中文摘要
不同社会人口群体的经济结果存在显著差异。在一个多维身份模型中，我们表明，经济参与的差异如何产生特定于身份的规范，而这些规范又如何将各群体历史上的代表性不足“正常化”并将其锁定。标准方法将各身份维度视为相互独立，而我们的分析揭示了基于种族、性别及其他特征的不平等与代表性不足之间的深层联系。仅针对单一身份维度、而不影响其他身份维度的“隔离式干预”通常是不可能的。旨在减少某一身份维度上代表性不足的干预，可能会加剧另一身份维度上的代表性不足。我们表明，如何通过一套采用（a）自融资补贴或（b）榜样的干预体系，消除每个维度上的代表性不足；其中的干预具有“交叉性”，即在各身份维度之间相互关联。当群体认同强度被内生化时，代表性不足会在长期内自行消失。视具体环境而定，平权行动可能加速，也可能减缓这一过程。
Abstract
Economic outcomes vary significantly across socio-demographic groups. In a model of multidimensional identity, we show how differences in economic participation can give rise to identity-specific norms which ‘normalize’ and lock in the historical underrepresentation of various groups. Whereas standard approaches treat identity dimensions as independent, our analysis reveals deep connections between inequality and underrepresentation based on race, gender, and other characteristics. ‘Sterilized interventions’ along a single identity dimension are generally impossible. Interventions that aim to reduce underrepresentation along one identity dimension can increase underrepresentation along another. We show how underrepresentation can be eliminated along every dimension, through a system of (a) self-financing subsidies or (b) role models, where interventions are ‘intersectional’, i.e., connected across identity dimensions. When the strength of group identification is made endogenous, underrepresentation disappears in the long run on its own. Depending on the environment, affirmative action can either speed up or slow down this process.