The Long-Term Effects of Meat Recalls on Futures Markets

The Long-Term Effects of Meat Recalls on Futures Markets Abstract Over the past twenty years, there has been an increasing trend in the number of food safety events. Despite increased safety control standards, foodborne disease outbreaks continue to impact the food supply. A common source of foodborne illness and fatal infection is beef, with potential causes being E. coli 0157: H7, Listeria Monocytogenes, and Salmonella. Certain companies have even been bankrupted, unable to overcome the social costs and economic losses associated with recalls. We look beyond the impact on processors to see the amount of impact transmitted through price channels to cattle farmers. We examine beef recalls over a twenty-year period through an accumulated two-year index to see if there is a prolonged effect of recalls on current weekly cattle prices. We find that food safety events act together, adversely impacting prices and decreasing farm-level revenue. The finding that economic harm to the beef industry from food safety events lingers for at least two years suggests that the industry should do more to reduce food safety incidents. Cattle futures, food safety, futures market, recalls, revenue loss In the past 20 years, there has been considerable research into the effects of food safety events on consumer demand, likely due to the increase in the number of recalls over the same period. From 1996 to 2016, the meat and poultry industry witnessed an increasing trend in the number of recalls. USDA Food Safety and Inspection Service (FSIS 2015) reports that, despite industry and government efforts to safeguard food, product contamination by adulterants continues to be an ongoing concern. Food safety issues can result in severe economic losses to food companies and even larger social costs for society, ranging from company bankruptcy to consumer illness or death. The CDC estimates that each year roughly 1 in 6 Americans (or 48 million people) get sick, 128,000 are hospitalized, and 3,000 die from foodborne diseases (Center for Disease Control and Prevention 2016). Recalls are designed to protect consumers by removing contaminated product from commerce, but also serve a secondary purpose of incentivizing companies to prevent foodborne disease outbreaks from happening. It seems obvious that food safety events generally lead to negative retail price and revenue impacts because the recalls and surrounding media attention associated with them suggest lower product quality or ineffective quality control, resulting in a loss of consumer confidence (Marsh, Schroeder, and Mintert 2004), lower expected utility from consumption, diminished sales, and lower consumer prices. In turn, traders in derivatives markets react to the sudden, negative news, resulting in a drop in futures prices for the implicated commodity. This makes sense as a decline in retail demand will transmit itself through the supply chain and result in reduced derived demand for the original farm- level products that are ingredients in the processed food being recalled. Thus, mistakes made by food processors can lead to economic losses for farmers. While there has been considerable research examining the effects of recalls on short-term price movements (McKenzie and Thomsen 2001; Lusk and Schroeder 2002; Moghadam, Schmidt, and Grier 2013), there is no research on the cumulative impact that food safety events play in shifting long-term consumer demand preferences. Lusk and Schroeder (2002) concluded that if meat recalls resulted in a systematic change in the structural demand for meat, the change only occurs gradually and must be examined through an extended period of time. If food safety events do have a more prolonged effect, then knowing this would help identify the effectiveness of recall policy. Therefore, this paper aims to build on existing literature by investigating the long-term impact of recalls on weekly live cattle futures markets using an author-constructed recall index. More specifically, the objective of this paper is to analyze if additional meat recalls have long-run impacts on weekly live cattle futures prices, and thus on farmers’ incomes. We find that food safety events do have surprisingly long-lasting impacts and thus, processors might want to invest more heavily in preventing food safety events from occurring. The rest of the paper is organized as follows. The next section presents a literature review related to the long-run impacts of recalls on futures market prices. This is followed by an account of the meat recall process, then the data used are described. Our empirical methodology is then explained, followed by our empirical results. Conclusions and policy implications are discussed in the last section. Literature Review Much of the early literature, while not explicitly studying the impacts of food safety events, identified health information as a demand factor in consumer response. Moschini and Meilke (1989) introduced modeling the structural change in demand for meat. In their seminal paper, these authors found that the dynamics of prices and incomes cannot fully explain consumer demand shifts from meat to chicken; however, dietary concerns provided the best explanation for shifts towards white meats. Kinnucan et al. (1997) examined the combined effects of advertising and health information on meat demand. These authors hypothesized that health information, such as linkages between cholesterol and heart disease, impacted beef demand, which was confirmed by the positive and negative cross- and own-price elasticity for poultry and beef, respectively. The first study conducted to determine the short-run impact of health information was done by Robenstein and Thurman (1996), which revealed insignificant effects of Wall Street Journal articles on livestock futures markets. Food safety information differs, however, from health information in that such information, for example pathogenic contamination, can result in short-term, long-term, or fatal consequences. These events are mostly unanticipated and the shock severity may cause a considerable market reaction. The shock of a food recall can cause a lack of consumer confidence, drive prices down, and ultimately impact traders’ decisions, which has been the focus of much research. For instance, Henson and Mazzocchi (2002) found that processors of beef, dairy, and pet food products were negatively affected by reported linkages between BSE and Creutzfeldt–Jacob disease. Thomsen and McKenzie (2001) studied the effects of recalls on shareholder wealth to find that serious health concerns did cause a drop in stock prices. However, in a second paper, McKenzie and Thomsen (2001) concluded that wholesale prices were impacted by recalls, yet this impact did not transmit to farm-level derivative prices. On the other hand, using food recalls as a proxy for food quality, Lusk and Schroeder (2002) found that medium-sized, serious food recalls did impact short-run farm-level derivatives prices for live cattle and lean hogs. Furthermore, Moghadam, Schmidt, and Grier (2013) revisited the McKenzie and Thomsen (2001) result to find that there was, in fact, a significant reaction in live cattle futures markets from E. coli 0157: H7 contamination. The researchers of the more recent publication concluded that modeling the role of recalls on futures prices was incomplete without the introduction of gradual effects. This finding suggests further investigation into the long-term effect of meat recalls. A number of studies explore the long-run effects of food safety on meat demand, many of which utilized food safety indexes for empirical methodology. One of the principal studies in this branch of research is Burton and Young (1997), which used media articles and popular press to measure the impact of food safety on beef demand. In their AIDS model, these authors found a short- and long-term effect of publicity on consumer expenditure for meat and a drop in meat market share. Patterson and Flake (1999) extended the literature concerning the effects of BSE, E. coli, and salmonella on meat demand by studying Associated Press articles. Both of these papers found a negative impact on demand, confirming that the market responds to food safety occurrences, and informational indexes have effective explanatory power. Piggott and Marsh (2004) incorporated similar empirical methodology with a food safety index for publicized recalls. Overall, their paper revealed small changes over time in average consumer