CAREER: Empirical Studies of Dynamics in Industrial Organization

本提案概述了实证研究和教学议程，以分析与产业组织研究相关的几个动态现象的经济学。这三个研究项目和教育活动研究旨在提高我们对所涉及的具体行业的理解，更广泛地说，有助于评估一些经济理论的相关性。前两个项目分别构建和估计消费者选择的动态模型，以研究临时降价（“销售”）和优惠券的影响。第三个模型使用行业动态的多智能体模型研究品牌引入。这个项目的教育部分提出了几个与研究项目相关的活动。第一个项目侧重于暂时降价，即所谓的“销售”，这是许多行业的常见做法。对销售有几种可能的解释，但这里的重点是跨期价格歧视。更具体地说，当消费者有一个预期的未来需求流，能够以成本存储一种消费商品，并面临价格变动时，PI建模了消费者的动态问题。在这个模型中，消费者会购买产品，或者是因为她想在当前期间消费，或者是因为她认为当前价格与预期未来价格之间的差额低于存储库存的成本。就后一种影响是重要的而言，它对在合并分析、新商品估值以及其他政策相关问题中使用从静态需求模型计算的需求弹性有影响。PI使用1000户家庭在104周内购买24种产品的数据，以检验模型隐含的相关性，并直接估计模型。第二个项目的重点是优惠券，经济学家通常将其视为价格歧视工具的教科书范例。在最近的一篇论文中，Nevo和Wolfram（1999）提出，为了充分理解息票的经济学，需要一个比标准静态垄断理论更丰富的模型。他们还发现一些证据表明，优惠券的模式符合：(1)寡头垄断环境下的价格歧视模型，该模型表明优惠券主要用于诱导从竞争品牌切换；(2)使用优惠券促进再购买。为了进一步研究这些问题，PI构建了一个考虑品牌忠诚度的消费者选择的动态模型。该模型是使用Nevo和Wolfram（1999）中使用的优惠券数据和第一个项目中描述的家庭购买数据的组合来估计的。第三个项目侧重于企业动态。在之前的工作中，PI使用静态竞争模型研究了即食谷物行业的价格竞争和合并（Nevo, 2000a, b），将公司提供的品牌作为给定的。然而，品牌引入是这个行业的一个重要维度，在其他行业也是如此。在这个项目中，PI分析了一个多智能体动态模型，在这个模型中，企业可以通过投资引入新品牌，停止使用旧品牌，并影响现有品牌的感知质量。这种投资可以看作是广告，产生的结果是随机的。PI检验了模型的马尔可夫完全均衡。在不同的假设下，模型的预测可以与实际的品牌导入进行比较，以便更好地了解品牌导入过程。此外，该模型还可用于模拟不同情景下的产业结构。后者对于反垄断政策尤其重要，因为在这个时代，人们关注的似乎是创新，而不是价格垄断。在教育方面，PI开发了研究生课程和学生工作坊，这两者都将训练学生使用上述方法，并更广泛地帮助他们撰写产业组织的论文。最后，该提案展示了拟议的研究如何也可以在本科水平的教学和建议。

This proposal outlines an empirical research and teaching agenda for analyzing the economics of several dynamic phenomena, relevant to the study of Industrial Organization. The three research projects and educational activities studies are intended to improve our understanding of the specific industries involved and, more broadly, to help assess the relevance of a number of economic theories. The first two projects construct and estimate a dynamic model of consumer choice in order to study the effects of temporary prices reductions ("sales") and coupons, respectively. The third model studies brand introduction using a multi-agent model of industry dynamics. The educational component of this project proposes several activities that connect to the research projects. The first project focuses on temporal price reductions, so called "sales", which are a common practice in many industries. There are several potential explanations of sales, but here the focus is on intertemporal price discrimination. More specifically, the PI models the consumer's dynamic problem when she has an expected stream of future demands, is able to store a consumption good, at a cost, and faces price movements. In this model, the consumer will purchase the product either because she wants to consume in the current period or because she believes the difference between the current price and the expected future price is lower than the cost of storing the inventory. To the extent that the latter effect is important, it has implications for the use of demand elasticities, computed from a static demand model, in merger analysis, valuation of new goods, as well as other policy relevant issues. The PI uses data on the purchases of 1,000 households of 24 products over 104 weeks, in order to examine correlations implied by the model and to estimate the model directly. The focus of the second project is coupons, which economists generally consider as a textbook example of a price discrimination tool. In a recent paper Nevo and Wolfram (1999) suggest that in order to fully understand the economics of coupons, a model richer than the standard static monopoly theory is needed. They also find some evidence that the patterns of couponing are consistent with: (1) models of price discrimination in oligopoly settings, which suggest that coupons are mainly used to induce switching from competing brands, and (2) the use of coupons to promote repurchase. In order to examine these issues further, the PI constructs a dynamic model of consumer choice allowing for brand loyalty. The model is estimated using a combination of the coupon data used in Nevo and Wolfram (1999) and the household purchase data described in the first project. The third project focuses on firm dynamics. In previous work the PI studied price competition and mergers in the ready-to-eat cereal industry (Nevo, 2000a, b) using static models of competition, taking as given the brands offered by the firms. However, brand introduction is an important dimension in this industry, as well as in other industries. In this project the PI analyzes a multi-agent dynamic model, in which firms can introduce new brands, discontinue old brands and impact the perceived quality of their existing brands by investing. The investment, which can be thought of as advertising, yields a random outcome. The PI examines the Markov Perfect equilibrium of the model. The predictions of the model, under different assumptions, can be compared to actual brand introduction in order to better understand the brand introduction process. Furthermore, the model can be used to simulate the industry structure under different scenarios. The latter is particularly important to anti-trust policy in an era when innovation, rather than price fixing, seems to be the focus. In terms of education, the PI develops a graduate curriculum and student workshop, both of which will train students in the above methods, and more broadly assist them in writing a thesis in industrial organization. Finally, the proposal demonstrates how the proposed research can also feed into teaching and advising at the undergraduate level.