Recipient Organization
CAL POLY CORPORATION
(N/A)
SAN LUIS OBISPO,CA 93407
Performing Department
(N/A)
Non Technical Summary
The COVID-19 pandemic brought several structural issues facing the US food supply chain into sharp relief. The labor force in the food system, from farms to supermarkets and all the interconnected segments in between, was a highly publicized constraint in getting food from the ground and onto plates. News stories in the popular press and trade publications regularly discussed the ongoing labor issues throughout the food system, covering factors such as high rates of turnover (Farm Bureau, 2022), shutdowns due to COVID-19 infections (AP News, 2021), wage inflation (AgAmerica, 2022), the loss of institutional knowledge (TSI, 2020; US Chamber of Commerce, 2023), and the difficulties in attracting young workers (CropTracker, 2022). A Google trends analysis for the term "labor shortage" readily reveals that, even among the public, concern and awareness for this issue has increased substantially since 2020.Despite the established importance and urgency of this issue by industry, government agencies, and increasingly the academic community, there is little published quantitative work addressing the issue. Available estimates of labor shortages, turnover rates, wage variation, and other metrics are predominantly anecdotal, which hampers investigating how these measures have changed in recent years. Understanding these measures in the context of a comprehensive and detailed demographic profile of the food and agriculture labor force can help the food system fully realize the quantifiable benefits of a diverse workforce (Roberson, 2019). Additionally, the US food industry, from farms (MacDonald 2020) to supermarkets (Zeballos et al., 2023, Hendrickson et al. 2021) and most sectors in between (MacDonald et al., 2023), has been defined by increased consolidation in recent decades. However, despite the known impacts of consolidation in other retail industries on the labor force (Rabbani and Raj, 2023, Guanziroli 2022), the labor force impacts of consolidation in the food system are not well understood and may be vital to understanding ongoing labor issues. This gap in knowledge is particularly apparent when considering how the impacts transcend an individual segment of the food supply chain. Medium- and long-term solutions have been proposed to address labor issues in the food supply chain (Luckstead et al. 2021, Larue 2020), but we are not aware of any work that has assessed the potential costs or benefits of these solutions. The proposed project addresses each of these knowledge and information gaps through four related, but individual objectives. Investigating these objectives will develop insights for industrial, policy, and academic audiences.
Animal Health Component
60%
Research Effort Categories
Basic
40%
Applied
60%
Developmental
(N/A)
Goals / Objectives
Objective R1: Develop generalized empirical facts about the evolution of labor in the food supply chain, 2000-present, focusing on geographic and demographic patterns by industrial sector.This objective serves as the foundation for the project and the remaining objectives. The primary output of R1 is a database of metrics, associations, and additional demographic and market structure variables that is expected to result in multiple papers, and to inform and support policy and research on employment beyond the duration of the project. To the best of our knowledge, these metrics have not been studied or discussed in published research for most industrial sectors in the food system.To facilitate comparisons across geographic areas that vary substantially in population size and density, our focus will be on employment metrics that are not direct functions of workforce size. We describe each metric of interest, show their formulas as defined by U.S. Census Bureau (2006) where applicable, and explain their intended use in filling the gaps in extant knowledge. All QWI measures pertain to time t (quarter or year), industry i, and county c, and subscripts are suppressed. The separation rate is the flipside of the hiring rate and measures the rate at which workers are leaving their jobs. Often referred to as the "quit rate" in the literature, economists have long studied the causes and consequences of this metric. It has been identified as a determinant of firm performance and, in some cases, has been used as a proxy for overall performance (Addison and Bellfield, 2001; Schmidt et al., 2018). As an example of an insight the research project will produce, R1 will yield the dispersion of separation rates by industry and demographics during COVID-19, to aid in the understanding of how the pandemic affected labor force participation.Turnover has been cited as a major cost across many industries, via search and onboarding costs and the loss of skill and institutional knowledge (Waldman et al., 2010; Davidson et al., 2010; Cosar et al., 2016; Nouri and Parker, 2020). Quantifying turnover throughout the food supply chain will allow the research team to identify demographics and other