response to food safety, with some periods of large, short-lived impacts. Marsh, Schroeder, and Mintert (2004) built upon this research by examining demand responses to food recalls from newspaper articles and FSIS meat product recalls. While meat product recalls significantly impacted demand, with elasticity measures indicating a change from meat to non-meat consumption, media indices had no effect. The early literature analyzed prices following USDA announcements, while more recent studies have incorporated media articles as a primary source of consumer information and decision-making. Schlenker and Villa-Boas (2009) compared the two information sources and found that independent news media created a more adverse impact on prices than a government report warning of the first BSE outbreak. Nevertheless, government reports on food scares are typically the first information that traders and consumers have access to. Therefore, it is still believed that USDA recall announcements contain relevant information to gauge how market participants respond to food safety events. This study will follow the recent literature in that an index measure will be used to quantify the food safety impact. However, being that information contained in the reports (recall size, severity and description) has consistently yielded significant results, the index will only include information that market participants could access through the USDA announcements. The existing literature has succeeded in determining the immediate market response of meat recalls on futures market prices. It is hypothesized in this paper that current cattle prices reflect many factors, one of which being food recalls from months (or years) past. It is then believed that the indexed effect of cumulative recalls will give a more accurate representation of how traders update prices in response to new outbreaks. Specifically, the results will show the revenue lost by cattle producers due to a cumulative, time- and severity-weighted index of recall events over the past two years. This is possible through using a rolling index, allowing for the inclusion of all recalls within a 24-month time window. The inclusion of all recalls removes a limitation in some previous work that dropped some recall data to achieve identification.1 Recall Background The FSIS is an agency of the USDA that handles product recalls for meat, poultry, and egg products. According to FSIS’s recall policy, a recall can begin in several ways. First, the company can identify an adulterant or other product defect and notify USDA officials immediately. The FSIS will investigate the claim and issue a Recall Notification Report (RNR) for non-serious, Class III recalls or publish a press release if the recall is serious (Class I or II). The recall process can also begin with FSIS, the state government, or third-party plant inspectors identifying a non-compliance issue that warrants product removal (USDA FSIS). The RNR or Recall Release contains all the information concerning the recall, such as company, reason for recall, date of recall, method of discovery, recall class, pounds recalled, reported consumer illness, and other related information. In addition to the reason for the recall and product, each report notifies the public of the recall severity by categorizing the event into one of three groups: Class I, Class II, or Class III. According to the FSIS website, “A Class I recall involves a health hazard situation in which there is a reasonable probability that eating the food will cause health problems or death,” (USDA FSIS). Class II involves a contaminant that presents a remote probability of health concern, and Class III involves situations expected to result in no negative health side effects. After a report has been issued on the size and class of the recall, the FSIS works with the company throughout the recall process to ensure all contaminated products are removed from commerce before closing the recall. Data Using Datastream software, weekly live cattle futures prices were gathered from the Chicago Mercantile Exchange (CME). The CME contains information for different commodities, contracts and frequency. Specifically, live cattle contracts were presented in daily, weekly, and monthly increments for the expiration months of February, April, June, August, October, and December. A frequency of weekly prices better fits the frequency of recall announcements and abstracts from the noise of daily price changes. To organize prices into a continuous series, contracts were rolled over to the next nearest contract at the start of the expiration month. Following the logic of the previous literature, it is believed that futures prices—specifically, live cattle futures—will respond most strongly to a USDA recall. The recall information was pulled from the USDA’s FSIS website catalogue of meat, poultry, and egg recalls from 1982 through 2017. The Recall Case Archive provides summary information on completed recalls beginning in 1994. We limit our examination of recalls to the period from January 1, 1995 to December 31, 2015. The information contained in each data frame is the press release information listed above, including company, reason for recall, date of recall, method of discovery, recall class, pounds recalled, and consumer illness—if there was one. For the purpose of this study, the information extracted and used for analysis will be company, date, recall class, and pounds recalled for beef recalls. Beef recalls were selected because they remain a common source of food contamination, which the literature has already closely analyzed. Since recalls often necessitate the removal of multiple meat products, the criterion was restricted to include only beef-specific events. A summary of these recalls is shown in table 1. Table 1 Number and Volume of Beef Recalls per Year Year All recalls # of recalls Total (lbs.) Mean (lbs.) Std. Dev. (lbs.) 1995 6 963,956.00 160,659.33 176,190.02 1996 7 171,734.00 24,533.43 56,337.99 1997 10 26,322,529.00 2,632,252.90 7,861,897.91 1998 18 4,946,491.00 274,805.06 635,835.08 1999 14 3,396,154.00 242,582.43 557,261.94 2000 25 2,925,476.00 117,019.04 230,259.44 2001 29 1,600,682.00 55,195.93 154,198.14 2002 41 24,133,913.00 588,632.02 2,981,869.68 2003 23 1,776,945.00 80,770.23 170,399.92 2004 14 1,509,515.00 107,822.50 153,897.00 2005 11 1,740,982.00 158,271.09 269,905.06 2006 16 5,021,153.00 313,822.06 1,069,512.08 2007 26 33,485,915.00 1,287,919.81 4,350,286.16 2008 20 150,549,632.00 8,363,868.44 33,720,083.09 2009 29 3,777,014.00 130,241.86 227,842.61 2010 26 24,561,504.00 944,673.23 3,092,721.06 2011 27 1,248,938.00 49,957.52 94,650.12 2012 12 124,777.00 10,398.08 15,055.14 2013 15 382,915.00 27,351.07 37,939.40 2014 9 11,253,356.00 1,250,372.89 2,870,911.44 2015 26 1,077,624.00 43,104.96 107,043.80 Year All recalls # of recalls Total (lbs.) Mean (lbs.) Std. Dev. (lbs.) 1995 6 963,956.00 160,659.33 176,190.02 1996 7 171,734.00 24,533.43 56,337.99 1997 10 26,322,529.00 2,632,252.90 7,861,897.91 1998 18 4,946,491.00 274,805.06 635,835.08 1999 14 3,396,154.00 242,582.43 557,261.94 2000 25 2,925,476.00 117,019.04 230,259.44 2001 29 1,600,682.00 55,195.93 154,198.14 2002 41 24,133,913.00 588,632.02 2,981,869.68 2003 23 1,776,945.00 80,770.23 170,399.92 2004 14 1,509,515.00 107,822.50 153,897.00 2005 11 1,740,982.00 158,271.09 269,905.06 2006 16 5,021,153.00 313,822.06 1,069,512.08 2007 26 33,485,915.00 1,287,919.81 4,350,286.16 2008 20 150,549,632.00 8,363,868.44 33,720,083.09 2009 29 3,777,014.00 130,241.86 227,842.61 2010 26 24,561,504.00 944,673.23 3,092,721.06 2011 27 1,248,938.00 49,957.52 94,650.12 2012 12 124,777.00 10,398.08 15,055.14 2013 15 382,915.00 27,351.07 37,939.40 2014 9 11,253,356.00 1,250,372.89 2,870,911.44 2015 26 1,077,624.00 43,104.96 107,043.80 Table 1 Number and Volume of Beef Recalls per Year Year All recalls # of recalls Total (lbs.) Mean (lbs.) Std. Dev. (lbs.) 1995 6 963,956.00 160,659.33 176,190.02 1996 7 171,734.00 24,533.43 56,337.99 1997 10 26,322,529.00 2,632,252.90 7,861,897.91 1998 18 4,946,491.00 274,805.06 635,835.08 1999 14 3,396,154.00 242,582.43 557,261.94 2000 25 2,925,476.00 117,019.04 230,259.44 2001 29 1,600,682.00 55,195.93 154,198.14 2002 41 24,133,913.00 588,632.02 2,981,869.68 2003 23 1,776,945.00 80,770.23 170,399.92 2004 14 1,509,515.00 107,822.50 153,897.00 2005 11 1,740,982.00 158,271.09 269,905.06 2006 16 5,021,153.00 313,822.06 1,069,512.08 2007 26 33,485,915.00 1,287,919.81 4,350,286.16 2008 20 150,549,632.00 8,363,868.44 33,720,083.09 