key: cord-1031301-r6g384l5
authors: Widmar, Nicole Olynk; Thompson, Nathanael M.; Bir, Courtney; Nuworsu, Eugene Kwaku Mawutor
title: Perception versus reality of the COVID-19 pandemic in U.S. meat markets
date: 2022-04-03
journal: Meat Sci
DOI: 10.1016/j.meatsci.2022.108812
sha: 18d195643fe865cd1753159c77cae7dce219de35
doc_id: 1031301
cord_uid: r6g384l5

Disruptions to meat markets during the COVID-19 pandemic spurred mass media attention. While media deeming the U.S. food system ‘broken’ garnered a great deal of attention, the actual production and meat availability data does not support this conclusion. The U.S. meat supply chain, while certainly strained and with measurable consequence during periods of adjustment, proved ultimately resilient and rebounded quickly. Increased attention on meat supply chains may drive continued efforts to improve resiliency, but analyses of online media and U.S. production and cold storage data do not support a narrative that the system ‘broke’, but was perhaps ‘strained’ and ‘responded efficiently’. Findings indicate that public sentiment about U.S. meat supply overall was not as dominated by pandemic-era concerns as may be hypothesized.

The World is interconnected both physically and virtually; webs of travel routes facilitate the movement of goods and people, and the World Wide Web spreads news and information contemporaneously. The COVID-19 pandemic is a testament to the level of our interconnectedness, physically and virtually, as both the virus and information (both correct and incorrect) has spread around the globe at rates previously unrecognized in conjunction with widespread disaster. The novel coronavirus was reported in Wuhan China on December 31, 2019, when Chinese authorities confirmed they were treating dozens of cases.

What started as a health emergency has become an economic emergency for nations around the globe (Ozili and Arun, 2020). Households faced economic and social challenges as school cancellations left parents torn between child care and economic survival. COVID-19 has been tragic for households impacted medically and economically, while instigating adjustments in eating behaviors due to restaurant closures, less spending on food away from home, and fewer shopping trips for many people seeking to limit exposure. Supply chain adjustments also impacted short-term and/or local availability of specific products, which was especially true of perishable food items, such as dairy (Wolf et al., 2021) and meat (Balagtas and Cooper, 2021) . Grocery stores struggled to keep up with rapid increases in demand for certain products and media attention on grocery prices was persistent throughout the pandemic as supply chains adjusted and prices for staple items increased (Wiener-Bronner, Further exacerbating immense pressure and external criticism, some meat industry leaders added fuel to the fire. For example, Tyson Foods board chairman John Tyson placed a full-page ad in the New York Times, The Washington Post, and the Arkansas Democrat-Gazette on Sunday April 26 th , 2020, entitled -A Delicate Balance: Feeding the Nation and Keeping Our Employees Healthy‖ where he stated -the food supply chain is breaking‖ (Tyson, 2020) . Not surprisingly, this simultaneously elicited national attention and outrage by many inside and outside the agriculture industry. The corresponding media storm likely intensified panic buying of staple food items and rushes to procure local foods (Westervelt, 2020) , not to mention contributing to already heightened anxiety levels. Recently, consumer advocacy groups have alleged that major meat packers fueled fears of shortages during the pandemic to boost demand and prices for meat products (Dorning, 2021) . Many months after the initial shocks to supply chains the meat industry remains under scrutiny by consumer groups and regulators. This scrutiny illustrates continued public relations, and potentially legal, challenges resulting from the potential for public concern regardless of supply chain performance.

Social media is credited with significantly changing the mental, emotional, and overall human cognition surrounding the COVID-19 pandemic. ‗Doomscrolling' (mostly negative) news during the COVID-19 era on online media platforms is known to fuel anxiety (Garcia-Navarro, 2020) ; that behavior combined with the growth in streaming and screen disasters using online media data can facilitate timely response and augment other data sources (Middleton et al., 2014) . Since major viral outbreaks happen with less frequency, social media studies are less numerous. Studying the use of social media in both the U.S. and Nigeria during the Ebola outbreak, Odlum and Yoon (2016) encouraged the use of social media for public-specific and literacy-appropriate messaging to control outbreaks.

