Prices in Space and Time

本提案旨在使用具有条形码或在线可获得的商品的价格数据来回答三个主要问题：商品价格和产品种类如何在不同空间变化？用线上价格代替线下价格来衡量通货膨胀有什么问题？价格和数量如何应对高频宏观经济冲击？为了回答这些问题，这个项目试图测量各个城市商品的确切价格指数。本项目首次尝试调查线上、线下和BLS价格指数之间差异的来源，并探索日常价格和消费数据如何应对宏观经济冲击。该项目将使用几个数据集。首先是美国的ACNielsen Homescan数据。其次是日经pos数据和ACNielsen Scantrak数据，涵盖了日本和许多国外的条形码零售销售数据。该项目将使用的第三个数据库有价格和点击通过大量零售产品的信息。这些数据加起来比任何经济学家或统计机构使用过的数据都要多得多。这将有助于构建每个国家数百万种产品的每日价格和需求信息，并将其与相同产品的线下销售数据进行比较。该项目旨在在多个方面取得突破。首先，对于学院派经济学家来说，拟议中的研究提供了第一个检验保罗？他提出的支撑其获得诺贝尔奖的新经济地理学理论的机制是正确的。克鲁格曼认为，更大的市场有更多的产品种类，应该有更低的可交易商品的价格指数。现有的跨城市价格衡量的一个主要问题是，它们没有比较相同的商品。因此，不可能知道在城市观察到的较高的商品价格是由于富裕的城市居民消费更高质量的商品还是由于相同商品的价格。更一般地说，该项目旨在展示如何使用条形码和在线房地产数据来衡量不同地区的生活成本。其次，该项目试图让我们了解基于互联网数据的价格指数在多大程度上是传统价格指数（如消费者价格指数）的良好替代品。经济学家和统计机构面临的一个主要问题是，网上价格与线下价格的追踪程度如何。美国劳工统计局（Bureau of Labor Statistics）等统计机构使用在线下手工收集的数据计算价格指数，这些数据无法反映频繁的网上价格变化，而这些变化与越来越多的消费者在网上购物有关。虽然越来越多的互联网数据的可用性有可能使这种类型的数据收集过时，但关于基于互联网数据的价格指数的准确性问题仍然存在。随着经济学家开始使用在线数据代替线下数据来调查许多重要的经济问题，了解在线价格与线下价格的总体密切程度是很重要的。作为起点，该项目试图研究不同市场的价格差异。虽然在某些狭义的市场（如书籍和隐形眼镜）已经做到了这一点？该项目将首次检验互联网价格是否普遍低于线下价格。该项目打算通过比较GPI数据和AC尼尔森的数据来做到这一点，AC尼尔森的数据包含了不同国家的大型商品商店（如沃尔玛）中750,000种不同UPC的价格信息。这种比较的一个额外好处是，该研究将能够验证点击率数据作为衡量市场份额的有效性。该项目的下一个目标是通过检查在线商家之间的价格变动来检查价格差异在不同类别的商家之间消散的速度（使用点击量数据控制其相关性）。这将有助于确定在线商家拥有多大的市场力量。该项目试图进行类似的分析，比较网上销售的商品与实体店销售的商品的价格动态，实体店销售的商品根据AC尼尔森的数据提供。生成的信息将有助于回答在线定价行为是否与零售价格有系统差异的问题。这是一个非常重要的问题，不仅对于理解市场互动，而且对于理解线上价格与线下价格的联系有多紧密。在某种程度上，这些价格是紧密相连的，这将进一步证明，人们可以用网上价格代替线下价格。这些信息还将使我们能够估计在线商店的发展对线下营销的影响。这个项目吗?S的目标也是利用数据来了解国际市场细分。特别是，通过比较不同国家的网上商品价格，这项研究将能够准确地估计国际成本冲击是如何在各国之间传递的（经济学家称之为“传递”）。这些类型的估计对于理解汇率变动如何影响价格和传播国际宏观经济冲击非常重要。最后，生成每日价格指数和观察每日需求的能力为评估定价和消费决策创造了许多令人兴奋的可能性。例如，如果仔细构建价格指数，理论上就有可能产生实时的每日通胀指标。这可能有助于研究宏观经济冲击对消费者价格的影响有多快。评估消费行为的困难之一是，消费和定价数据往往以季度或月度的频率出现。因此，通常很难确定哪个事件对决定定价和消费行为至关重要。每天处理价格和点击量数据的能力，将有助于理解宏观经济冲击是如何传递给需求、房地产和商品价格的。

Abstract This proposal aims to use price data for the universe of goods that have barcodes or are available online to answer three main questions: How do goods prices and product variety vary across space? What are the problems with using online prices as a substitute for offline prices to measure inflation? How do prices and quantities respond to high frequency macroeconomic shocks? In order to answer these questions, this project seeks to measure exact price indexes for goods across cities. This project is the first endeavor to investigate the sources of differences between online, offline, and BLS price indexes and to explore how daily price and consumption data respond to macroeconomic shocks.The project will make use of several datasets. The first is ACNielsen Homescan data for the US. The second is Nikkei-POS data and ACNielsen Scantrak data covering retail sales at the barcode level for Japan and a number of foreign countries. The third database that the project will use has price and click-through information for large number retail products. Jointly, this is vastly more data than has ever been used by any economist or statistical agency. This will enable to construct daily price and demand information for millions of products in each country and compare it to data on offline sales of the same products.The project aims to make breakthroughs in a number of dimensions. First, for academic economists, the proposed research provides the first test of whether Paul R. Krugman?s proposed mechanism that underlies his Nobel Prize winning theory of New Economic Geography is correct. Krugman argues that larger markets have greater product variety and should have lower price indexes for tradable goods. A major problem with existing cross-city measures of prices is that they do not compare identical goods. Hence, it is not possible to know whether, whether observed higher goods prices in cities are due to wealthier urban residents consuming higher quality items or due to prices of identical items. More generally, the project aims to demonstrate how barcode and online real estate data can be used to measure cost of living across locations. Second, the project seeks us to understand the extent to which