2009 29 3,777,014.00 130,241.86 227,842.61 2010 26 24,561,504.00 944,673.23 3,092,721.06 2011 27 1,248,938.00 49,957.52 94,650.12 2012 12 124,777.00 10,398.08 15,055.14 2013 15 382,915.00 27,351.07 37,939.40 2014 9 11,253,356.00 1,250,372.89 2,870,911.44 2015 26 1,077,624.00 43,104.96 107,043.80 Year All recalls # of recalls Total (lbs.) Mean (lbs.) Std. Dev. (lbs.) 1995 6 963,956.00 160,659.33 176,190.02 1996 7 171,734.00 24,533.43 56,337.99 1997 10 26,322,529.00 2,632,252.90 7,861,897.91 1998 18 4,946,491.00 274,805.06 635,835.08 1999 14 3,396,154.00 242,582.43 557,261.94 2000 25 2,925,476.00 117,019.04 230,259.44 2001 29 1,600,682.00 55,195.93 154,198.14 2002 41 24,133,913.00 588,632.02 2,981,869.68 2003 23 1,776,945.00 80,770.23 170,399.92 2004 14 1,509,515.00 107,822.50 153,897.00 2005 11 1,740,982.00 158,271.09 269,905.06 2006 16 5,021,153.00 313,822.06 1,069,512.08 2007 26 33,485,915.00 1,287,919.81 4,350,286.16 2008 20 150,549,632.00 8,363,868.44 33,720,083.09 2009 29 3,777,014.00 130,241.86 227,842.61 2010 26 24,561,504.00 944,673.23 3,092,721.06 2011 27 1,248,938.00 49,957.52 94,650.12 2012 12 124,777.00 10,398.08 15,055.14 2013 15 382,915.00 27,351.07 37,939.40 2014 9 11,253,356.00 1,250,372.89 2,870,911.44 2015 26 1,077,624.00 43,104.96 107,043.80 The reasons for recalls range from labeling to bacteria contamination. From 1995 through 2015, the leading cause of beef recalls was E.coli O157:H7. The second-leading reason for a recall fell in the category “Other”, which included misbranding, extraneous materials, or a non-life-threatening chemical/spoilage exposure. Therefore, the data reflect higher Class I and Class II/III recalls for the former and latter, respectively. Table 2 displays a chart with the various types of recall reasons and number of occurrences for each. Table 2 Type of Recalls 1995–2015 All recalls Year Listeria monocytogenes E.coli O157: H7 Allergen Other bacteria Other Processing 1995 0 5 0 0 1 1 1996 3 2 0 0 2 2 1997 0 6 0 0 4 0 1998 1 12 0 1 4 1 1999 2 9 0 0 3 2 2000 5 18 0 0 2 0 2001 5 21 1 1 1 1 2002 3 31 1 1 5 2 2003 3 10 1 1 8 1 2004 2 5 4 0 2 1 2005 3 5 1 1 1 0 2006 0 8 2 1 5 0 2007 1 22 0 0 3 2 2008 2 15 1 0 2 0 2009 4 13 1 3 8 0 2010 2 11 4 2 7 2 2011 2 12 5 2 6 2 2012 1 3 1 1 6 0 2013 1 5 3 2 4 1 2014 0 1 2 0 6 2 2015 1 7 9 0 9 0 Total 41 221 36 16 89 20 All recalls Year Listeria monocytogenes E.coli O157: H7 Allergen Other bacteria Other Processing 1995 0 5 0 0 1 1 1996 3 2 0 0 2 2 1997 0 6 0 0 4 0 1998 1 12 0 1 4 1 1999 2 9 0 0 3 2 2000 5 18 0 0 2 0 2001 5 21 1 1 1 1 2002 3 31 1 1 5 2 2003 3 10 1 1 8 1 2004 2 5 4 0 2 1 2005 3 5 1 1 1 0 2006 0 8 2 1 5 0 2007 1 22 0 0 3 2 2008 2 15 1 0 2 0 2009 4 13 1 3 8 0 2010 2 11 4 2 7 2 2011 2 12 5 2 6 2 2012 1 3 1 1 6 0 2013 1 5 3 2 4 1 2014 0 1 2 0 6 2 2015 1 7 9 0 9 0 Total 41 221 36 16 89 20 Table 2 Type of Recalls 1995–2015 All recalls Year Listeria monocytogenes E.coli O157: H7 Allergen Other bacteria Other Processing 1995 0 5 0 0 1 1 1996 3 2 0 0 2 2 1997 0 6 0 0 4 0 1998 1 12 0 1 4 1 1999 2 9 0 0 3 2 2000 5 18 0 0 2 0 2001 5 21 1 1 1 1 2002 3 31 1 1 5 2 2003 3 10 1 1 8 1 2004 2 5 4 0 2 1 2005 3 5 1 1 1 0 2006 0 8 2 1 5 0 2007 1 22 0 0 3 2 2008 2 15 1 0 2 0 2009 4 13 1 3 8 0 2010 2 11 4 2 7 2 2011 2 12 5 2 6 2 2012 1 3 1 1 6 0 2013 1 5 3 2 4 1 2014 0 1 2 0 6 2 2015 1 7 9 0 9 0 Total 41 221 36 16 89 20 All recalls Year Listeria monocytogenes E.coli O157: H7 Allergen Other bacteria Other Processing 1995 0 5 0 0 1 1 1996 3 2 0 0 2 2 1997 0 6 0 0 4 0 1998 1 12 0 1 4 1 1999 2 9 0 0 3 2 2000 5 18 0 0 2 0 2001 5 21 1 1 1 1 2002 3 31 1 1 5 2 2003 3 10 1 1 8 1 2004 2 5 4 0 2 1 2005 3 5 1 1 1 0 2006 0 8 2 1 5 0 2007 1 22 0 0 3 2 2008 2 15 1 0 2 0 2009 4 13 1 3 8 0 2010 2 11 4 2 7 2 2011 2 12 5 2 6 2 2012 1 3 1 1 6 0 2013 1 5 3 2 4 1 2014 0 1 2 0 6 2 2015 1 7 9 0 9 0 Total 41 221 36 16 89 20 Index Construction Significant consideration and research was taken in developing our index. Lusk and Schroeder (2002) determined that seriousness and size of recall determine market response. Other papers also identified recall class and size as key variables in price response to food safety events. Pozo and Schroeder (2016) found recall size to be significant, causing abnormal stock returns to become more negative when increased by 1%. The literature consistently finds that recalls involving serious health consequences or death have significant impact on financial variables (Thomsen and McKenzie 2001; Pozo and Schroeder 2016). Therefore, this study treats recall size and severity as primary measures for the index construction, yielding a higher index value for more serious recalls. We might have tried to incorporate media coverage into the index construction, but decided that converting media mentions into a quantitative measure was too subjective. Following Lusk and Schroeder (2002), this paper includes size categories as follows: size1 = less than 1,162 lbs., size2 = between 1,162 and 4,516 lbs., size3 = between 4,516 and 32,000 lbs., size4 = between 32,000 and 175,288 lbs., and size5 = greater than 175,288 lbs. For each of these categories, we assign an index of odd numbers 1 through 9, with 9 being assigned to the category with the most pounds recalled. Next, we index the classes by 3, 2, and 1 for Class I, II and III, respectively, giving a larger weight to events more likely to cause a serious illness. These two numbers are then summed to arrive at the contribution to the index for each recall event. While most previous research focused on the short-term impact of recalls and food safety information, we are taking a long-term view, hypothesizing that consumers build up an impression on the safety of foods from cumulative exposure to media coverage of food safety incidents such as recalls. Piggott and Marsh (2004) did test for such an effect in a meat demand system, and found no impact of food safety information on prices beyond a calendar quarter. Thus, if our hypothesis is correct, our results will differ from their findings. Clearly, any such longer-term, cumulative index is ad hoc in nature. Therefore, sensitivity analysis will be conducted on the length and shape that should be imposed on the index that reflect the rate of decay for past information’s effect on cattle prices. We explain the baseline index specification now and then report sensitivity analysis findings in the empirical results section. The two-year frame for our baseline index was divided into four subperiods that consist of six months of recall data for each group. All of the recalls in a six-month period were summed together to find the total index measure per six months. Each subperiod is then weighted decreasingly by distance from the current week. The most recent six months have a weight of one, the following six months have a weight of 0.75, and the remaining two six-month subperiods receive weights of 0.5 and 0.25, respectively. Then all the weighted values are summed to arrive at the final index value for that week. In turn, this window will move through the entire dataset of recalls, re-indexing every week, creating an index based on a rolling window of two years of recalls. We calculated the index from December 31, 1996 through December 22, 2015, totaling 989 weekly observations. Empirical Methodology Based on demand theory, we treat the food safety index as a demand shifter.2 This method was successfully used by Marsh, Schroeder, and Mintert (2004) and Attavanich, McCarl, and Bessler (2011). We use a simple autoregressive model of live cattle futures prices with the addition of several exogenous variables. The explanatory variables included in this study were chosen from previous literature and economic theory, specifically as it relates to food recall events. The main exogenous variable consists of a two-year accumulated recall index. Since beef, like most agricultural commodities, demonstrates seasonality, the empirical model also includes dummy variables for months, and yearly fixed effects are included to account for inflation and trends in tastes and preferences. The logarithm of the price series was used as the dependent variable and when tested for nonstationarity with an augmented Dickey-Fuller test was found to be stationary. Two lagged values of logged futures prices were included as regressors to account for past prices influencing present prices. Modeled autocorrelation and partial autocorrelation graphs justify the specification. Thus, the equation used to model prices is given by lnPt=β0+β1Indext+β2ln⁡Pt-1+β3ln⁡Pt-2+∑j=212θjMonthjt+∑k=219δkYearkt+ɛt (1) where P is the nearby