Performing Department
Agricultural & Resource Econ
Non Technical Summary
The U.S. food retail landscape has important implications for farmers, consumers, and workers that extend beyond its direct economic importance. Yet previous research on its structure and conduct is somewhat dated, implemented primarily at the aggregate or national level, or descriptive. The proposed project aims to develop new critical knowledge about the evolving food retail entry and market structure at the establishment level and measure its impacts on competition and employment, using a novel dataset and modern economic modeling tools that provide richer and more reliable measures of these impacts in local food markets. We propose to accomplish this in four stages: (1) a comprehensive evaluation of entry, exit, and post-entry performance by retail formats and regions; (2) estimation of industrial organization models of entry and market structure, accounting for store heterogeneity, imperfect information, and dynamics, including entry costs; (3) impacts of these changes on incumbent sales and retail employment, with consideration to spatial competition, retail formats, and the COVID-19 pandemic; and (4) writing up and disseminating this new knowledge to academia, policy markets, and industry. A corollary impact is human capital development by training graduate students in two universities. The new knowledge created by this project will help inform federal policies aimed at promoting competition in the food supply chain and supporting employment development in the evolving food retail landscape.
Animal Health Component
0%
Research Effort Categories
Basic
10%
Applied
80%
Developmental
10%
Goals / Objectives
The proposal has the following goals/objectives:Objective 1: Measure and examine the patterns of entry and exit in food retailing, considering the heterogeneity of establishments over time and space. We will quantify patterns of firm entry and exit by types of stores and measure post-entry performance and the persistence of firm entry and exit patterns.Objective 2: Ascertain the implications of food retail entry for the competition in food and beverage retail markets. We will utilize entry data to develop and examine a model of entry and competition in non-metro U.S. food markets to determine threshold levels of entry to attain competitive behavior.Objective 3: Quantify the impact of food retail entry and market structure on sales performance for various types of retail grocery stores. We will develop and implement a model of entry based on panel data of firms operating in food retail markets with particular attention to the location and types of competitors over time and space.Objective 4: Assess the consequences of changing food retail entry and market structure for employment. We will develop and estimate a model of employment by retail formats to ascertain the impact of entry on food retail employment over time.
Project Methods
We will first compile a novel establishment-level database to describe the dynamics of food retailing. Our primary data source will be the NETS database, along with covariates from other data sources such as county-level information from the County Business Patterns , employment. Since our NETS access will be for the next five years with a two-year lag (1990-2023), we will benefit from access to data collected during the COVID-19 pandemic. Therefore, our statistical models will include the effects of the pandemic in the set of dependent variables. We will exploit the geographic variation of local COVID cases and deaths to capture possible effects of COVID-19 on entry, exit, competition, and employment. We will do this in several steps:1. Measure and examine the patterns of entry and exit in food retailing, considering the heterogeneity of establishments over time and space.The focus, dates, and breath of our descriptive analysis expand significantly beyond what Stevens et al. (2021) did in their recent ERS report. In particular, we will examine entry, exit, the persistence of entry, and post-entry performance, including the COVID-19 pandemic, which they did not address since their analysis ended in 2015. In addition, we will decompose employment productivity by retail formats and entrants, incumbents, and exiting establishments.We will adopt NAICS 445 (food and beverage retailing in a fixed location) as the broad definition of the industry boundary. For the geographic boundary, we will follow Rossi-Hansberg, Sarte, and Thracher (2018) and define a market as consisting of a ZIP code or a county.The second step is to obtain a typology of retail formats. The U.S. Census reports food retail groups in NAICS 445 as consisting of three broad groups: grocery (except convenience stores), convenience stores (except those with fuel pumps), and specialty stores. Then we will identfy: (1)Types of entrants and their relative importance in terms of size and numbers; (2) the persistence of entry and exit patterns, and (3) post-entry performance of new entrants, based on market shares and average size.2. Ascertain the implications of entry for the competition in food retail markets. To quantify the effects of market entry on retail competition, we rely on the framework initially proposed by Bresnahan and Reiss (1991, herein BR) and the substantial literature that has followed (see for a comprehensive review, Berry and Reiss, 2007). The model utilizes information on market structures to empirically establish entry thresholds and then uses these thresholds to assess differences in markups and competition in a given industry or market. Thus, we will first follow Berry and Reiss (1991) to ascertain how profit margins of incumbents change with changes in the number of competitors.We plan to relax the assumptions of the basic model in several directions. First, we will relax the homogeneous firm restriction and introduce market, variable cost, and fixed cost shifters that reflect the "true" retail market structure as reflected by the mix of formats. Second, we will explore models of incomplete information (Seim, 2000, and follow-up work). Third, we also plan to extend the basic static model to account for dynamic folloding Aguirregaberria and Mira (2007), Baiari, Benkard and Leving (2007), and Pakes, Ostrovsky, and Berry (2007).3. Quantify the impact of food retail entry and market structure on sales performance for various types of retail grocery stores.It is essential to examine the location of competitors relative to an individual incumbent grocery store rather than relying on a summary index, like the HHI, for a given market area. Thus, we will consider the spatial dispersion of competitors in each market area, in addition to the types of retailers (e.g., Walmart or a Dollar store). Both elements have been ignored in the food retail literature so far. We will follow the standard practice in the industrial organization literature and focus on clearly identifiable markets by considering individual markets in rural and isolated areas (Berry and Reiss, 2007) rather than addressing all establishments in the NETS dataset. Unlike previous industrial organization research that focuses on narrow case studies of market structure, and which is mostly centered on entry in specific geographic markets,this project is national in scope, relying on panel identification of parameters with a significantly larger dataset. In addition, we observe sales for each establishment and, thus, are not operating with limited data.For manageability, our approach is to use a parsimonious model by simplifying Mazzeo's (2002) model, which extends the Berry and Weiss (1992) model to consider different types of competitors in lieu of quality product types.4.Assess the consequences of changing food retail entry and market structure for employment.The market boundaries for labor markets are distinct from those for products or food retail markets. On the one hand, labor markets can be substantially larger than the county or zip code of the retail outlet. On the other hand, employees in grocery stores could work for other retail outlets, but also FedEx or Amazon Fulfillment Centers for that matter. Following previous work, we adopt the spatial concept of the commuting zone to analyze the implications of food retail entry on labor market outcomes (USDA, 2019). Since this spatial concept relates closely to that of the U.S. counties, the collection of covariates is greatly facilitated. We are particularly interested in the effects of entry and market structure on employment by alternative retail formats. To this end, we will build and estimate an employment model follwoing some of the Walmart literature; in particular, Basker (2005) and Newmark et al. (2008) for the impact of entry of entry of new retail formats on retail employment.