Social Interactions: Theory, Econometrics, and Policy Implications

1. 本提案由一系列相互关联的研究项目组成，探索社会互动的理论建模和统计识别，以及不确定经济环境下的政策评估。项目的第一部分将分析社会互动的模型。社会互动模型被用来解释一系列社会经济结果，在理解群体间持续存在的不平等方面似乎尤为重要。这里提出的工作将在几个方向上扩展现有的文献。一个主要主题是社会互动的计量经济学鉴定。虽然在许多情况下，社会因素在理解个人社会经济结果方面似乎很重要。拟议的研究将开发新的策略来识别社会互动，特别关注当研究人员无法测量个人群体的所有特征时，如何识别不同类型的社会互动。这项研究将有助于把关于社会互动的经验主张建立在比目前更坚实的基础上。其次，本研究将整合社会互动效应模型和群体形成模型，并分析这种整合的理论和计量意义。在居民区等环境中，社会互动是群体构成的主要决定因素，因此完整的不平等理论必须考虑到这一点。该建议的第二部分将开发工具，以促进对替代性政府政策的评估。其中一个主题将探讨当政策制定者不知道他希望影响的现象的真实模型时的政策设计。模型不确定性的发生可能是因为对某些感兴趣的结果的理论决定因素的一般问题的不确定性，对定量指定理论的方式的不确定性，或者是因为对观察单位的测量和异质性问题的不确定性，这些似乎是社会科学中特别突出的因素。分析将使用最近在统计学中发展并开始在经济学中应用的模型平均方法来解决模型不确定性问题。最后，该提案描述了研究政府政策设计中的基本限制的研究。政府政策的选择通常会影响某些利益结果的平均值和波动水平。在控制理论文献中，存在许多关于特定频率权衡的结果，这意味着任何试图稳定某些结果的政府政策都将不可避免地减少某些频率的波动，同时加剧其他频率的波动。这些权衡似乎没有在社会科学背景下得到利用。2. 拟议活动的更广泛影响拟议的研究可能有三个更广泛的影响。首先，它代表了一种整合个人行为和群体结果的社会学和经济学观点的努力。因此，它试图打破跨学科的障碍，可能在社会科学中具有普遍价值。其次，这项工作可能被证明与政策相关。对模型不确定性的分析将有助于开发适合于明确评估政策效果的计量经济学工具。第三，对设计限制的分析代表了一个新的视角，即政府政策如何稳定不同频率下的动态社会经济结果，以及不同频率下的稳定如何被权衡。最后，作者计划编写一本研究生水平的社会互动专著，以促进学生的培训。

1. Intellectual Merit of the Proposed ActivityThis proposal consists of a set of interrelated research projects that explore the theoretical modeling and statistical identification of social interactions and the evaluation of policies in uncertain economic environments. The first part of the project will analyze models of social interactions. Social interactions models have been proposed to explain a range of socioeconomic outcomes and seem especially important in understanding persistent inequality across groups. The work proposed here will extend the existing literature in several directions. One main theme is the econometric identification of social interactions. While in many contexts, social factors seem important in understanding individual socioeconomic outcomes. The proposed research will develop new strategies for identifying social interactions with particular attention to how one can identify different types of social interactions when a researcher cannot measure all of the characteristics of an individual's group. This research will help place empirical claims aboutsocial interactions on a much firmer basis than is currently the case. Second, the proposed research will integrate models of the effects of social interactions with models of group formation and analyze the theoretical and econometric implications of this integration. In contexts such as residential neighborhoods, social interactions are a primary determinant of group compositions and so a complete theory of inequality must account for this.The second part of the proposal will develop tools to facilitate the evaluation of alternative government policies. One theme will explore policy design when the policymaker does not know the true model of the phenomena he wishes to influence. Model uncertainty may occur because of uncertainty over general questions of the theoretical determinants of some outcome of interest, uncertainty about the way to quantitatively specify a theory, or because of uncertainty about issues of measurement and heterogeneity across units of observations, to name what seem to be particularly salient factors in social science. The analysis will address model uncertainty issues using model averaging methods that have recently been developed in statistics and have begun to receive application in economics. Finally, the proposal describes research to study fundamental limits in the design of government policies. Government policies are typicallychosen to affect both the mean and level of fluctuations of some outcome of interest. Within the control theory literature, there exist a number of results on frequency-specific tradeoffs that imply that any government policy that attempts to stabilize some outcome will inevitably reduce fluctuations at certain frequencies while exacerbating fluctuations at others. These tradeoffs do not appear to have been exploited in social science contexts. 2. Broader Impacts of the Proposed ActivityThe proposed research potentially has three broader impacts. First, it represents an effort to integrate sociological and economic perspectives on individual behavior and group outcomes. As such, it attempts to break down interdisciplinary barriers and may have general value in the social sciences. Second, the work may prove to be policy relevant. The analysis of model uncertainty will facilitate the development of econometric tools that are appropriate for the explicit evaluation of the effects of policies. Third, the analysis of design limits represents a new perspective on how government policies can stabilize dynamic socioeconomic outcomes at different frequencies, and how stabilization at different frequencies may be traded off. Finally, the authors plan to produce a graduate level monograph on social interactions that shouldfacilitate the training of students.