the dynamics of domestic violence
play

The Dynamics of Domestic Violence and Sauer Learning about the - PowerPoint PPT Presentation

The Dynamics of Domestic Violence Anderberg, Mantovan The Dynamics of Domestic Violence and Sauer Learning about the Match Introduction The Data The Model Dan Anderberg, Noemi Mantovan , Robert M. Sauer Solution and


  1. The Dynamics of Domestic Violence Anderberg, Mantovan The Dynamics of Domestic Violence and Sauer Learning about the Match Introduction The Data The Model Dan Anderberg, ∗ Noemi Mantovan ∗∗ , Robert M. Sauer ∗∗∗ Solution and Estimation Estimation ∗ RHUL, IFS, CESifo, ∗∗ Bangor University, ∗∗∗ RHUL, IZA Results Counterfactual Experiments November 8, 2017 Conclusion

  2. Abuse is Widespread The Dynamics of Domestic Violence Anderberg, Mantovan and Sauer Crime Survey of England and Wales 2015 Introduction Over 8% of women experienced domestic abuse The Data Domestic abuse accounts for 20% of all reported violent The Model incidents Solution and Estimation Highest rate of repeat victimization of any type of crime Estimation Results Counterfactual Experiments Conclusion

  3. Economic Research on Abuse The Dynamics of Domestic Violence Anderberg, Has mostly focused on Mantovan and Sauer variation by educational attainment, labour market conditions, culture and social norms Introduction other triggers such as emotional cues and instrumental The Data violence The Model impact of law enforcement, welfare and cash-transfer Solution and Estimation policies Estimation No studies on dynamic and simultaneous links between Results abuse, labour supply, partnership status and fertility Counterfactual Experiments Conclusion

  4. Our Contribution The Dynamics of Domestic Violence Estimate dynamic model of labour supply, partnership status Anderberg, and fertility with learning about partner’s abusive type Mantovan and Sauer Builds on Brian, Lillard and Stern (2006) Introduction women choose partnership status and learn about type but The Data abstract from labour supply and fertility is exogenous The Model Builds on Bowlus and Seitz (2006) Solution and women choose partnership status and labour supply but Estimation no learning about type and fertility is exogenous Estimation Results Builds on Keane and Wolpin (2010) Counterfactual women choose labour supply, partnership status and Experiments fertility but no abuse or learning Conclusion

  5. Main Research Questions The Dynamics of Domestic In our more comprehensive environment, we address the Violence following questions: Anderberg, Mantovan What is the e ff ect of uncertainty about partner’s violent and Sauer nature? Introduction does it lead to delays in marriage-specific investments, The Data most notably fertility? The Model What are the labour supply responses of women facing Solution and possible domestic violence? Estimation do certain labour supply choices trigger domestic abuse? Estimation Results What is the e ff ect of female “empowerment” on abuse Counterfactual rates? Experiments Conclusion through higher wages more generous childcare support

  6. Avon Longitudinal Study of Parents and Children The Dynamics of Domestic ALSPAC also known as “Children of the 90s” survey Violence Pregnant women with estimated delivery dates between Anderberg, Mantovan April 1991 and December 1992 and Sauer Questions on abuse annually until child was 6 years old Introduction was partner physically cruel The Data was partner emotionally cruel The Model subjective measure aligns with individual’s expectations Solution and “any” abuse gives similar incidence as British Crime Survey Estimation Estimation Drop non-white women and other standard restrictions Results 9,359 women between ages of 17 and 40 Counterfactual Experiments 56,926 woman-year observations Conclusion over 80 percent with observations for all seven years impute wages from UK Labour Force Survey

  7. Descriptive Statistics at Baseline The Dynamics of Domestic Violence Means at Mid-Pregnancy Anderberg, Mean St. Dev Mantovan and Sauer Age 28.1 4.5 Introduction Married .96 .19 The Data Marriage Duration 4.8 3.5 The Model Has Child .55 .50 Solution and Number Children .78 .89 Estimation Low Qualification .24 .43 Estimation Results Medium Qualification .38 .49 Counterfactual Experiments High Qualification .37 .49 Conclusion N 9,359

  8. Descriptive Statistics - Domestic Abuse The Dynamics of Domestic Violence Anderberg, Physical Emotional Any Mantovan and Sauer Mean .024 .087 .092 Introduction N 56,926 56,926 56,926 The Data The Model Any Abuse Time t+1 Solution and 0 1 Estimation Estimation 0 .943 .057 Results Time t 1 .505 .495 Counterfactual Experiments Conclusion

  9. Domestic Abuse by Age The Dynamics of Domestic .15 Violence Anderberg, Mantovan and Sauer Introduction .1 The Data The Model Solution and Estimation .05 Estimation Results Counterfactual Experiments 0 Conclusion 17 − 24 25 − 31 32 − 45 Physical Emotional Any

