The Economic Determinants of Domestic Violence

描述（由申请人提供）：家庭暴力的经济决定因素安娜·艾泽尔，布朗大学经济学助理教授，美国每天大约有1.4万名妇女被她们的亲密伴侣殴打，其中4人被杀害，这促使前卫生部长C. Everett Koop将家庭暴力称为“美国最重要的健康问题”。尽管家庭暴力很普遍，代价也很高，但关于家庭暴力的经济决定因素，仍有许多问题有待了解。我建议分析妇女的相对收入对殴打发生率的影响。关于相对收入对暴力的影响，现有的理论是模棱两可的。一个讨价还价的经济模型预测，当一个女人的资源相对于她丈夫的资源增加时，针对她的暴力就会减少。其他理论，其中的交换理论，预测的是相反的。要确定这种关系的性质，还得靠实证研究。利用调查数据考察妇女工资对暴力的影响的现有实证研究普遍发现了一种负相关关系——工资较低的妇女遭受更多的暴力。然而，这项工作在两个主要方面受到限制。首先是测量的问题。调查容易漏报，通常不允许计算稳定的小区域（州/地方）估计值，因为需要的样本太大。第二，这些研究未能建立家庭暴力与妇女工资之间的因果关系，例如，未能解释被忽略的变量偏差或反向因果关系（内生性）的可能性。也就是说，他们发现的负面关系可能归因于混杂因素（导致暴力和低工资的女性未观察到的属性），或者是高水平的暴力可能导致女性收入下降，而不是相反。我提出两个策略来克服这些障碍。首先，我使用多个管理数据集来制定家庭暴力的措施。这些数据包括：因袭击而住院、报警、因家庭暴力而被捕以及亲密伴侣谋杀率。我还使用了个人层面的调查数据，以及来自加州妇女健康调查的不太严重的暴力、威胁和“控制行为”的自我报告。虽然没有一种措施是完美的，但这些措施加在一起可以捕捉到各种各样的暴力行为。使用这些多种暴力措施将揭示暴力关系的不同方面如何受到妇女相对收入变化的影响。为了克服第二个障碍并建立收入与家庭暴力之间的因果关系，我将利用某些行业由女性主导（例如服务业）而其他行业由男性主导（例如建筑业）这一事实。这些部门需求的增加分别导致女性和男性工资的外生增长。通过关注当地劳动力市场状况和外生劳动力需求冲击，我可以确定相对收入变化对家庭暴力的影响。基于这项工作，我计划探讨家庭暴力对孕产妇和儿童健康的影响以及经济资源的潜在缓解作用。以前的工作已经证实，当母亲控制更多的家庭资源时，儿童的健康状况会得到改善。这些进步很大程度上归因于女性在子女身上投入了更多的物质。然而，我的研究表明了另一种潜在的机制：减少暴力。利用加州个人层面的出生和死亡率数据与怀孕期间因殴打住院的数据相关联，我使用工具变量策略提供了暴力对出生结果影响的第一个因果估计。接下来是对暴力影响生育结果的途径的探索性分析——通过钝性创伤的直接影响或增加产妇冒险和延迟寻求产前护理的间接影响（遵循Newberger， 1992）。这项工作将成为今后工作的基础，这些工作将审查妇女资源和家庭暴力在决定一般妇幼保健方面所起的作用。在美国，每天大约有1.4万名女性受到亲密伴侣的殴打，其中4人被杀害，这促使前卫生部长C. Everett Koop将家庭暴力称为“美国最重要的健康问题”。尽管家庭暴力很普遍，代价也很高，但关于殴打及其后果的经济决定因素仍有许多问题有待了解。我建议1)分析妇女的相对收入对殴打发生率的影响，2)估计怀孕期间暴力对分娩结果的影响。更好地了解殴打的原因和后果将有助于为家庭暴力政策提供信息。

DESCRIPTION (provided by applicant): The Economic Determinants of Domestic Violence Anna Aizer, Asst Professor of Economics, Brown University Everyday roughly 14 thousand women in the US are battered and four are killed by their intimate partners, prompting former Surgeon General C. Everett Koop to label domestic violence "the single most important health issue in the US." Despite the prevalence and high costs of domestic violence, much remains to be understood about the economic determinants of battering. I propose to analyze the impact of women's relative earnings on the prevalence of battering. Existing theory is ambiguous with respect to the impact of relative income on violence. An economic model of bargaining predicts that as a woman's resources increase relative to those of her husband, violence against her should decline. Other theories, exchange theory among them, predict the opposite. It is left to empirical work to determine the nature of the relationship. Existing empirical research examining the impact of women's wages on violence using survey data have generally found a negative relationship - women with lower wages experience more violence. However, this work is limited in two main respects. First is the problem of measurement. Surveys are prone to underreporting and do not typically allow one to calculate stable small area (state/local) estimates since prohibitively large samples would be required. Second, these studies fail to establish a causal relationship between domestic violence and women's wages by, for example, failing to account for the potential for omitted variable bias or reverse causality (endogeneity). That is, the negative relationship they find may be attributed to confounding factors (unobserved attributes of women that result in both violence and lower wages), or the fact that higher levels of violence may cause women's earnings to decline, not the other way around. I propose two strategies to overcome these obstacles. First, I use multiple administrative datasets to develop measures of domestic violence. These data include: hospitalizations for assaults, calls to the police, arrests for domestic violence as well as intimate partner homicide rates. I also use individual level survey data with self- reports of less severe violence, threats and "controlling behavior" from the California Women's Health Survey. While and no one measure is perfect, together these measures capture a wide range of violent behavior. Using these multiple measures of violence will shed light on how different aspects of violent relationships are affected by changes in women's relative earnings. To overcome the second obstacle and establish a causal relationship between earnings and domestic violence, I will take advantage of the fact that certain industries are dominated by women (e.g., services) and others by men (e.g. construction). Increases in demand in these sectors results in exogenous increases in female and male wages, respectively. By focusing on local labor market conditions and exogenous labor demand shocks, I can identify the impact of changes in relative income on domestic violence. Based on this work, I plan to explore the impact of domestic violence on maternal and child health and the potential mitigating effects of economic resources. Previous work has established that child health improves when mothers control a greater share of the household resources. These improvements have largely been attributed to women's greater material investments in their children. However, my research suggests another potential mechanism: reductions in violence. Using individual-level California natality and mortality data linked to data on hospitalization for assault during the pregnancy, I provide the first causal estimates of the impact of violence on birth outcomes using an instrumental variable strategy. This is followed by exploratory analysis of the pathways by which violence affects birth outcomes - through the direct effect of blunt trauma or the indirect effects of increased maternal risk taking and delays in seeking prenatal care (following Newberger, 1992). This work will form the basis for future work that will examine the role that women's resources and domestic violence play in determining maternal and child health generally. Everyday roughly 14 thousand women in the US are battered and four are killed by their intimate partners, prompting former Surgeon General C. Everett Koop to label domestic violence "the single most important health issue in the US." Despite the prevalence and high costs of domestic violence, much remains to be understood about the economic determinants of battering and its consequences. I propose to 1) analyze the impact of women's relative earnings on the prevalence of battering and 2) estimate the costs of violence during pregnancy on birth outcomes. A better understanding of both the causes and consequences of battering will help to inform domestic violence policies.