Prominent mass media coverage of the challenges faced by the meat supply chain during the COVID-19 pandemic are hypothesized to have influenced public perception of meat markets. This analysis uses online media listening and analytics, including natural language processing and net sentiment analysis, fundamentally the degree of positivity or negativity associated with the topic, to derive insight into the U.S. public's relative attention (quantitatively) and sentiment (qualitatively and quantitatively) surrounding the U.S. meat markets during 2019 and 2020.

The total volume of online media/search results, comprised of online media on these topics, is presented over time, enabling the documentation of ‗the anatomy of COVID-19 in U.S. meat markets' in online media space.

To further examine what happened to meat markets during this time, public perception, in the form of U.S. online media about meat markets, is also directly compared/contrasted to actual meat production, cold stocks of meat and poultry products, and overall availability of meat in the U.S.. Fundamentally, this analysis seeks to determine if the online media reflects reality in the U.S. meat markets; does perception reflect reality? Or, did online and social media volume and sentiment (perception) reflect an over or under reaction relative to the actual national meat production and availability situation (reality). Gaining understanding of this unique disaster in terms of online media versus reality may inform future disaster responses and aid in shaping communication with the public about food during times of stress.

The supply-side reality of the U.S. meat market situation, in terms of production and overall availability of meat products in U.S. markets is summarized using weekly production data in 2019 and 2020. The public's overall perception of the state of those markets in both 2019 and 2020 were quantified and summarized using online and social media searches and analytics.

Total meat production was calculated as the sum of beef, pork, chicken and turkey production for federally inspected and commercially slaughtered livestock and poultry. Weekly beef and pork production data were obtained from the USDA National Agricultural Statistics Service (NASS) (2021). Weekly chicken and turkey production data were obtained from the poultry slaughter reports released weekly by USDA Agricultural Marketing Service (AMS) (2021).

Since weekly chicken and turkey data is not available from USDA NASS, poultry slaughter numbers were converted from per head basis to total production by multiplying the total number of slaughtered animals by average weight and dressing percentage. Monthly meat stocks, measured/presented as millions of pounds of pork, chicken, and beef in cold storage were obtained from USDA NASS (2021). study affiliated stimulus. The positive aspects of social media listening, when done with the intention of understanding public perceptions or the public viewpoints on a topic of societal or public interest, is that the search results occurred sporadically or naturally and not in response to researcher stimuli. Online media exists in the form of comments, posts, reviews, responses to ongoing conversations, etc. (Widmar et al., 2020) . Whilst precise questioning and experimental design are the cornerstones of social science research based on surveys/questionnaires and focus groups, the lack of design is perhaps the appeal behind analyzing online and social media data. Natural language processing to discern sentiment Primary search terms parameterized a search which attempted to quantify online and social media, including news media, social media posts, blogs and other online chatter, and verbiage generally related to meat and U.S. meat markets. Fifty-three primary meat search terms were identified using USDA Agricultural Marketing Service (AMS) retail reports for beef, pork, chicken and turkey (USDA AMS, 2021). Researchers combined common terms and incorporated common hashtags to reference inputs to meat production lines, such as cattle and feedlot, and reference to meat cases and counters, to develop the eventual list of 53 J o u r n a l P r e -p r o o f Researchers employ both inclusionary and exclusionary search terms, refined in a process called search tuning, to tailor searches to the specific subject matter intended. There are many colloquial phrases, sports teams, and online games for example, that mention animals or meat products that were not the intended topic of study. Researchers developed a total of 250 exclusionary terms (Appendix A), including terms such as sacrificial lamb, you're a sheep, beefed up, Farmville, and Peppa Pig, among others. Simultaneously, 121 exclusionary domains were identified (Appendix A), including boardgamegeek.com, texashuntingforum.com, tractorybynet.com, and a variety of others determined by researchers to be contributing media unrelated to the subject matter intended. A total of 135 authors were also excluded during tuning processing (Appendix A).