  10. Domestic Abuse by Education The Dynamics of Domestic Violence .1 Anderberg, Mantovan and Sauer .08 Introduction .06 The Data The Model Solution and .04 Estimation Estimation Results .02 Counterfactual Experiments 0 Conclusion Low Medium High Physical Emotional Any

  11. Descriptive Statistics - Work, Partnership, Fertility The Dynamics of Domestic Violence Anderberg, Mantovan and Sauer Mean N Nonemployed .471 53,746 Introduction The Data Part-time .345 53,746 The Model Full-time .184 53,746 Solution and Married .937 56,926 Estimation Birth .121 37,876 Estimation Results Counterfactual Experiments Conclusion

  12. LPMs with Fixed E ff ects The Dynamics of Domestic Violence Anderberg, Ab(t-1,t) UE(t) Div(t-1,t) B(t-1,t) Mantovan and Sauer (1) (2) (3) (4) Introduction Ab(t-1,t) -.018 The Data Ab(t-2,t-1) .030** -.027** The Model PT(t-1) -.009* Solution and Estimation FT(t-1) .027** Estimation Controls Yes Yes Yes Yes Results N 33,015 31,485 34,482 35,033 Counterfactual Experiments Conclusion

  13. Optimization Problem The Dynamics of Domestic Violence Anderberg, Mantovan and Sauer Discrete choice dynamic programming problem Introduction At the start of each period t, a woman chooses to be The Data in non-employment, part-time or full-time work, The Model k t ∈ { 0 , 1 , 2 } Solution and single or married m t ∈ { 0 , 1 } (marriage o ff er probability ς ) Estimation pregnant or not f t ∈ { 0 , 1 } Estimation Results Counterfactual Experiments Conclusion

  14. Abuse Environment The Dynamics of Domestic Violence Abuse is a semi-endogenous stochastic process Anderberg, Mantovan Males of two possible unknown types: "non-violent and Sauer nature" and "violent nature" Introduction A violent man r = 0 will abuse z t = 1 with probability χ k 0 The Data A non-violent man r = 1 will abuse z t = 1 with probability The Model χ 1 < χ k Solution and 0 Estimation φ t is belief partner is non-violent type at time t (in state Estimation Results space) Counterfactual φ t = φ b at start of new partnership: proportion of Experiments Conclusion non-violent types in population

  15. Learning Dynamics The Dynamics of Belief about partner’s nature updated according to Bayes’ Domestic Violence rule (law of motion) Anderberg, Updating belief partner is non-violent when z t − 1 = 0 (no Mantovan and Sauer abuse last period): Introduction φ t − 1 ( 1 − χ 1 ) The Data φ t | z t − 1 = 0 = 0 ) . The Model φ t − 1 ( 1 − χ 1 ) + ( 1 − φ t − 1 )( 1 − χ k Solution and Estimation Updating belief partner is non-violent type when z t − 1 = 1: Estimation Results φ t − 1 χ 1 Counterfactual φ t | z t − 1 = 1 = . Experiments φ t − 1 χ 1 + ( 1 − φ t − 1 ) χ k 0 Conclusion Belief enters utility flow thus a ff ecting all three choice dimensions

  16. Utility Flow and Consumption The Contemporaneous Utility Dynamics of Domestic Violence U t = µ k t C 1 − λ Anderberg, t − ¯ t Ψ m Ψ z m t + Ψ n � � + Mantovan t t 1 − λ and Sauer t = ψ m + ε m Ψ m t Introduction ⇣ ⌘ Ψ z ¯ φ t χ 1 + ( 1 − φ t ) χ k t ψ z The Data t = 0 The Model Ψ n t = β n 1 n t − β n 2 n 2 t + f t ε f Solution and t Estimation n t + 1 = n t + f t Estimation Results Consumption Counterfactual Experiments Conclusion ( � w t + w h � τ t − c t if m t = 1 C t = if m t = 0 w t − c t

  17. Wage O ff ers and Child Care Costs The Dynamics of Wage O ff ers Domestic Violence Anderberg, ⇣ ⌘ 3 x 2 w k β k 0 + β k 1 a + β k 2 x t + β k t + ε k t = exp Mantovan t and Sauer ⇣ ⌘ w h β h 0 + β h 1 a + β h 2 t t + β h 3 t 2 t + ε h t = exp Introduction t The Data Pr ( a = 1 | q ) Pr ( a = 0 | q ) = exp ( β a 0 + β a 1 d q = 1 + β a 2 d q = 2 ) The Model Solution and Estimation x t + 1 = x t + k t Estimation k = 1 , 2 Results Counterfactual Experiments Child Care costs Conclusion c t = ρ k t ( β c 1 n t + β c 2 n 2 t ) − ( β c 3 n t + β c 4 n 2 t )( 1 − m t )

Recommend


More recommend