Gender and Conflict by UzLj3VMZ

VIEWS: 0 PAGES: 27

									         Gender and Conflict:
       General Introduction and
        case-study for Burundi
World Bank Workshop on Gender and Conflict, June 10, 2010


                  Philip Verwimp
             Solvay Brussels School of
            Economics and Management
            Université Libre de Bruxelles
               Objectives
• Study the impact of violent conflict on
  gender inequality
• Focus on outcomes that have longer term
  effect, such as schooling, labor market
  participation, marriage and fertility
• Not a study of the direct effects of
  violence, such as torture, kidnapping, rape
• Sample of 6 countries with recent
  experience of violent conflict
              Main methods
• Large scale household data
• Difference-in-Differences analysis,
  use temporal and spatial variation in the extent
  of conflict to compare cohorts exposed to conflict
  with those not exposed in their area of residence
• Interpret the findings taking account of the
  country-specific context of gender and conflict in
  each of the six sample countries.
         Status of the research
• Today we present preliminary results
• Final results towards the end of this year
• Effect of conflict on gender is an empirical question
  Underlying question: how do we conceptualize a
  conflict shock ? Is it something similar to the better
  studied and understood economic shocks ?
• We do not find the same effects in each country
• Pre-existing gender inequalities as well as the type
  and duration of the conflict play and important role
• Results also depend on how conflict is measured
            Intensity of Conflict in Sample Countries

                  Estimates of Anual Battle Fatalities                                                                                         Estimates of Anual Battle Fatalities
                             Estimated Number of Deaths (Burundi x 1,000)
8000




                                                                                                                              100




                                                                                                                                                                                                                      800
                                                                                                                      15000
6000




                                                                                                                                80




                                                                                                                                                                                                                      600
                                                                                                                      10000




                                                                                                                                                                                             Rwanda (x1,000)
                                                                                                  Tajikistan




                                                                                                                                60
4000




                                                                                                                                                                                                                      400
                                                                                                                      5000




                                                                                                                                40
2000




                                                                                                                                                                                                                      200
                                                                                                                                20
       0




                                                                                                                      0


           1990         1995                                                  2000      2005                   2010



                                                                                                                                    0




                                                                                                                                                                                                                      0
                                                                            Año
                                                                                                                                        1990      1995             2000                2005                    2010
                           East_Timor                                                Colombia                                                                      Año
                           Nepal                                                     Tajikistan
                                                                                                                                                           Burundi                 Rwanda
                       Source: PRIO Battle Deaths DATASET 3.0
                                                                                                                                                    Source: PRIO Battle Deaths DATASET 3.0
        Case-Study on Burundi
•   1993-2005 civil war
•   Ethnic, regional, political
•   300.000 people killed
•   GDP per capita decreased by almost 40%
•   Civilian suffering
•   Gendered impact
•   Peace agreement
•   Consociational system of governance
Spatial and temporal distribution of
      the civil war in Burundi
  Province level Poverty (>national mean)
before the War and Duration of War (#years)
     Variables of interest and data set used
• Independent:
Exposure to conflict: number of years that a child is exposed to violent
  conflict during the age of 7-12 or 7-14 in her/his area of residence

• Dependent:
Completion of at least primary schooling (grade 6)

• Surveys used
   - Burundi Household Priority Survey by the World Bank and
     ISTEEBU 1998-2007, 1.400 rural households in all but 3 provinces
  - Demographic and Health Survey (EDS, UNFPA, 2002)
     5.000 rural households, all provinces, including displacement
     camps
Exposure to Conflict during Primary School Age
Exposure (ii)
     Primary School Completion
     by Conflict Area and Cohort
25
20
15
10
 5
 0




      old cohort   young cohort   old cohort    young cohort
          no conflict area              conflict area
               Percentage females by cohorts and
                    conflict, no-conflict areas
                60
                40
female ratio




                20
                 0




                     old cohort   young cohort   old cohort   young cohort
                         no conflict area              conflict area
                 Some caution
• In Burundi the minority of educated citizens has been
  strongly targeted in massacres and genocidal violence
• Violent conflict at least affects the human capital base of
  a country in two ways
  - those who finished school: intentionally attacked
  - those who are in school at the time of the
    conflict : lower school attainment
• Survey we work with is by definition a sample of
  survivors
• When we find effects : they will be underestimate of the
  true effect
                  Long term trend in
              primary school completion
                    male                                 female
.4
.2
 0




     30        40    50    60       70     30      40         50   60   70
                                  age07
                           Fitted values        (mean) prim
Graphs by 2. Sexe
Trend reversal by gender in younger cohorts

                                    male                                        female
       .2
     .15
       .1
     .05




               20              25          30            35     20         25            30   35
                                                       age07
                                                Fitted values        (mean) prim
           Graphs by 2. Sexe
Males only, by conflict, no-conflict area

                       no_conflict_area                               conflict_area
    .4
    .2
     0




         20            25           30             35     20         25          30   35
                                                 age07
                                          Fitted values        (mean) prim
    Graphs by confl3
Females only, by conflict, no-conflict area

                      no_conflict_area                               conflict_area
   .4
   .3
   .2
   .1
    0