determinants of turnover, and in turn to estimate the impacts of turnover on output, productivity, and food prices in R3.The replacement hiring rate measures the share of separations that are being replaced. This is not to be confused with the unemployment insurance replacement rate. Our measurement of the replacement rate is positively associated with firm performance, as it serves as a counterbalance to separation rates and turnover (Juranek et al., 2020). The complete statistical portrait of the labor force in an industrial sector should include the replacement rate to fully understand the dynamics of the labor force, and to understand where losses in some demographic groups might be offset by gains in others.Objective R2: Identify the impacts of market concentration and mergers and acquisitions (MA) on the labor force in the industry, including but not limited to metrics such as wages, total employment, and turnover.To estimate market concentration and changes therein for the grocery and foodservice sectors, we will rely primarily on the Nielsen TDLinx and the National Establishment Time Series (NETS) datasets, respectively. Collaborator Marchesi is an expert at using both datasets and will ensure access for the research team for the years 2004-2021 (and later as they become available) during the duration of the project. These datasets consist of store-level observations for supermarkets and restaurants, including precise location, ownership structure, and categorical annual revenues. The PD and co-PD have published multiple papers using TDLinx data to estimate retail market concentration (Cai et al., 2018; Rahkovsky and Volpe, 2018; Volpe and Cho, 2019; Ma et al., 2019). The primary metric for measuring concentration is the Herfindahl-Hirschman Index (HHI).We also rely on Nielsen TDLinx data to measure the impact of MA events. As mentioned previously, the research team has developed a list of MA events in retail, wholesaling, manufacturing, and distribution occurring between 2008 and 2022 using data from the Food Institute. Using these events and Circana Retail Panel data we can group stores into three groups; those that were involved in an MA event, those that were competitors with a store involved in and MA event, and those stores unrelated or unaffected by the MA event. The panel nature of the employment measures, such as wages, total employment, and turnover and the grouping of stores in the TDLinx data facilitate a difference in difference approach to identifying the causal impact of MA events on employment indicators throughout the food supply chain. We plan to estimate a regression equation, with variations informed by preliminary results.Objective R3: Estimate the impacts of employment dynamics, including hirings, separations, wages, and turnover, on productivity and retail food prices in the U.S.Following a similar approach to Gunderson et al. (2016), we calculate the cost of both a TFP basket and a "low-cost" TFP basket. The TFP basket price uses all available products at a store in each week in the calculation, whereas the "low-cost" TFP basket price only considers products in the lowest decile of price. We make this distinction because low-income households tend to purchase cheaper goods. Including all goods may inflate the realized price of the TFP basket. Therefore, we calculate the TFP basket price for each store andweek.This median price is calculated using all items in the TFP group i at store j in week k for the TFP basket price and using items in the first decile of items in TFP group i at store j in week k for the "low-cost" TFP basket price. The Circana Retail Panel data from 2012 to 2021 are used to calculate the TFP basket prices for all stores in the sample. The expectation is that the sample will continue to be available throughout the project years, so we can continue to calculate TFP basket prices for future years relevant to our proposed project. Using these prices, we plan to employ a similar empirical approach to R2.Objective R4: Undertake a cost/benefit analysis of proposed solutions to labor shortages in the food system, particularly for production agriculture. With the statistical foundation of R1 and the impact findings of R2 and R3, R4 is the culmination of the proposed project. The first step of this objective is to compile proposed solutions to labor challenges throughout the food supply chain. This will be done through a literature review of primarily industry and government publications, as well as conversations with industry partners and stakeholders in IFPA, FMI, NGA, and CGA. These trade associations include member companies across nearly all industrial sectors of interest. Much of the conversation to date has focused on the production sector. As an example, Calvin et al. (2022) summarizes the USDA's perspective on addressing labor shortages in specialty crop production, and identifies change the mix of domestic commodity production, increasing automation, and expanding the H-2A visa program while decreasing program costs. We intend to investigate the potential costs and benefits of each.