An additional search filter developed to search for mentions of COVID-19 in the 2020 calendar year, within the findings of the primary meat search described, was developed. To develop these COVID-19 search terms preliminary data were collected weekly, on Sundays, for the previous Sunday through Saturday starting after the first cases appeared on December On Twitter, retweets are a process where a person can, for all intents and purposes, quote another person's post. Additional text may or may not be added. During analysis, retweets can be removed to reduce the amplification of the opinion if they are not providing additional information. Or, retweets can be kept to capture the influence of a broader issue (Gasco et al., 2019) . This analysis is on the general perceptions, from the online public, about meat markets, availability, and use in the US. Therefore, we wanted to be able to identify spikes in the overall number of posts and mentions, so we retained re-tweets in our analysis. In terms of sentiment analysis, the original tweet, and any additional text that was added to a quoted or -retweeted‖ tweet, was analyzed separately.

Top terms mentioned, domains, and sources summarizing search findings were summarized for 2019 and for 2020 independently, to facilitate comparisons. A source provides a general idea of where a post appeared, for example on a news cite. A more detailed example of where the post appeared, for example cnn.com, is a domain.

U.S. meat production in April and May 2020 was 1.5 billion lbs., or 10%, less than the same ‗breaking' (Tyson, 2020), one may posit that there was a massive amount of concern in the U.S. public about meat supply chains. However, the reality was that there was relatively small volumes of chatter or public conversation devoted to meat supply during the COVID pandemic in 2020 (Figure 1) . Furthermore, investigation into the meat supply, including production and stocks in cold storage, reveal that the meat supply chain was responsive, and that while production decreased measurably, there was not a shortage in the U.S. meat market. May 2020 saw a notable drop in total meat production, relative to the 2019-2020 average meat production, coming in 1.1 billion lbs. below average. Unexpectedly, U.S. meat in cold storage increased during May 2020. This may reflect the complexities being faced in the meat market during this time period as stocks were attempted to be replenished alongside continuing strength of demand for supermarket and retail food channels. Notably, the study of total meat may obscure some of the intricacies of meat demand during this time period when consumers were adjusting to at-home cooking and consumption, leaving certain meat items popularized in restaurants, like beef tenderloins or chicken wings, in abundance while ground beef and other at-home cooking staples remained hard to find in some places (Balagtas and Cooper, 2021; Hobbs, 2021; Ortez et al., 2021) .

While notable decreases in meat production existed in April and May 2020, meat production had rebounded by the first week of June 2020 as processing plants reopened during the month of May (Fu, 2020) . Furthermore, and generally missing from many critiques of the U.S. meat supply chain, is the understanding that immediately after the hardest months of the pandemic in the meat production system, the U.S. still had 1.7 billion lbs. of meat in cold storage (USDA NASS, 2021). Thus, production indeed fell measurably in the Spring of 2020, but there was not a shortage of meat in the U.S. supply chain or food system. It is important to note that although there was not a meat shortage, there were shortages of some specific products. For example, few retail grocery stores have butchers, meaning that beef products come to the store pre-rendered (Peel, 2021) . This limited the grocery stores' ability to adjust to increased beef demand caused by a shift from food service to retail (Peel, 2021) . Additional issues such as differences in the make-up of ground beef J o u r n a l P r e -p r o o f between retail and food service, resulted in specific bottlenecks that cold storage could not overcome for specific products (Peel, 2021) . However, meat, although perhaps not exactly the specific product the consumer wanted, was still readily available. Appendix B displays the total meat in cold storage for individual meat products.

The average net sentiment in the general meat search was 32 (Figure 4) . Broken down by year, the average net sentiment was 33 in 2019, with a standard deviation of 9, and 31 in 2020, with a standard deviation of 11. The overall net sentiment about meat in U.S. online indeed fell during the acute COVID-19 impact period of April and May 2020, but remained positive (24) on a -100 to +100 scale. While the net sentiment of search results in the general meat search did deviate over time, there were very few structural changes in meat sentiment seen. There is a notable drop in the net sentiment in the last week of May 2020. However, further investigation reveals news coverage about chickens killed in a fire in Pennsylvania (Stauffer, 2020 ) and a number of other non-COVID-19 related occurrences were the primary influences of this shift in sentiment. (Taylor et al., 2020; Saitone et al., 2021) and faced significant adjustments with closures that sent reverberations across the nation as livestock systems faced challenges with systems ‗backing up' as processing facilities closed or slowed (Marchant-Forde and Boyle, 2020; McCarthy and Danley, 2020; Balagtas and Cooper, 2021; Hobbs, 2021) . Taken together, the average net sentiment for the meat search with the COVID-19 filter was -27 for the January -May 2020 time period.