price indexes based on internet data are good substitutes for conventional price indexes, such as the Consumer Price Index. A major question for economists and statistical agencies is how well online prices track offline prices. Statistical agencies such as the Bureau of Labor Statistics calculate price indexes using data manually collected at offline locations that cannot reflect the high frequency online price changes that are relevant to consumers who are increasingly shopping at online stores. While the increasing availability of internet data has the potential to render this type of data collection obsolete, questions remain regarding the accuracy of price indexes based on internet data. As economists start using online data in place of offline data to investigate many important economic questions, it is important to know how closely online prices track offline prices in general. As a starting point, the project seeks to examine price differences across markets. While this has been done for certain narrowly defined markets -- e.g., books and contact lenses ? this project will be the first to examine whether Internet prices are lower than offline prices in general. The project intends to do this by comparing GPI data with AC Nielsen data that contains price information on 750,000 different UPC's available in mass-merchandising stores like Wal-Mart for a variety of different countries. One of the added benefits of this comparison is that the research will be able to verify the validity of click-through data as a measure of market share.Next goal of the project is to examine how rapidly price differences dissipate across different classes of merchants by examining price movements across online merchants (controlling for their relevance using click-through data). This will be useful for determining how much market power online merchants have. The project seeks to conduct a similar analysis comparing price dynamics of goods available online with those available in brick-and-mortar establishments that sell goods available in AC Nielsen data. The generated information will be useful to answer the question of whether online pricing behavior differs systematically from retail prices. This is a very important question for understanding not only market interactions but also for understanding how closely online prices are linked to offline prices. To the extent that these prices are closely linked, it will provide further validation that one can use online prices as a substitute for offline prices. The information will also enable to obtain estimates for how the development of online stores affects offline marketing.The project?s goal is also to make use of data to understand international market segmentation. In particular, by comparing prices of online goods in different countries, this research will be able to estimate precisely how international cost shocks are transmitted across countries (what economists term ?pass-through?). These types of estimates are extremely important for understanding how exchange rate movements affect prices and propagate international macroeconomic shocks.Finally, the ability to produce daily price indexes and observe daily demand creates a number of exciting possibilities for evaluating pricing and consumption decisions. For example, if one constructs price indexes carefully, it is theoretically possible to produce daily measures of inflation in real time. This potentially could be useful for examining how fast macroeconomic shocks appear in consumer prices. One of the difficulties of evaluating consumption behavior is that consumption and pricing data tend to come out at a quarterly or monthly frequency. As a result, it is often very difficult to determine which event was critical in determining pricing and consumption behavior. The ability to work with price and click-through data at a daily frequency will help to understand how macroeconomic shocks are transmitted to demand as well as real estate and goods prices.