futures price with t = 1,…, 886 denoting the week in our sample, Index denotes the accumulated index, Monthjt is a set of monthly dummy variables, the Yearkt are the year dummies to provide yearly fixed effects, and ɛ is the stochastic error term. Empirical Results Equation (1) is estimated using OLS. The main variable of interest, Index, is statistically significant at the 5% level and the coefficient for the index variable is negative (−0.0000893), which supports our hypothesis that an accumulation of recalls will put downward pressure on cattle prices. Table 3 provides the full regression results for our model. Table 3 Regression Results Log (cattle price) Coefficient Standard error p values Intercept 0.47414*** 0.072 <0.001 Index −0.00009** 0.000 0.011 Log(P)t-1 0.86462*** 0.035 <0.001 Log(P)t-2 0.02359 0.033 0.475 Year  1998 −0.00323 0.004 0.455  1999 0.01002** 0.005 0.034  2000 0.01865*** 0.006 0.001  2001 0.02420*** 0.007 0.001  2002 0.02318** 0.008 0.003  2003 0.03662*** 0.010 <0.001  2004 0.04039*** 0.008 <0.001  2005 0.03700*** 0.007 <0.001  2006 0.03269*** 0.007 <0.001  2007 0.04833*** 0.008 <0.001  2008 0.05361*** 0.010 <0.001  2009 0.04062*** 0.008 <0.001  2010 0.06293*** 0.011 <0.001  2011 0.07920*** 0.013 <0.001  2012 0.08032*** 0.012 <0.001  2013 0.07764*** 0.012 <0.001  2014 0.09908*** 0.014 <0.001  2015 0.09028*** 0.015 <0.001 Month  February 0.00519 0.004 0.155  March 0.00124 0.003 0.702  April −0.01381*** 0.004 <0.001  May −0.00379 0.003 0.260  June −0.00089 0.003 0.798  July 0.00002 0.003 0.995  August 0.00653* 0.004 0.068  September 0.00419 0.003 0.187  October 0.00935*** 0.003 0.004  November 0.00442 0.004 0.211  December 0.00607 0.004 0.124 R2 0.9927 Log (cattle price) Coefficient Standard error p values Intercept 0.47414*** 0.072 <0.001 Index −0.00009** 0.000 0.011 Log(P)t-1 0.86462*** 0.035 <0.001 Log(P)t-2 0.02359 0.033 0.475 Year  1998 −0.00323 0.004 0.455  1999 0.01002** 0.005 0.034  2000 0.01865*** 0.006 0.001  2001 0.02420*** 0.007 0.001  2002 0.02318** 0.008 0.003  2003 0.03662*** 0.010 <0.001  2004 0.04039*** 0.008 <0.001  2005 0.03700*** 0.007 <0.001  2006 0.03269*** 0.007 <0.001  2007 0.04833*** 0.008 <0.001  2008 0.05361*** 0.010 <0.001  2009 0.04062*** 0.008 <0.001  2010 0.06293*** 0.011 <0.001  2011 0.07920*** 0.013 <0.001  2012 0.08032*** 0.012 <0.001  2013 0.07764*** 0.012 <0.001  2014 0.09908*** 0.014 <0.001  2015 0.09028*** 0.015 <0.001 Month  February 0.00519 0.004 0.155  March 0.00124 0.003 0.702  April −0.01381*** 0.004 <0.001  May −0.00379 0.003 0.260  June −0.00089 0.003 0.798  July 0.00002 0.003 0.995  August 0.00653* 0.004 0.068  September 0.00419 0.003 0.187  October 0.00935*** 0.003 0.004  November 0.00442 0.004 0.211  December 0.00607 0.004 0.124 R2 0.9927 Note: *significant at 10%, **significant at 05%, ***significant at 01% Table 3 Regression Results Log (cattle price) Coefficient Standard error p values Intercept 0.47414*** 0.072 <0.001 Index −0.00009** 0.000 0.011 Log(P)t-1 0.86462*** 0.035 <0.001 Log(P)t-2 0.02359 0.033 0.475 Year  1998 −0.00323 0.004 0.455  1999 0.01002** 0.005 0.034  2000 0.01865*** 0.006 0.001  2001 0.02420*** 0.007 0.001  2002 0.02318** 0.008 0.003  2003 0.03662*** 0.010 <0.001  2004 0.04039*** 0.008 <0.001  2005 0.03700*** 0.007 <0.001  2006 0.03269*** 0.007 <0.001  2007 0.04833*** 0.008 <0.001  2008 0.05361*** 0.010 <0.001  2009 0.04062*** 0.008 <0.001  2010 0.06293*** 0.011 <0.001  2011 0.07920*** 0.013 <0.001  2012 0.08032*** 0.012 <0.001  2013 0.07764*** 0.012 <0.001  2014 0.09908*** 0.014 <0.001  2015 0.09028*** 0.015 <0.001 Month  February 0.00519 0.004 0.155  March 0.00124 0.003 0.702  April −0.01381*** 0.004 <0.001  May −0.00379 0.003 0.260  June −0.00089 0.003 0.798  July 0.00002 0.003 0.995  August 0.00653* 0.004 0.068  September 0.00419 0.003 0.187  October 0.00935*** 0.003 0.004  November 0.00442 0.004 0.211  December 0.00607 0.004 0.124 R2 0.9927 Log (cattle price) Coefficient Standard error p values Intercept 0.47414*** 0.072 <0.001 Index −0.00009** 0.000 0.011 Log(P)t-1 0.86462*** 0.035 <0.001 Log(P)t-2 0.02359 0.033 0.475 Year  1998 −0.00323 0.004 0.455  1999 0.01002** 0.005 0.034  2000 0.01865*** 0.006 0.001  2001 0.02420*** 0.007 0.001  2002 0.02318** 0.008 0.003  2003 0.03662*** 0.010 <0.001  2004 0.04039*** 0.008 <0.001  2005 0.03700*** 0.007 <0.001  2006 0.03269*** 0.007 <0.001  2007 0.04833*** 0.008 <0.001  2008 0.05361*** 0.010 <0.001  2009 0.04062*** 0.008 <0.001  2010 0.06293*** 0.011 <0.001  2011 0.07920*** 0.013 <0.001  2012 0.08032*** 0.012 <0.001  2013 0.07764*** 0.012 <0.001  2014 0.09908*** 0.014 <0.001  2015 0.09028*** 0.015 <0.001 Month  February 0.00519 0.004 0.155  March 0.00124 0.003 0.702  April −0.01381*** 0.004 <0.001  May −0.00379 0.003 0.260  June −0.00089 0.003 0.798  July 0.00002 0.003 0.995  August 0.00653* 0.004 0.068  September 0.00419 0.003 0.187  October 0.00935*** 0.003 0.004  November 0.00442 0.004 0.211  December 0.00607 0.004 0.124 R2 0.9927 Note: *significant at 10%, **significant at 05%, ***significant at 01% The results indicate that a single recall, even of sizable and serious proportions, cannot increase the index enough to push farm-level prices downward by an economically significant amount. However, the accumulation of recalls does add up to a price effect, transmitted through the supply chain, that is large enough to matter to farmers, processors, and retailers. To show that beef industry actors are being impacted enough to induce action, we need to put these results more clearly in context. Economic Impact Fewer recalls means increased farm revenue, according to our results, but how much? As one way to put the numbers in context, we computed the annual change in the index each year of our sample. The annual index change in table 4 below gives the cumulative change for one year of recall information. The recall index increased the most in 1998 (+80.75), while its biggest drop was in 2012 (-100.75). Using these index change values, the estimated coefficients of our model, average settlement prices for those two years, and accounting for the effect of lagged prices in our model, we computed rough estimates of the expected change in equilibrium futures prices (since these index changes are quite large, these figures are only approximate). That amounted, for example, to −$0.87/cwt and $2.07/cwt changes in price for 1998 and 2012, respectively. Applying these numbers to USDA total disappearance statistics, and factoring in the average yield of meat per animal, elevated levels of recalls cost cattle farmers roughly $454 million in 1998.3 On the opposite side, thanks to less frequent and serious recalls in 2012, farm-level revenues are estimated to have increased $1.07 billion nationwide. Such gains and losses, once transmitted to processors and retailers through higher or lower prices suggest the potential for more investment in food safety being profitable. Table 4 Economic Gains and Losses From Food Safety Date Annual index change % price change Avg yearly loss ($/cwt) Tot. beef cons. (Billion lbs.) Loss in dollars 12/30/1997 31 −0.27435 −0.18 25.61 −$93,480,435.31 12/29/1998 80.75 −0.71464 −0.46 26.31 −$240,211,084.78 12/28/1999 −8.75 0.077438 0.05 26.94 $27,519,866.97 12/26/2000 62 −0.5487 −0.39 27.34 −$210,661,302.34 12/26/2001 24.75 −0.21904 −0.16 27.03 −$86,377,785.50 12/31/2002 63 −0.55755 −0.39 27.88 −$215,281,809.15 12/30/2003 −40 0.354 0.28 27 $152,647,693.34 12/21/2004 −81.5 0.721275 0.6 27.75 $333,276,244.77 12/27/2005 −50.75 0.449138 0.39 27.75 $216,479,512.76 12/26/2006 −7.5 0.066375 0.06 28.14 $32,158,613.69 12/26/2007 7.5 −0.06638 −0.06 28.14 −$35,346,902.54 12/30/2008 −9 0.07965 0.08 27.19 $41,143,339.27 12/22/2009 33.75 −0.29869 −0.25 26.84 −$135,605,483.35 12/28/2010 5 −0.04425 −0.04 26.39 −$22,223,049.58 12/27/2011 −11.5 0.101775 0.12 25.54 $59,786,730.04 12/26/2012 −100.75 0.891638 1.1 25.75 $567,097,572.72 12/31/2013 −25.75 0.227888 0.29 25.48 $147,298,381.29 12/30/2014 −32.25 0.285413 0.43 24.68 $211,896,288.16 12/22/2015 33.25 −0.29426 −0.43 24.77 −$212,294,668.12 Date Annual index change % price change Avg yearly loss ($/cwt) Tot. beef cons. (Billion lbs.) Loss in dollars 12/30/1997 31 −0.27435 −0.18 25.61 −$93,480,435.31 12/29/1998 80.75 −0.71464 −0.46 26.31 −$240,211,084.78 12/28/1999 −8.75 0.077438 0.05 26.94 $27,519,866.97 12/26/2000 62 −0.5487 −0.39 27.34 −$210,661,302.34 12/26/2001 24.75 −0.21904 −0.16 27.03 −$86,377,785.50 12/31/2002 63 −0.55755 −0.39 27.88 −$215,281,809.15 12/30/2003 −40 0.354 0.28 27 $152,647,693.34 12/21/2004 −81.5 0.721275 0.6 27.75 $333,276,244.77 