Data needs for consumer and retail firm studies.

Growing concentration in the retail grocery sector raises new economic questions that are difficult to answer with existing data sources. The data problems are due in large part to concentration in the retail data industry, where data are collected for commercial rather than academic research. Currently available grocery-level datasets are extremely expensive, are not properly randomized, and lack critical information.

To focus our discussion, we address data needs for industrial organization and marketing, nutrition and food safety, and government policy studies. The growing concentration at the grocery retail level raises a variety of industrial organization and marketing questions, such as: Has this greater concentration increased market power or changed the vertical relationship between manufacturers and other suppliers with retailers? Has the entry of low-price superstores fundamentally changed the services provided, the degree of product differentiation, the provision of private label products, and other actions by traditional supermarkets? What caused the mergers to occur?

Similarly, we want to know if greater concentration has affected the nation's nutrition and food safety, such as by making catastrophic food safety disasters more likely. Have increased product differentiation and lower prices from changes in retailing contributed substantially to alarming increases in rates of obesity?

Finally, we want to know how government rules and regulations have affected these markets and consumers. To protect consumers' health, the government has imposed restrictions on selling certain goods when food safety issues arise (e.g., mad cow disease and E. coli in lettuce and spinach). The government also provides nutritional and other label information (e.g., concerning health foods and organic foods) to help consumers make more informed food choices. What have been the effects of these laws and regulations on markets and on the health of various groups of consumers? We discuss the increase in concentration at the retail level, commercial databases, data needs for a number of important research areas, and possible solutions.

Concentration in Retail Markets

Grocery retailing markets are much more concentrated today than they were two decades ago. This increased concentration has altered the relationship between manufacturers and retailers. Although most existing empirical studies based on grocery scanner data implicitly presume that manufacturers set prices and retailers passively add on a competitive markup, there is substantial evidence (e.g., Villas-Boas) that such a description of the market is no longer true, if it ever was.

Mergers and acquisitions by large grocery retailers, including Kroger Co., Albertson's, Ahold USA, and Safeway, have significantly increased concentration ratios. Between 1997 and 2000, more than 4,100 U.S. supermarkets were acquired, representing $69 billion in sales. The four-firm concentration ratio (C4) increased from 16.6 percent in 1992 to 35.5 percent in 2005 (see figure 1). This trend toward increased concentration has continued with Supervalu's acquisition of third-ranked Albertson's in 2006 and the growth of Wal-Mart (Kaufman 2007).

[FIGURE 1 OMITTED]

Companies that were not involved in the food business two decades ago, such as Wal-Mart and Target, now account for a significant share of consumers' food-at-home expenditures. Since 1994, nontraditional food retailers (supercenters, warehouse clubs, mass merchandisers, drugstores, and dollar stores) have steadily increased their market share by about 28 percentage points to 31.6 percent in 2005. Led by Wal-Mart, most of this growth is attributed to supercenters that commanded 17.1 percent of the food-at-home retail markets in 2005 (Kaufman 2007).

It took Wal-Mart just four years of aggressive supercenter growth to become the largest U.S. grocery chain by 2002. Wal-Mart's large share is due to its relatively low prices, which are driven by scale economies and efficient operations based on buying directly from suppliers. Wal-Mart's approach has started a domino effect, significantly changing the retail food market's landscape. Warehouse club and mass-merchandisers have adopted this strategy, further intensifying price competition as more consumers have switched from shopping at supermarkets to low-price, large-scale operations.

Many supermarkets and other traditional grocery retailers have reacted by expanding their operations through merger and acquisition strategies, introducing a wider variety of new products (e.g., organic and natural foods, upgrade store brands, and convenience foods), promoting new store formats, introducing self-checkout stations, expanding frequent shopper card programs, and offering online home shopping services. Some researchers contend mergers and acquisitions are driven by a search for efficiencies associated with consolidation as supermarkets are increasingly pressured to meet price competition from non-traditional food retailers like Wal-Mart. Others contend that mergers increase the market power of supermarkets and increase prices for consumers.

Growing retail concentration has not only changed the nature of competition at the retail level, it has greatly affected the vertical relations along the marketing chain. As a result of the competitive pressures from Wal-Mart and other nontraditional formats, many firms in the grocery industry have resorted to what the industry refers to as efficient consumer response. These methods are designed to enhance timely, accurate, continuous, consistent flow of products that are matched to consumer demands. The initiative focuses on reengineering activities in the selection of product assortments, product replenishment, product promotions, and new product introductions. Information on the type and extent of these business practices are not readily available, thus impeding efforts to examine their impact on prices and consumer welfare. Further, many researchers believe that the now larger retail vendors are exercising their increased oligopsony power to lower prices paid to suppliers and increasingly charging manufacturers slotting fees, which are lump-sum fees for carrying a new product or continuing to carry an existing one.

Commercial DataBases

Agricultural economists have studied a variety of demand, health, marketing, and industrial organization questions using data from grocery chains or proprietary retail grocery scanner data. Stores' loyalty card datasets do not include detailed information on household demographics and are potentially subject to more measurement errors due to infrequent use of loyalty cards or use of someone else's card for convenience. Moreover, grocery chains rarely make their databases available to researchers.

Today, the only two major firms providing such scanner data are Information Resources, Inc. (IRI) and Nielsen (formerly known as AC-Nielsen). Their datasets are constructed primarily for marketing purposes and are used by retailers, manufacturers, and farm commodity groups. Usually, these firms charge researchers prices comparable to those they charge their commercial customers, so that a dataset covering only a few commodities for the most recent year may cost hundreds of thousands of dollars.

The current major point-of-sale or store scanner data sources are IRI's InfoScan and Nielsen's ScanTrack. Store scanner data are collected at cash registers, while household scanner data are obtained from a sample of households that scan their purchases after each shopping trip. Over the past ten years, IRI and Nielsen also have begun to track grocery purchases by specific households. Nielsen's household scanner dataset is Homescan and IRI's is Consumer Network. (Knowledge Networks is also developing a household-based scanner data panel.)

These datasets provide richer household demographic information than are available in store scanner data (Muth, Siegel, and Zhen 2007). Because IRI and Nielsen instruct the household scanner data panelists to scan all purchases from all outlets, the datasets from household-based scanner data are more complete than grocery datasets of purchases of individual households collected through loyalty card users.

In addition to being expensive, commercial datasets come with significant restrictions on how they may be used (e.g., brand market shares may not be reported) and do not provide all critical information needed for many important research topics. For example, although feasible, they do not have information on whether a specific low-income household is a Women, Infants, and Children (WIC) program participant, they do not provide any details on retailers' cost of operation (e.g., wholesale prices), and the household scanner databases tack prices of nonpurchased items for demand studies.

Because scanner data are proprietary and are not primarily designed for academic research, detailed documentation on sampling and data collection procedures and statistical properties of the data are not readily available. Although few academic papers that use IRI and Nielsen data discuss the quality of these datasets, there is good reason to question whether these firms use proper random sampling techniques. In the store-based scanner data, large, traditional supermarket chains are over-represented (because they supply data and hence are included with certainty, as opposed to smaller stores that are sampled). In addition, store-based scanner data may not adequately include new sources of food sales (Wal-Mart supercenters and other big box stores, and WIC-only stores).