Food CPI, Prices, and Expenditures:
A Workshop on the Use of Scanner Data in Policy Analysis

Overview

As markets become more segmented and contracts replace spot market transactions, the declining volume of available data associated with spot transactions becomes less representative and therefore less useful for research. As a result, researchers are increasingly turning to retail scanner data to decipher market workings. Not only are such data plentiful (although expensive), but with links to demographics of individual households, the data provide a window on distributional issues. The voluminous quantity of the data, while an asset, can also present researchers with special challenges.

On June 9, 2003, the Economic Research Service and the Farm Foundation conducted a workshop on use of household scanner data in food policy analysis. The workshop provided a forum for participants to discuss strategies in using scanner data. Panels and presentations were led by representatives from the Federal Government and higher education.

The workshop kicked off with six presentations on ongoing research using scanner data. Topics included demand issues related to nutritionally enhanced foods, nonalcoholic beverages, and fruits and vegetables. Presentations also addressed the effects of supermarket promotions on consumer purchase decisions and the use of demand estimation for policy simulation.

Addressing Policy Issues Using Micro Scanner Data Moderator: Harry Kaiser, Cornell University
Helen Jensen, Iowa State University	Demand for Enhanced Foods and the Value of Nutritional Enhancements of Food
James Binkley, Purdue University	The Demand for Functional Foods and Identifying Marketing Strategies of These Foods
Cesar Costantino, University of Maryland	Consumer Search Inside the Supermarket
Oral Capps, Texas A&M University	Demand Projections Segmented by Income for the Highly Competitive Non-Alcoholic Beverage Complex Using the A.C. Nielsen HomeScan Panel Data
Steven Yen, University of Tennessee	Demand for Fruits and Vegetables: An Analysis of HomeScan Data
Jeff Perloff, University of California, Berkeley	Use of Demand Estimation for Policy Simulation

Scanner data allow for a very accurate representation of items purchased by consumers, and this additional information can be used when calculating price indices and price changes, a topic addressed in a panel discussion during the workshop’s second morning session. Panel members also examined the costs and benefits of incorporating this information.

Panel Discussion: Scanner Data and Price Indices Moderator: Ephraim Leibtag, Economic Research Service
Mick Silver, Cardiff University	Scanner Data and Price Indices
Walter Lane, Bureau of Labor Statistics	Uses of Point of Sale Scanner Data at BLS
Marshall Reinsdorf, Bureau of Economic Analysis	Roundtable Discussion

Afternoon sessions focused on methodological challenges encountered by researchers when using scanner data, which are voluminous and are recorded at a high frequency. One presentation addressed organizational designs that researchers face when purchasing scanner data. Another discussed the consequences of conducting economic research using scanner data in light of the fact that most price variation is generated from sales, which cause dramatic increases in quantities purchased. The volume of scanner data can also make it necessary to aggregate the data over some dimension. The workshop addressed statistical methods to determine the appropriate aggregation of scanner data and the consequences of different levels of aggregation.

Methodological and Data Challenges of Using Scanner Data Moderator: David Davis, Economic Research Service
Mike Harris, Economic Research Service	Properties of Scanner Data
Dan Hosken, Federal Trade Commission	The Importance of Sales in High Frequency Supermarket Scanner Data
Oral Capps, Texas A&M University	Aggregating Scanner Data

The workshop’s final session examined methodologies in conducting econometric estimation with scanner data. Discussions focused on empirical models of industrial organization and issues faced when conducting hedonic regressions using scanner data.

Estimation Using Scanner Data
Tirtha Dhar, while at the University of Wisconsin	Estimation Using Scanner Data
Mick Silver, Cardiff University	Estimation Using Scanner Data

Summaries of the papers can be found on the Farm Foundation website.