12/27/2005 −50.75 0.449138 0.39 27.75 $216,479,512.76 12/26/2006 −7.5 0.066375 0.06 28.14 $32,158,613.69 12/26/2007 7.5 −0.06638 −0.06 28.14 −$35,346,902.54 12/30/2008 −9 0.07965 0.08 27.19 $41,143,339.27 12/22/2009 33.75 −0.29869 −0.25 26.84 −$135,605,483.35 12/28/2010 5 −0.04425 −0.04 26.39 −$22,223,049.58 12/27/2011 −11.5 0.101775 0.12 25.54 $59,786,730.04 12/26/2012 −100.75 0.891638 1.1 25.75 $567,097,572.72 12/31/2013 −25.75 0.227888 0.29 25.48 $147,298,381.29 12/30/2014 −32.25 0.285413 0.43 24.68 $211,896,288.16 12/22/2015 33.25 −0.29426 −0.43 24.77 −$212,294,668.12 Table 4 Economic Gains and Losses From Food Safety Date Annual index change % price change Avg yearly loss ($/cwt) Tot. beef cons. (Billion lbs.) Loss in dollars 12/30/1997 31 −0.27435 −0.18 25.61 −$93,480,435.31 12/29/1998 80.75 −0.71464 −0.46 26.31 −$240,211,084.78 12/28/1999 −8.75 0.077438 0.05 26.94 $27,519,866.97 12/26/2000 62 −0.5487 −0.39 27.34 −$210,661,302.34 12/26/2001 24.75 −0.21904 −0.16 27.03 −$86,377,785.50 12/31/2002 63 −0.55755 −0.39 27.88 −$215,281,809.15 12/30/2003 −40 0.354 0.28 27 $152,647,693.34 12/21/2004 −81.5 0.721275 0.6 27.75 $333,276,244.77 12/27/2005 −50.75 0.449138 0.39 27.75 $216,479,512.76 12/26/2006 −7.5 0.066375 0.06 28.14 $32,158,613.69 12/26/2007 7.5 −0.06638 −0.06 28.14 −$35,346,902.54 12/30/2008 −9 0.07965 0.08 27.19 $41,143,339.27 12/22/2009 33.75 −0.29869 −0.25 26.84 −$135,605,483.35 12/28/2010 5 −0.04425 −0.04 26.39 −$22,223,049.58 12/27/2011 −11.5 0.101775 0.12 25.54 $59,786,730.04 12/26/2012 −100.75 0.891638 1.1 25.75 $567,097,572.72 12/31/2013 −25.75 0.227888 0.29 25.48 $147,298,381.29 12/30/2014 −32.25 0.285413 0.43 24.68 $211,896,288.16 12/22/2015 33.25 −0.29426 −0.43 24.77 −$212,294,668.12 Date Annual index change % price change Avg yearly loss ($/cwt) Tot. beef cons. (Billion lbs.) Loss in dollars 12/30/1997 31 −0.27435 −0.18 25.61 −$93,480,435.31 12/29/1998 80.75 −0.71464 −0.46 26.31 −$240,211,084.78 12/28/1999 −8.75 0.077438 0.05 26.94 $27,519,866.97 12/26/2000 62 −0.5487 −0.39 27.34 −$210,661,302.34 12/26/2001 24.75 −0.21904 −0.16 27.03 −$86,377,785.50 12/31/2002 63 −0.55755 −0.39 27.88 −$215,281,809.15 12/30/2003 −40 0.354 0.28 27 $152,647,693.34 12/21/2004 −81.5 0.721275 0.6 27.75 $333,276,244.77 12/27/2005 −50.75 0.449138 0.39 27.75 $216,479,512.76 12/26/2006 −7.5 0.066375 0.06 28.14 $32,158,613.69 12/26/2007 7.5 −0.06638 −0.06 28.14 −$35,346,902.54 12/30/2008 −9 0.07965 0.08 27.19 $41,143,339.27 12/22/2009 33.75 −0.29869 −0.25 26.84 −$135,605,483.35 12/28/2010 5 −0.04425 −0.04 26.39 −$22,223,049.58 12/27/2011 −11.5 0.101775 0.12 25.54 $59,786,730.04 12/26/2012 −100.75 0.891638 1.1 25.75 $567,097,572.72 12/31/2013 −25.75 0.227888 0.29 25.48 $147,298,381.29 12/30/2014 −32.25 0.285413 0.43 24.68 $211,896,288.16 12/22/2015 33.25 −0.29426 −0.43 24.77 −$212,294,668.12 For the nineteen years, changes in the level of recalls were responsible for decreasing revenue for nine years and increasing it for ten. Table 1 indicated that recalls soared from 2000 to 2002. It was also in these years that E. Coli O157: H7 was attributed to more recalls and contaminated more food products than any other time frame on record. Therefore, these results reflect the significant losses that farmers faced when the beef industry was experiencing distress during this time. In 2000, there were nineteen confirmed cases, nineteen likely cases, and forty-nine suspected cases of E. coli O157: H7 originating from several Wendy’s stores in Oregon (Knowlton 2000). In the same year a young girl died and 65 people fell ill from an outbreak stemming from Sizzler restaurants in Wisconsin. In 2002, the third largest recall in history occurred when ConAgra distributed ground beef contaminated with E. coli O157: H7. The outbreak sickened about nineteen people in six different states, forcing ConAgra to recall over 19 million pounds of ground beef. Another company the same year recalled over 2.8 million pounds of ground beef after fifty-seven people in seven states fell ill to E. coli O157: H7 contaminated ground beef. By examining the data, one can see that the only other years to result in losses comparable to 2000–2002 are the years that also experienced abnormally high E. coli O157: H7 outbreaks, such as 1998 and 2007. In 2007, the second-largest beef recall in history (21.7 million) occurred due to E. coli O157: H7 contamination in ground beef. Interestingly, the year with the most pounds recalled, 2008, benefitted economically at $41 million. This indicates that recall severity plays as fundamental a role in prices as the size and number of the recalls. Specifically, the results support the findings of Moghadam (2013), who finds E. coli O157: H7 recalls responsible for adverse returns in live cattle markets. Moreover, this study adds to his work by including some very large recalls that were excluded in the years studied in the previous literature. To ensure that the empirical results are not overly dependent on our method of index construction, we carried out sensitivity analysis on the weights used when aggregating recalls in the four six-month subperiods of our two-year rolling recall window. Results for the coefficient on the recall Index variable with alternative weighting schemes are shown in table 5, and appear suitably robust: all are statistically significant, negative, and within one-half of the estimated standard error for the original coefficient estimate. Table 5 Index Weight Robustness Check Results Weights Estimated coefficient t-value (1, 0.75, 0.50, 0.25) baseline −0.0000893 −2.53 (1, 0.50, 0.25, 0.125) −0.0000737 −2.21 (1, 0.67, 0.33, 0.17) −0.0000799 −2.32 (1, 1, 1, 1) −0.0000947 −3.40 Weights Estimated coefficient t-value (1, 0.75, 0.50, 0.25) baseline −0.0000893 −2.53 (1, 0.50, 0.25, 0.125) −0.0000737 −2.21 (1, 0.67, 0.33, 0.17) −0.0000799 −2.32 (1, 1, 1, 1) −0.0000947 −3.40 Table 5 Index Weight Robustness Check Results Weights Estimated coefficient t-value (1, 0.75, 0.50, 0.25) baseline −0.0000893 −2.53 (1, 0.50, 0.25, 0.125) −0.0000737 −2.21 (1, 0.67, 0.33, 0.17) −0.0000799 −2.32 (1, 1, 1, 1) −0.0000947 −3.40 Weights Estimated coefficient t-value (1, 0.75, 0.50, 0.25) baseline −0.0000893 −2.53 (1, 0.50, 0.25, 0.125) −0.0000737 −2.21 (1, 0.67, 0.33, 0.17) −0.0000799 −2.32 (1, 1, 1, 1) −0.0000947 −3.40 Further, we reran the model with the four six-month subperiods entered as separate indices, allowing the estimation algorithm to determine the optimal weights. Taking the resulting four estimated coefficients and normalizing by the first period estimated coefficient, we arrive at the weights that would maximize our model’s R2. Those weights are (1, 0.75, 0.64, and 0.67). The t-values on the estimated coefficients are −2.70, −2.23, −1.50, and −2.23, respectively, on the four subperiods from most recent to the longest lags. Thus, we conclude that the long-term, cumulative effects hypothesized in our original index are, in fact, real and statistically significant. This is in contrast to the finding of Piggott and Marsh (2004), who performed a similar test and found that effects of food safety information did not persist past one calendar quarter. Additional robustness checks were performed on the components of the index itself: the size and severity of the recalls. We constructed indices with only the severity of recalls component, only the size of recalls, with the weights on severity changed from (3, 2, 1) in the baseline index to (5, 3, 1), thus placing more weight on more severe recalls, and utilizing the pounds of recalls rather than the size classes in the baseline specification. The resulting coefficient estimates on the index are displayed in table 6. While the magnitude changes (because the mean index value is changing), the coefficient estimates remain negative throughout all these alternatives with roughly the same statistical precision. Table 6 Index Component Robustness Check Results Index component Estimated coefficient t-value Severity only −0.00029 −3.12 Size class only −0.00011 −1.93 Severity weights (5, 3, 1) −0.0000731 −3.16 Size in pounds, not size class −0.00030 −3.26 Index component Estimated coefficient t-value Severity only −0.00029 −3.12 Size class only −0.00011 −1.93 Severity weights (5, 3, 1) −0.0000731 −3.16 Size in pounds, not size class −0.00030 −3.26 Table 6 Index Component Robustness Check Results Index component Estimated coefficient t-value Severity only −0.00029 −3.12 Size class only −0.00011 −1.93 Severity weights (5, 3, 1) −0.0000731 −3.16 Size in pounds, not size class −0.00030 −3.26 Index component