Discrete Choice and Hedonics: Identification, Estimation, and Application to Housing and New Goods

这项研究项目有两个主要目标。首先，该项目导出了识别和估计随机系数离散选择模型的新方法，特别关注具有有限维误差项的离散选择模型，该模型有时被称为“纯特征”模型。这些模型是分析政策变化影响的更好的经济模型。其次，这些新方法被应用于两个重要的实证应用：住房选择和新住房政策的分析，以及个人计算机最近的创新给个人带来的福利收益的衡量。这些计量方法除了改善了模型的理论经济性质外，还有两个明显优于现有方法的优势。该项目开发了一种新的识别策略，使用了享乐定价方程，它不需要最近文献中广泛使用的工具。为了实现识别，最近的文献通常假设可观察的产品特征与不可观察的产品特征不相关。这样的假设与长期均衡相冲突，在长期均衡中，企业根据竞争对手的产品特征来最佳地选择其产品的特征。由于定价方程被用来识别未观察到的产品特征，因此它可以用于个体效用最大化，从而非参数地识别个体品味系数。个人随机系数的恢复很重要，因为它允许研究人员评估政策变化的分配后果，也因为它提供了一个更丰富的总需求体系。这些技术被用来评估两个行业的重要实证问题。第一个被考虑的行业是住房。具体地说，扩展了对Bajari和Kahn(2000)的分析，以解决为什么种族隔离仍然存在的问题。这项计划亦会研究现行的房屋政策，特别是在何处兴建新的公共房屋单位的问题。其次，个人电脑创新所带来的个人福利是在过去十年中衡量的。我们的方法非常适合这个问题，因为它强调个体的异质性。特别是，调查人员能够查明消费者福利收益的来源(例如，产品质量与产品价格)和福利收益在不同个人群体之间的分配，评估这些问题是对“数字鸿沟”形式和程度的精确衡量。调查人员还重新探讨了如何为计算机等质量变化迅速的行业构建理想的价格指数这一重要问题。

There are two primary goals for this research project. First, the project derives new methods for iden-tification and estimation of random coefficient discrete choice models, focusing particularly on discrete choice models with a finite dimensional error term sometimes called "pure hedonic" models. Such models are superior economic models with which to an-alyze the effects of a policy change. Second, these new methods are applied to two important empirical applications: an analysis of housing choice and new housing policy, and measurement of the welfare gains to individuals from recent innovation in personal computers.Besides improving the theoretical economic properties of the model, these econometric methods have two distinct advantages over existing methods. The project develops a new identification strategy, using a hedonic pricing equation, which does not require the instruments that have been widely used in the recent literature. In order to achieve identification, the recent literature has typically made the assumption that observable product characteristics are uncorrelated with unobserv-able product characteristics. Such an assumption is in conflict with a long run equilibrium in which firms optimally choose their products' characteristics conditional on the characteristics of their rivals' products. Because the pricing equation is used to identify unobserved product characteristics, it can then be used for individual utility maximization to nonparametrically identify individual taste coefficients. Recovery of individual random coefficients is important because it allows the researcher to evaluate the distributional consequences of a policy change, and because it provides for a richer aggregate demand system.These techniques are used to evaluate important empirical questions in two industries. The first industry considered is housing. Specifically, the analysis of Bajari and Kahn (2000) is extended to address the question of why racial segregation persists. The project also studies current housing policy, specifically the question of where to locate new public housing units. Second, individual welfare from innovation in personal computers is measured over the last decade. Our approach is well suited to this problem because of its emphasis on individual heterogeneity. In particular, the investigators are able to break out the source of consumer welfare gains (e.g., product quality versus product price) and the distribution of welfare gains across different groups of individuals, evaluating such questions as a precise measurement of the form and extent of the "digital divide". The investigators also revisit the important issue of how to construct an ideal price index for an industry with rapid quality change like computers.