        20            25           30             35     20         25          30   35
                                                age07
                                         Fitted values        (mean) prim
   Graphs by confl3
          Length of Exposure to Conflict
         and Primary School Completion

                    male                                female
.2
.1
 0




     0          2    4     6        8      0      2          4   6   8
                                exposure
                           Fitted values       (mean) prim
Graphs by 2. Sexe
EDS data set
    Length of Exposure and
  Number of Years of Schooling

3.5
      3
2.5
      2




          0   2            4                 6    8
                        exposure

                  Fitted values    (mean) years
Primary Schooling, Conflict and Gender
            .25
              .2
% primary



            .15
              .1
            .05
                  0




                      old         young          old          young
                      no_conflict_area             conflict_area
                                          male   female
            Variables in the estimation
• 2 cohorts, 1977-1982, 1982-1987
• Dependent variable: completion of primary schooling
• Regressors:
  - exposure to the conflict (number of years, during
        primary school age )
  - male/female
  - interaction between exposure and gender
• Control variables:
  - linear trend (age) or year of birth fixed effects
  - Household characteristics
  - province fixed effects
• Robustness checks and channels of impact
   - other measures of exposure to violent conflict
   - other outcome variables
Dependent variable:           BHS, 2007               EDS, 2002
Child completed 6 years
of primary schooling          All         All         All         All



                              (1)         (2)         (3)         (4)
Years of Civil War                        -0.483***   -0.272**    -0.326***
Exposure during school
age
                              -0.354***
                              (0.118)     (0.180)     (0.094)     (0.111)
Child is Female               -0.018      -0.111      -0.438***   -0.477***

                              (0.194)     (0.216)     (0.083)     (0.094)
Exposure*Female                           0.226                   0.105
Interaction                               (0.20)                  (0.115)
Intercept                     0.75        0.82        -1.85***    -0.76***

                              (1.02)      (1.1)       (0.52)      (0.92)
Number of children            840         840         5175        5175
Wald Chi      square   test                           136.23***   139.84**
statistics                    56.18***    56.51***
Dependent variable:               HPS, 2007               EDS, 2002
Child completed 6 years of
                                  All         All         All         Poor HH     Non-poor HH
primary schooling
                                  (1)         (2)         (3)         (4)         (5)
Years of Civil War Exposure       -0.501**    -0.552**    -0.241***   -0.191      -0.276**
during school age

                                  (0.219)     (0.22)      (0.118)     (0.166)     (0.162)
Female Child                      -0.024      0.034       -0.34***    -0.26       -0.453**
                                  (0.26)      (0.27)      (0.139)     (0.186)     (0.190)
Exposure*Female interaction       0.36*       0.38*       0.020       -0.08       0.153

                                  (0.22)      (0.23)      (0.134)     (0.213)     (0.178)
Household Head Is Educated        0.51**      0.58**      0.30***     0.34**      0.21

                                  (0.28)      (0.28)      (0.113)     (0.12)      (0.17)
Household Head is Female          0.11        0.098       0.34***     0.48***     0.194
                                  (0.31)      (0.31)      (0.114)     (0.156)     (0.172)
Household Head’s Age              0.006       0.002       0.002       0.034       0.003
                                  (0.008)     (0.008)     (0.007)     (0.01)      (0.01)
HouseholdWealth                   -           -           0.054***    -           -
(livestock, 1993)
                                                          (0.015)
Intercept                         0.83        -1.37*      -3.62***    -5.38***    -4.31***
                                  (1.26)      (0.82)      (0.55)      (0.86)      (1.42)
Number of children                648         648         2582        1512        1072
Wald Chi square test statistics                           203.05***
                                  62.87***    101.74***               191.99***   320.25***
Dependent variable:                   HPS, 2007                 EDS, 2002
Child completed 6 years of
                                      Building    Intensity     Camp         Moved
primary schooling
                                      destroyed   of conflict   experience   residence
                                      (1)         (2)           (3)          (4)
School      destroyed      (village
level)                                -0.82
                                      (0.951)
Number of people killed                           -0.065**
(village level)                                   (0.061)
Time spent in a displacement                      -             -0.016
camp (in years)                       -
                                                                (0.068)
Number of times moved                             -             -            -0.18**
Residence                             -
                                                                             (0.092)
Female Child                          0.23        0.23          -0.34***     -0.29**
                                      (0.26)      (0.25)        (0.122)      (0.132)
Shock variable*female                 -0.196      -0.009        -0.02        -0.11
Interaction                           (1.32)      (0.014)       (0.08)       (0.128)
Intercept                             -1.15       -0.063        -5.05***     -4.78***
                                      (1.2)       (1.18)        (0.83)       (0.81)
Number of children                    648         648           2640         2640
Wald Chi2 test-statistic              62.19***    59.18***      228.05***    242.12***
         Preliminary conclusions
• Boys and girls exposed to violent conflict during school-age have a
  smaller chance to complete primary schooling in Burundi
• The effect we measure, 30% less chance to complete for every
  additional year of conflict is an underestimate because
   – it is not the only devastating effect of conflict: educated adults
      run a higher risk to be killed
   – Because of that, our control group in the conflict
      affected area is biased towards lower educated adult
• The effect is larger for boys which seems logic as they have more to
  loose, in particular boys from non-poor families
• So exposure to conflict in Burundi seems not to work as an
  economic shock, because non-poor households cannot protect
  themselves from it.
• Exposure to conflict (fighting) in the area of residence as well as its
  intensity and the frequency of displacement seems to be channels
  driving the result
• Children from poor families: household characteristics !

								
To top