Estimated coefficient t-value Severity only −0.00029 −3.12 Size class only −0.00011 −1.93 Severity weights (5, 3, 1) −0.0000731 −3.16 Size in pounds, not size class −0.00030 −3.26 Having checked the sensitivity of our results to different specifications of the index in terms of shape, length of time, precise way of quantifying size and severity of recalls, and relative weighting of severity and poundage of the recall, we find the empirical results comfortingly robust. All the alternative specifications have qualitatively and quantitatively similar results, and when separated out even the final six month period (18–24 months) had a statistically significant and negative price impact. We still prefer our baseline index weights due to the more theoretically pleasing shape (recall memories fade within a few years), but adjusting those admittedly ad hoc assumptions has little effect on our results. Conclusions and Policy Implications This study investigates the gradual impact of food safety events and their surrounding negative publicity (measured through recalls) over an extended period of time, as was suggested by but not done in previous food safety literature. Our results offer important information for the cattle industry. Whereas before the impact of recalls on daily cattle prices was known and believed to be temporary, this research uncovers a more prolonged impact. Here, the results indicate not only a significant impact on weekly commodity prices, but they show that accumulated recalls act together. Thus, traders’ responses to a recall event are based on present and past knowledge of food safety events for the same product. Consumer theory applied to the interaction between traders and consumers suggests that consumers are substituting meat for other products in the short and medium term, or the market is experiencing a long-run structural change in meat caused by consumers’ perceptions of meat due to recalls. The structural change conclusion is consistent with Moschini and Meilke (1989). Previous literature used longer-term indices in meat demand models, focused on retail consumption, while effects on futures prices were modeled assuming only short-term effects. We have married these two bodies of literature by uncovering long-term effects at the commodity price level. Since futures prices reflect derived demand for future retail consumption, it makes sense that long-term consumption changes should be transmitted to futures markets. We provide results for the effect of beef recalls on the cattle industry over the past twenty years. Based on the results, recalls are costing cattle producers economically significant amounts of money when the losses to the industry as a whole are considered. Disaggregating these losses across individual firms may result in only marginal changes in farm revenue. Therefore, it is uncertain whether recall costs financially incentivize producers to invest in food safety safeguards, although it would certainly be an appropriate subject for research supported by check-off funds. Recall costs, both private and public, influence recall policy. By better understanding the costs of a recall, business owners can more accurately make investment decisions for food safety technologies and protocols. When you view cattle as an input in the production of beef, it seems safe to assume the input (cattle) price is lower because of a decline in consumer demand for beef. This leads to a downward shift in the derived demand for wholesale (processed) beef and for both live and feeder cattle. If processors and retailers operate on margins that are percentage mark-ups (as is common), while lower input costs might sound good, they will also lead to shrinking margins. That means that processors and retailers will share some of the cost of recalls in the form of lower prices, margins, and volumes. The new knowledge that recalls can adversely impact prices for as long as two years should encourage processors, in particular, to invest more heavily in food safety efforts to prevent recalls from happening. The fact that recall costs are shared across all industry participants and are long-lasting should encourage the industry to support efforts by the USDA to more stringently regulate food safety in the meat industry. Finally, because larger size recalls come with larger price impacts, the industry should support efforts toward traceability so that when food safety issues do arise, any resulting recalls will be as small as possible. More importantly, if this information influences companies to make even minor investments in food safety or changes to safety protocol, it could move the meat industry toward higher food safety standards, thus decreasing illnesses and death from contaminated beef. Footnotes 1 The previous literature (e.g., McKenzie and Thomsen 2001; Moghadam et al. 2013) typically used event study methods for estimation. To have an accurate measure of normal returns, surrounding recall events were dropped so that the abnormal return of one event did not influence the normal return estimation of another. This guaranteed the independence required for the test statistics; however, it potentially led to losing meaningful data from the analysis. 2 While it is certainly true, as pointed out by an associate editor, that recalls shift supply inward, the previous empirical research surveyed above has found negative price impacts to food safety incidents. This suggests that while we are really capturing a reduced form, or net, effect, the downward shift in demand dominates the inward shift in supply to produce a negative price impact overall. 3 Since beef disappearance is in pounds of meat and futures prices are in pounds of cattle, we need to convert between these two using an estimated yield for beef. This number is variable, but generally somewhere in the 40s or 50s in terms of percent (e.g., a heifer or steer yields 45% of its weight in retail beef pounds (Nold, 2013). We use 50% here to make our loss estimates somewhat conservative. References Attavanich W. , McCarl B.A. , Bessler D. . 2011 . The Effect of H1N1 (Swine Flu) Media Coverage on Agricultural Commodity Markets . Applied Economic Perspectives & Policy 33 2 : 241 – 59 . Google Scholar CrossRef Search ADS Burton M. , Young T. . 1997 . Measuring Meat Consumers’ Response to the Perceived Risks of BSE in Great Britain . Risk Decision and Policy 2 1 : 19 – 28 . Google Scholar CrossRef Search ADS Center for Disease Control and Prevention. 2016 . Estimates of Foodborne Illness in the United States. Available at: http://www.cdc.gov/foodborneburden/2011-foodborne-estimates.html. Accessed May 11, 2018. Henson S. , Mazzocchi M. . 2002 . Impact of Bovine Spongiform Encephalopathy on Agribusiness in the United Kingdom: Results of an Event Study of Equity Prices . American Journal of Agricultural Economics 84 2 : 241 – 58 . Google Scholar CrossRef Search ADS Kinnucan H.W. , Xiao H. , Hsia C. , Jackson J.D. . 1997 . Effects of Health Information and Generic Advertising on U.S. Meat Demand . American Journal of Agricultural Economics 79 1 : 13 – 23 . Google Scholar CrossRef Search ADS Knowlton S. 2000 . Meat Thought to be Coli Culprit: A Case at a Second Wendy’s in Tualatin has been Linked to the Outbreak. Available at: http://www.about-ecoli.com/ecoli_outbreaks/news/meat-thought-to-be-e-coli-culprit/. Accessed May 11, 2018. Lusk J.L. , Schroeder T.C. . 2002 . Effects of Meat Recalls on Futures Market Prices . Agricultural and Resource Economics Review 31 1 : 47 – 58 . Google Scholar CrossRef Search ADS Marsh T.L. , Schroeder T.C. , Mintert J. . 2004 . Impacts of Meat Product Recalls on Consumer Demand in the USA . Applied Economics 36 9 : 897 – 909 . Google Scholar CrossRef Search ADS McKenzie A.M. , Thomsen M.R. . 2001 . The Effect of E. Coli O157: H7 on Beef Prices . Journal of Agriculture and Resource Economics 26 2 : 431 – 44 . Moghadam A.K. , Schmidt C. , Grier K. . 2013 . The Impact of E. Coli O157: H7 Recalls on Live Cattle Futures Prices: Revisited . Food Policy 42 : 81 – 7 . Google Scholar CrossRef Search ADS Moschini G. , Meilke K.D. . 1989 . Modeling the Pattern of Structural Change in U.S. Meat Demand . American Journal of Agricultural Economics 71 2 : 253 – 61 . Google Scholar CrossRef Search ADS Nold R. 2013 . How Much Meat Can You Expect From a Fed Steer? South Dakota State University Extension. Available at: http://igrow.org/livestock/beef/how-much-meat-can-you-expect-from-a-fed-steer/. Accessed May 11, 2018. Patterson P. , Flake O. . 