The second distinct time period in Figure 5 is the Summer of 2020, during which the average net sentiment of the meat search with the COVID-19 filter applied rose substantially to +18. This time period saw plants reopening, meat production returning to pre-pandemic levels, rebuilding of U.S. cold stocks of meat, and improvements in meat processing plant worker health reports as testing became more widely available and guidelines were put in place to try to protect worker health and safety (Fu, 2020) . The third and final phase of 

J o u r n a l P r e -p r o o f Journal Pre-proof Perception of the general public, and that of the media generated in 2020, regarding the U.S. meat supply chain functionality and performance versus reality of U.S. total meat production is worthy of investigation. Given the increased speed of person-to-person communication through social and interactive online media and ability of online media to shape and amplify conversations, the use of online and social media to discern perception of the U.S. meat industry is fitting and timely. Furthermore, the ability of the U.S. meat industry to recognize and respond to public concern may be necessary to respond transparently about product availability during disasters in the future.

The increase in total search mentions associated with COVID-19 was the smallest of the identified peaks in mentions throughout the 2019-2020 time period studied. The volume of mentions of COVID-19 in reference to the meat supply were timely in matching the timing of production volume being depressed due to plant closures, slowdowns, and COVID-19 challenges. Timing between the online media mentions and real-world happening aligns, although we cannot disentangle whether plant closures or production availability caused these increases or if they were or due to other mentions like safety among meat plan workers or related discussions. Simultaneous occurrences were seen but causality cannot be established.

Total public interest, measured as the percent of total online media mentions of COVID-19 relative to the total mentions of meat overall, was relatively small. Yet, a number of high-ranking meat industry leaders, along with major journalistic outlets and some academic or industry groups, warned of ‗impending doom' in the meat industry which never materialized. In today's high-speed information networks, the potential for sparking mass panic, especially during times of distress, is worrying. Reflection on the perceptionbroadly speakingversus reality, measured in actual pounds of production and the performance of the supply chain in delivering product to the U.S. marketplace is important to prevent unnecessary concern and potential panic in the future. Furthermore, directed investment in J o u r n a l P r e -p r o o f Journal Pre-proof building resiliency for future strains faced in the supply chain requires accurate understanding of the strain sustained, rather than over (or under) reactions fueled by speculation.

Despite the ongoing conversations about the pandemic revealing vulnerability in the meat supply chain, a direct comparison between perception and reality reveals that total meat production in the U.S. showed resiliency and efficient recovery after the April and May 2020 declines due to COVID-19-in this case, perception did reflect reality. Therefore, calls to create redundancies in the meat supply chain through disaggregation or local, direct-toconsumer marketing channels, while intuitively appealing, are not a panacea and lack rationale (Lusk et al., 2020; Hobbs, 2021; Ma and Lusk, 2021) . Much has been written about the complex tradeoffs between a supply chain that is efficient and produces affordable food for consumers during -normal‖ times and one that is redundant, flexible, and resilient to -abnormal‖ shocks (Lusk et al., 2020; Hobbs, 2021) . Therefore, proposals aimed at addressing the structure of the meat supply chain in the name of the pandemic need to be 

Smithfield sued for profiting from Covid meat-shortage fears

Earthquake twitter

Using value modeling to evaluate social media messages: The case of Hurricane Irene

Meat processing plants are reopening. Is it safe for workers to return? The Counter

Your ‗doomscrolling' breeds anxiety

Beyond sound level monitoring: Exploitation of social media to gather citizens subjective response to noise

Must-Know vocab for COVID-19 From Droplets to Zoonotic

Rapid assessment of disaster damage using social media activity

Tweeting about emergency: A semantic network analysis of government organizations' social media messaging during Hurricane Harvey

Beef and pork marketing margins and price spreads during COVID-19

Concentration and resiliency in the U.S. meat supply chains

COVID-19 effects on livestock production: a one welfare issue

Leveraging Twitter to gauge evacuation compliance: spatiotemporal analysis of Hurricane Matthew

