 4298 1794 2504

General Sample Matched Sample - IDEN

Page 14: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Results 3: Total sampleProportion of women victims of domestic violence: treated and counterfactual

(matched sample)

33.6

44

27.17

34

0

15

30

45

60

Treated Treated counterfactual Treated Treated counterfactual

Physical Psychological

Page 15: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Results 4: Rural sampleProportion of women victims of domestic violence: treated and counterfactual

(matched rural sample)

37.1642

31.3836

0

15

30

45

60

Treated Treated counterfactual Treated Treated counterfactual

Physical Psychological

Page 16: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Results 5: Urban sampleProportion of women victims of domestic violence: matched and counterfactual

(urban matched sample)

31.17

23 24.3 24.3

0

10

20

30

40

50

Treated Treated Counterfactual Treated Treated Counterfactual

Physical Psychological

Page 17: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

>> The program helped reducing the rate of domestic physical violence in around 10%

>> This effect is driven, basically by the effects in rural areas, where the decrease in domestic violence is larger (in terms of coeffcients)

>> The decrease in the rate of domestic physical violence is accompanied by a smaller effect on psychological domestic violence, both in the total sample and in the rural sample

>> In urban areas, the effect takes the opposite direction. There is an increase in physical domestic violence, but no statistically significant effect on the rate of psychological domestic violence (Explanations?)

Conclusions

Page 18: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Tabla 1. Domestic Violence Against Women

Psychological Physical

Income Quintile First 22 % 36 % Second 24 % 43 % Third 28 % 46 % Fourth 26 % 42 % Fifth 21 % 33 % Área of residence Urban 26 % 42 % Rural 22 % 38 % Total 24.9 40.9 Authors’ calculations. Base don DHS.

Page 19: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Variables GENERAL SAMPLE

(SE) MATCHED SAMPLE

(SE)

Age at first marriage

2.5344 (0.8184) 2.5453 (0.8174) Age 33.8475 (8.4068) 34.0190 (8.3628) Age squared

1216.3240 (569.5543) 1227.2140 (570.0313) Years of education 8.0475 (4.5795) 7.4993 (4.6795) Total household members 5.4809 (2.2310) 5.4228 (2.0215) Number of children under 5 0.8197 (0.8643) 0.8346 (0.8504) Electricity (dummy) 0.7098 (0.4539) 0.6707 (0.4700) Household in urban area

0.6110 (0.4875) 0.5512 (0.4974) Second quintile 0.2483 (0.4320) 0.2914 (0.4544) Third quintile 0.2259 (0.4182) 0.2125 (0.4091) Fourth quintile 0.2049 (0.4036) 0.1717 (0.3772) Fifth quintile 0.1251 (0.3308) 0.0952 (0.2935) 2 unions or more 0.1021 (0.3028) 0.0832 (0.2762) Duration of marriage 5-9 years 0.1994 (0.3996) 0.1991 (0.3993) Duration of marriage 10-15 years 0.1937 (0.3952) 0.1947 (0.3960) Duration of marriage >15 0.4395 (0.4963) 0.4444 (0.4970) Partner drinks alcohol (dummy) 0.7548 (0.4302) 0.7553 (0.4300) Father ever bit her mother (dummy) 0.4541 (0.4979) 0.4467 (0.4972)

Descriptive: Control Variables

Page 20: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

Table 3. Location determinants of MBJ

Results of the estimated probit model (marginal effects)

Dependent variable: district attended by MBJ = 1 Absolute poverty index 0.4957 (0.2482)** Population density logarithm 0.0279 (0.0070)*** Dummy provincial capital -0.0875 (0.0181)*** Classification of poverty: Extreme poverty 0.0268 (0.1756) Classification of poverty: Very poor 0.0329 (0.1503) Classification of poverty: Poor 0.0066 (0.1303) Classification of poverty: Regular 0.0518 (0.1418) Km. to provincial capital 0.0003 (0.0001)** Main access road: Road and/ or rail -0.0523 (0.0255)** Main access road: Air or Rural road -0.1601 (0.0302)*** Main access road: Waterway 0.0851 (0.0993) Main access road: Small rural road and/ or non-motorized road

-0.1621 (0.0146)***

Andes region -0.0788 (0.0477)* Rainforest -0.0765 (0.0282)*** Log of Altitude logarithm 0.0459 (0.0117)*** Malnutrition rate -0.0030 (0.0015)* Posts deficit 0.0011 (0.0008) Classrooms deficit -0.0005 (0.0004) Observations 1733 R-squared 0.683 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Page 21: The Impact of Court Decentralization on Domestic Violence Against Women Raúl Andrade Jimena Montenegro March 2009

THANK YOU