1999 . Health, Food Safety and Meat Demand . American Journal of Agricultural Economics 81 ( 5) : 1304 . Piggott N.E. , Marsh T.L. . 2004 . Does Food Safety Information Impact U.S. Meat Demand? American Journal of Agricultural Economics 86 1 : 154 – 74 . Google Scholar CrossRef Search ADS Pozo V.F. , Schroeder T.C. . 2016 . Evaluating the costs of meat and poultry recalls to food firms using stock returns . Food Policy 59 : 66 – 77 . Google Scholar CrossRef Search ADS Robenstein R.G. , Thurman W.N. . 1996 . Health Risk and the Demand for Red Meat: Evidence from Futures Markets . Review of Agricultural Economics 18 4 : 629 – 41 . Thomsen M.R. , McKenzie A.M. . 2001 . Market Incentives for Safe Foods: An Examination of Shareholder Losses from Meat and Poultry Recalls . American Journal of Agricultural Economics 82 3 : 526 – 38 . Google Scholar CrossRef Search ADS U.S. Department of Agriculture. 2015 . FSIS Food Recalls. Available at: https://www.fsis.usda.gov/wps/portal/fsis/topics/food-safety-education/get-answers/food-safety-fact-sheets/production-and-inspection/fsis-food-recalls/fsis-food-recalls. Accessed May 11, 2018. Schlenker W. , Villas-Boas S.B. . 2009 . Consumer and Market Responses to Mad Cow Disease . American Journal of Agricultural Economics 91 4 : 1140 – 52 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Economic Perspectives and Policy Oxford University Press

The Long-Term Effects of Meat Recalls on Futures Markets

Loading next page...
 
/lp/ou_press/the-long-term-effects-of-meat-recalls-on-futures-markets-eLwNGiRjPB
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
2040-5790
eISSN
2040-5804
D.O.I.
10.1093/aepp/ppy010
Publisher site
See Article on Publisher Site

Abstract

Abstract Over the past twenty years, there has been an increasing trend in the number of food safety events. Despite increased safety control standards, foodborne disease outbreaks continue to impact the food supply. A common source of foodborne illness and fatal infection is beef, with potential causes being E. coli 0157: H7, Listeria Monocytogenes, and Salmonella. Certain companies have even been bankrupted, unable to overcome the social costs and economic losses associated with recalls. We look beyond the impact on processors to see the amount of impact transmitted through price channels to cattle farmers. We examine beef recalls over a twenty-year period through an accumulated two-year index to see if there is a prolonged effect of recalls on current weekly cattle prices. We find that food safety events act together, adversely impacting prices and decreasing farm-level revenue. The finding that economic harm to the beef industry from food safety events lingers for at least two years suggests that the industry should do more to reduce food safety incidents. Cattle futures, food safety, futures market, recalls, revenue loss In the past 20 years, there has been considerable research into the effects of food safety events on consumer demand, likely due to the increase in the number of recalls over the same period. From 1996 to 2016, the meat and poultry industry witnessed an increasing trend in the number of recalls. USDA Food Safety and Inspection Service (FSIS 2015) reports that, despite industry and government efforts to safeguard food, product contamination by adulterants continues to be an ongoing concern. Food safety issues can result in severe economic losses to food companies and even larger social costs for society, ranging from company bankruptcy to consumer illness or death. The CDC estimates that each year roughly 1 in 6 Americans (or 48 million people) get sick, 128,000 are hospitalized, and 3,000 die from foodborne diseases (Center for Disease Control and Prevention 2016). Recalls are designed to protect consumers by removing contaminated product from commerce, but also serve a secondary purpose of incentivizing companies to prevent foodborne disease outbreaks from happening. It seems obvious that food safety events generally lead to negative retail price and revenue impacts because the recalls and surrounding media attention associated with them suggest lower product quality or ineffective quality control, resulting in a loss of consumer confidence (Marsh, Schroeder, and Mintert 2004), lower expected utility from consumption, diminished sales, and lower consumer prices. In turn, traders in derivatives markets react to the sudden, negative news, resulting in a drop in futures prices for the implicated commodity. This makes sense as a decline in retail demand will transmit itself through the supply chain and result in reduced derived demand for the original farm- level products that are ingredients in the processed food being recalled. Thus, mistakes made by food processors can lead to economic losses for farmers. While there has been considerable research examining the effects of recalls on short-term price movements (McKenzie and Thomsen 2001; Lusk and Schroeder 2002; Moghadam, Schmidt, and Grier 2013), there is no research on the cumulative impact that food safety events play in shifting long-term consumer demand preferences. Lusk and Schroeder (2002) concluded that if meat recalls resulted in a systematic change in the structural demand for meat, the change only occurs gradually and must be examined through an extended period of time. If food safety events do have a more prolonged effect, then knowing this would help identify the effectiveness of recall policy. Therefore, this paper aims to build on existing literature by investigating the long-term impact of recalls on weekly live cattle futures markets using an author-constructed recall index. More specifically, the objective of this paper is to analyze if additional meat recalls have long-run impacts on weekly live cattle futures prices, and thus on farmers’ incomes. We find that food safety events do have surprisingly long-lasting impacts and thus, processors might want to invest more heavily in preventing food safety events from occurring. The rest of the paper is organized as follows. The next section presents a literature review related to the long-run impacts of recalls on futures market prices. This is followed by an account of the meat recall process, then the data used are described. Our empirical methodology is then explained, followed by our empirical results. Conclusions and policy implications are discussed in the last section. Literature Review Much of the early literature, while not explicitly studying the impacts of food safety events, identified health information as a demand factor in consumer response. Moschini and Meilke (1989) introduced modeling the structural change in demand for meat. In their seminal paper, these authors found that the dynamics of prices and incomes cannot fully explain consumer demand shifts from meat to chicken; however, dietary concerns provided the best explanation for shifts towards white meats. Kinnucan et al. (1997) examined the combined effects of advertising and health information on meat demand. These authors hypothesized that health information, such as linkages between cholesterol and heart disease, impacted beef demand, which was confirmed by the positive and negative cross- and own-price elasticity for poultry and beef, respectively. The first study conducted to determine the short-run impact of health information was done by Robenstein and Thurman (1996), which revealed insignificant effects of Wall Street Journal articles on livestock futures markets. Food safety information differs, however, from health information in that such information, for example pathogenic contamination, can result in short-term, long-term, or fatal consequences. These events are mostly unanticipated and the shock severity may cause a considerable market reaction. The shock of a food recall can cause a lack of consumer confidence, drive prices down, and ultimately impact traders’ decisions, which has been the focus of much research. For instance, Henson and Mazzocchi (2002) found that processors of beef, dairy, and pet food products were negatively affected by reported linkages between BSE and Creutzfeldt–Jacob disease. Thomsen and McKenzie (2001) studied the effects of recalls on shareholder wealth to find that serious health concerns did cause a drop in stock prices. However, in a second paper, McKenzie and Thomsen (2001) concluded that wholesale prices were impacted by recalls, yet this impact did not transmit to farm-level derivative prices. On the other hand, using food recalls as a proxy for food quality, Lusk and Schroeder (2002) found that medium-sized, serious food