How does the new coronavirus compare with the flu? Scientific American

Crisis communication patterns in social media during Hurricane Sandy

COIVID-19 morbidity and mortality in U.S. meatpacking counties

A decade before coronavirus, I covered the swine flu pandemic. Here's how it compares to COVID-19. ABC News

4-alarm Martic Township blaze killed 42,000 chickens; firefighters saved 2 attached buildings, chief says. LNP Lancaster Online

Livestock plants and COVID-19

2021. Weekly Poultry Slaughtered Under Federal Inspection (NW_PY017)

USDA National Agricultural Statistics Service (NASS). 2021. Quick Stats. Accessed

As food supply chain breaks down, farm-to-door CSAs take off

Social media sentiment as an additional performance measure? Examples from iconic theme park destinations

The anatomy of natural disasters on online media: hurricanes and wildfires. Natural Hazards

CNN Exclude Terms @Blacknmild, @YouTube, pig out, @sheep-in-clouds, Peppa pig, peppa pig, pump milk, hogging the, Egg Town, cattle call, hog tied, GO CHICKENS, go chickens, go bulls, ranch village, farmville, #Farmville, #farmville, Farmville, chicken-radio, chicken radio, LA rams, LA Rams, playing chicken, hen party, I bet ewe, sheeple, The Silence of the Lambs, silence of the lambs, Silence of the Lambs, starting beef, bush hog, too chicken, play cicken, cash cow, cash-cow, eggs in one basket, wolf in sheeps clothing, wolf in sheep's clothing, Gaston, Shaun the Sheep, shaun the sheep, hog wild, wild boars, wild boar, guinea pigs

Bay of Pigs, bay of pigs, kicken' chicken, kickin' chicken, commie sheep, smart sheep, tell the other sheep, Dirty Pig, dirty pig, you are a chicken, you're a chicken, stuffed animal, bloated pig, lipstick on a pig, spring chicken, name is cow, wild hog, Wild Hog, lamb of god, lamb of God, electric sheep, #lambofgod, dishonor on my cow, chicken shit

Jumbo Eye pencil, milkmakeup, lambs of dalamud, Im a goat, I'm a goat, I am a goat, I am a sheep, I'm a sheep, you're a sheep, chicken nugget animation, pig-boy, anime, chicken fucking, chicken fucker, chicken fuckers, boobs, y'all beef, FUCK YOUR SHEEP, hog-arm

Turkey Brooke Park, pork to be added to the bill, gang shit, now I got beef, Greenland Shark, have beef wit me, no serious beef, PS5, Greece, Madrid, midfielder, soundtrack, going HAM, old town road, Geoffrey Bacon, Nintendo, #GeorgeFloydProtests, GeorgeFloyd, George Floyd, parliament, Turkey earthquake Exclude Domains Boards.4chan.org, breitbart.com, steamcommunity.com, forums.hardwarezone.com.sg, hogville.net, forums.digitalspy.com, bigfooty.com, hotcopper.com.au, uk.answers.yahoo.com, au.answers.yahoo.com, tigerdroppings.com, forums.spacebattles.com, forums.penny-arcade.com, gamefaqs.com, ign.com, nz.answers.yahoo.com, boardgamegeek.com, thefarmingforum.co.uk, forum.supercell.com, dailymail.co.uk, topix.com, worldstarhiphop.com, boards.ie, thehill.com, medium.com, nairaland.com, neogaf.com, texashuntingforum.com, article.wn.com, dailycaller.com, fark.com, forums.sherdog.com, infowars.com, stuff.co.nz, trueviralnews.com, play.google.com, thestudentroom.co.uk, forum.lowyat.net, theland.com.au, pastebin.com, J o u r n a l P r e -p r o o f defence.pk, s13.zifboards.com, 24hourcampfire.com, mediaite.om, investorshub.advfn.com, vanguardngr.com, tmz.com, mmo-champion

chickensoupbot:twitter.com, faridabadtutor:twitter.com, eggsbenedict-and-butter:tumblr.com, barntiques859:twitter.com, botburgers:twitter.com, gaylord69bot:twitter.com, RamsViews:twitter.com, RatioChristi:twitter.com, NewpetSuppliesN:twitter.com