Collaborative Research: Dynamic Demand for New Durable Goods: An Empirical Model and Applications to Pricing and Welfare

在许多行业中，消费者不仅要选择购买什么，还要选择何时购买。然而，经济学家在这种情况下缺乏令人信服的总需求模型，许多论文在明显具有动态特征的行业中指定静态模型。为了解决这一问题，本研究扩展了先前的几条研究路线，以指定和估计为汇总数据设计的第一个消费者需求动态随机系数模型。本研究使用这种方法来分析两个可见的消费电子行业，数码相机和DVD播放器的数据集。本研究的主要焦点是开发新的方法来估计消费者需求的动态与汇总数据。利用这些方法，本文研究了这些行业中几个重要的应用问题。特别是，这项研究试图评估为什么这些行业的价格会大幅下跌，这是许多新的耐用品市场的普遍特征。虽然跨期价格歧视、进入和成本下降的典型解释早已得到承认，但以前没有工作对其相对重要性提供经验评价。本研究亦探讨这些产业的福利及物价指数。如果消费者不能处理随着时间的推移数量增加的情况，或者消费者利用当前的价格变化来预测最佳购买时间，那么标准价格指数就会出现问题。在提出的模型中自然考虑了这些特征，并且在考虑的数据集中预计是重要的。因此，本研究的另一个贡献是，它基于理论和经验上一致的模型，为这些问题提供了答案。这项研究有两个重要的更广泛的含义。首先，提出了一种估算消费者需求的新方法。实施该方法的计算机代码和算法将由作者在万维网上发布。所提出的方法可以被其他研究人员广泛应用于其他行业。其次，研究结果可能对政策制定者产生更广泛的重要影响。例如，科斯猜想表明，如果消费者知道未来价格会下跌，那么垄断者即使受到进入壁垒的保护，也可能几乎没有市场力量。对这一问题的实证分析可能对合并分析和竞争管理机构面临的其他问题很重要。研究如何评估价格指数是创建这些指数的政府机构非常感兴趣的，这些指数以各种各样的方式影响着经济。

In many industries, consumers choose not only what to purchase but when to purchase. However, economists lack compelling models of aggregate demand for this case and many papers specify static models in industries that clearly have dynamic features. In order to address this problem, this study extends several previous lines of research to specify and estimate the first dynamic random coefficients model of consumer demand designed for aggregate data. This study uses this method to analyze little-studied data sets on two visible consumer electronics industries, digital cameras and DVD players. The principal focus of this study is to develop new methods to estimate the dynamics of consumer demand with aggregate data.Using these methods, this work studies several important applied questions in these industries. In particular, this research seeks to evaluate why prices fall so dramatically in these industries, a ubiquitous feature of many new durable goods markets. While the typical explanations of intertemporal price discrimination, entry and cost declines have long been recognized, no previous work has provided an empirical evaluation of their relative importance. This study also investigates welfare and price indices in these industries. Standard price indices are problematic when consumers can delay purchase because indices do not handle situations in which quantity expands over time or when consumers use current price changes to predict the optimal purchase time. These features are naturally accounted for in the proposed model and are expected to be important in the data sets under consideration. Thus, another contribution of this study is that it provides answers to these questions, based on theoretically and empirically coherent models.There are two important broader implications of this study. First, a new methodology for estimating consumer demand is developed. Computer code and algorithms for implementing the methodology will be distributed on the World Wide Web by the authors. The proposed methods can be widely used by other researchers for other industries. Second, the results could have important broader implications for policy makers. For instance, the Coase conjecture suggests that if consumers know prices will fall in the future, a monopolist may have little market power even if it is protected by barriers to entry. Empirical analysis of this issue could be important for merger analysis and other questions faced by a competition authority. Research about how to evaluate price indices are of significant interest to government agencies that create these indices, and these indices impact the economy in a large variety of ways.