Project Methods
Objective R1: Develop generalized empirical facts about the evolution of labor in the food supply chain, 2000-present, focusing on geographic and demographic patterns by industrial sector.This objective serves as the foundation for the project and the remaining objectives. The primary output of R1 is a database of metrics, associations, and additional demographic and market structure variables that is expected to result in multiple papers, and to inform and support policy and research on employment beyond the duration of the project. To the best of our knowledge, these metrics have not been studied or discussed in published research for most industrial sectors in the food system.The hiring rate measures hires as a percent of average employment. The hiring rate has been shown to contain forward-looking information about firms and industries that is not explained or predicted using traditional measures, such as investment rates (Belo et al., 2014). It serves to illustrate variation in the fraction of workers entering new jobs and will provide insights into which sectors of the food supply chain are attracting the new workers at the fastest rate, and how this has changed over time.Turnover has been cited as a major cost across many industries, via search and onboarding costs and the loss of skill and institutional knowledge (Waldman et al., 2010; Davidson et al., 2010; Cosar et al., 2016; Nouri and Parker, 2020). Quantifying turnover throughout the food supply chain will allow the research team to identify demographics and other determinants of turnover, and in turn to estimate the impacts of turnover on output, productivity, and food prices in R3.Objective R2: Identify the impacts of market concentration and mergers and acquisitions (MA) on the labor force in the industry, including but not limited to metrics such as wages, total employment, and turnover.To estimate market concentration and changes therein for the grocery and foodservice sectors, we will rely primarily on the Nielsen TDLinx and the National Establishment Time Series (NETS) datasets, respectively. Collaborator Marchesi is an expert at using both datasets and will ensure access for the research team for the years 2004-2021 (and later as they become available) during the duration of the project. These datasets consist of store-level observations for supermarkets and restaurants, including precise location, ownership structure, and categorical annual revenues.We will also experiment with and consider other options such as the Comprehensive Concentration Index, which has the advantage of accounting for the total number of smaller firms in a market (Plastun et al., 2018), and may be salient in our case given the potential role of independent supermarkets in urban markets. We will consider multiple definitions of a market based on population size, density, and driving distance to food retail events for robustness. Collaborator Scharadin has contributed to a line of research showing that significant differences may exist as the market boundary definition varies, particularly for rural areas (Scharadin et al. 2022).We also rely on Nielsen TDLinx data to measure the impact of MA events. As mentioned previously, the research team has developed a list of MA events in retail, wholesaling, manufacturing, and distribution occurring between 2008 and 2022 using data from the Food Institute. Using these events and Circana Retail Panel data we can group stores into three groups; those that were involved in an MA eveObjective R3: Estimate the impacts of employment dynamics, including hirings, separations, wages, and turnover, on productivity and retail food prices in the U.S.Similar to market concentration in R2, the first step in understanding the impact of labor market dynamics on agricultural production and food prices is operationalizing these terms. Measuring price impacts requires the creation and use of robust and representative price measures, which is a computationally intensive component of the proposed research. Past studies have used individual product prices, e.g., mean price of selected products over a given time period (Allain et al., 2013; Huang and Stiegert, 2010) or calculated a price index weighted by the share of each product in the average consumer's market (Aalto-Setala, 2002; Barros et al., 2006). Both appropriations have major limitations. Choosing one product provides a very narrow view of the impact and basing a price index of consumer baskets may underestimate price increases if consumers substitute to cheaper goods in response to a price increase. Therefore, we calculate the price to meet the TFP because it provides multiple methodological and application advantages, but two primary advantages of this approach are that this:Considers the prices of all goods in each market and weights the price index by TFP suggested allocations, which tracks the price of a consistent basket of goods across time.Uses a well-established price measure in the diet quality literature (Christensen and Bronchetti, 2020; Crockett et al., 1992) and is directly related to federal policy.The traditional approach to measuring labor productivity is to estimate equations, typically translog, based on the Cobb-Douglas production function. Ratchford (2003), Restuccia et al. (2008), Wahdat and Lusk (2023), and many others have applied variations of this Cobb-Douglas estimation to measure productivity in food industrial sectors. Cosic and Steuerle (2021) used QWI data in this setting. The Bureau of Labor Statistics (BLS) maintains the Detailed Industry Productivity database, which is ideal for our purposes and allows us to measure factors affecting labor productivity in the food system directly in a reduced-form setting.Objective R4: Undertake a cost/benefit analysis of proposed solutions to labor shortages in the food system, particularly for production agriculture. With the statistical foundation of R1 and the impact findings of R2 and R3, R4 is the culmination of the proposed project. The first step of this objective is to compile proposed solutions to labor challenges throughout the food supply chain. This will be done through a literature review of primarily industry and government publications, as well as conversations with industry partners and stakeholders in IFPA, FMI, NGA, and CGA. These trade associations include member companies across nearly all industrial sectors of interest. Much of the conversation to date has focused on the production sector. As an example, Calvin et al. (2022) summarizes the USDA's perspective on addressing labor shortages in specialty crop production, and identifies change the mix of domestic commodity production, increasing automation, and expanding the H-2A visa program while decreasing program costs. We intend to investigate the potential costs and benefits of each.Another major component of R4 will be a review of the literature and industry conversations to understand and quantify costs, and in some cases benefits, that cannot be estimated via the proposed project. As examples, Kuhn and Yu (2021) estimate that a 10% increase in retail turnover is equivalent to a 0.6% wage increase, while Wahdat and Lusk (2023) identify the downstream impacts of labor impacts in the food supply chain, which transcend individual sectors. We expect that most, if not all, of our estimations of costs and benefits will be based on back-of-the envelope calculations, and there is a long literature on these calculations specific to agriculture. For example, Lowenberg-DeBoer et al. (2020) reviewed 18 studies on the economics of automation in field crop production. Most of the reviewed studies are based on calculations that would be refined based on our findings in R1, R2, and R3.