recalls did impact short-run farm-level derivatives prices for live cattle and lean hogs. Furthermore, Moghadam, Schmidt, and Grier (2013) revisited the McKenzie and Thomsen (2001) result to find that there was, in fact, a significant reaction in live cattle futures markets from E. coli 0157: H7 contamination. The researchers of the more recent publication concluded that modeling the role of recalls on futures prices was incomplete without the introduction of gradual effects. This finding suggests further investigation into the long-term effect of meat recalls. A number of studies explore the long-run effects of food safety on meat demand, many of which utilized food safety indexes for empirical methodology. One of the principal studies in this branch of research is Burton and Young (1997), which used media articles and popular press to measure the impact of food safety on beef demand. In their AIDS model, these authors found a short- and long-term effect of publicity on consumer expenditure for meat and a drop in meat market share. Patterson and Flake (1999) extended the literature concerning the effects of BSE, E. coli, and salmonella on meat demand by studying Associated Press articles. Both of these papers found a negative impact on demand, confirming that the market responds to food safety occurrences, and informational indexes have effective explanatory power. Piggott and Marsh (2004) incorporated similar empirical methodology with a food safety index for publicized recalls. Overall, their paper revealed small changes over time in average consumer response to food safety, with some periods of large, short-lived impacts. Marsh, Schroeder, and Mintert (2004) built upon this research by examining demand responses to food recalls from newspaper articles and FSIS meat product recalls. While meat product recalls significantly impacted demand, with elasticity measures indicating a change from meat to non-meat consumption, media indices had no effect. The early literature analyzed prices following USDA announcements, while more recent studies have incorporated media articles as a primary source of consumer information and decision-making. Schlenker and Villa-Boas (2009) compared the two information sources and found that independent news media created a more adverse impact on prices than a government report warning of the first BSE outbreak. Nevertheless, government reports on food scares are typically the first information that traders and consumers have access to. Therefore, it is still believed that USDA recall announcements contain relevant information to gauge how market participants respond to food safety events. This study will follow the recent literature in that an index measure will be used to quantify the food safety impact. However, being that information contained in the reports (recall size, severity and description) has consistently yielded significant results, the index will only include information that market participants could access through the USDA announcements. The existing literature has succeeded in determining the immediate market response of meat recalls on futures market prices. It is hypothesized in this paper that current cattle prices reflect many factors, one of which being food recalls from months (or years) past. It is then believed that the indexed effect of cumulative recalls will give a more accurate representation of how traders update prices in response to new outbreaks. Specifically, the results will show the revenue lost by cattle producers due to a cumulative, time- and severity-weighted index of recall events over the past two years. This is possible through using a rolling index, allowing for the inclusion of all recalls within a 24-month time window. The inclusion of all recalls removes a limitation in some previous work that dropped some recall data to achieve identification.1 Recall Background The FSIS is an agency of the USDA that handles product recalls for meat, poultry, and egg products. According to FSIS’s recall policy, a recall can begin in several ways. First, the company can identify an adulterant or other product defect and notify USDA officials immediately. The FSIS will investigate the claim and issue a Recall Notification Report (RNR) for non-serious, Class III recalls or publish a press release if the recall is serious (Class I or II). The recall process can also begin with FSIS, the state government, or third-party plant inspectors identifying a non-compliance issue that warrants product removal (USDA FSIS). The RNR or Recall Release contains all the information concerning the recall, such as company, reason for recall, date of recall, method of discovery, recall class, pounds recalled, reported consumer illness, and other related information. In addition to the reason for the recall and product, each report notifies the public of the recall severity by categorizing the event into one of three groups: Class I, Class II, or Class III. According to the FSIS website, “A Class I recall involves a health hazard situation in which there is a reasonable probability that eating the food will cause health problems or death,” (USDA FSIS). Class II involves a contaminant that presents a remote probability of health concern, and Class III involves situations expected to result in no negative health side effects. After a report has been issued on the size and class of the recall, the FSIS works with the company throughout the recall process to ensure all contaminated products are removed from commerce before closing the recall. Data Using Datastream software, weekly live cattle futures prices were gathered from the Chicago Mercantile Exchange (CME). The CME contains information for different commodities, contracts and frequency. Specifically, live cattle contracts were presented in daily, weekly, and monthly increments for the expiration months of February, April, June, August, October, and December. A frequency of weekly prices better fits the frequency of recall announcements and abstracts from the noise of daily price changes. To organize prices into a continuous series, contracts were rolled over to the next nearest contract at the start of the expiration month. Following the logic of the previous literature, it is believed that futures prices—specifically, live cattle futures—will respond most strongly to a USDA recall. The recall information was pulled from the USDA’s FSIS website catalogue of meat, poultry, and egg recalls from 1982 through 2017. The Recall Case Archive provides summary information on completed recalls beginning in 1994. We limit our examination of recalls to the period from January 1, 1995 to December 31, 2015. The information contained in each data frame is the press release information listed above, including company, reason for recall, date of recall, method of discovery, recall class, pounds recalled, and consumer illness—if there was one. For the purpose of this study, the information extracted and used for analysis will be company, date, recall class, and pounds recalled for beef recalls. Beef recalls were selected because they remain a common source of food contamination, which the literature has already closely analyzed. Since recalls often necessitate the removal of multiple meat products, the criterion was restricted to include only beef-specific events. A summary of these recalls is shown in table 1. Table 1 Number and Volume of Beef Recalls per Year Year All recalls # of recalls Total (lbs.) Mean (lbs.) Std. Dev. (lbs.) 1995 6 963,956.00 160,659.33 176,190.02 1996 7 171,734.00 24,533.43 56,337.99 1997 10 26,322,529.00 2,632,252.90 7,861,897.91 1998 18 4,946,491.00 274,805.06 635,835.08 1999 14 3,396,154.00 242,582.43 557,261.94 2000 25 2,925,476.00 117,019.04 230,259.44 2001 29 1,600,682.00 55,195.93 154,198.14 2002 41 24,133,913.00 588,632.02 2,981,869.68 2003 23 1,776,945.00 80,770.23 170,399.92 2004 14 1,509,515.00 107,822.50 153,897.00 2005 11 1,740,982.00 158,271.09 269,905.06 2006 16 5,021,153.00 313,822.06 1,069,512.08 2007 26 33,485,915.00 1,287,919.81 4,350,286.16 2008 20 150,549,632.00 8,363,868.44 33,720,083.09 2009 29 3,777,014.00 130,241.86 227,842.61 2010 26 24,561,504.00 944,673.23 3,092,721.06 2011 27 1,248,938.00 49,957.52 94,650.12 2012 12 124,777.00 10,398.08 15,055.14 2013 15 382,915.00 27,351.07 37,939.40 2014 9 11,253,356.00 1,250,372.89 2,870,911.44 2015 26 1,077,624.00 43,104.96 107,043.80 Year All recalls # of recalls Total (lbs.) Mean (lbs.) Std. Dev. (lbs.) 1995 6 963,956.00 160,659.33 176,190.02 1996 7 171,734.00 24,533.43 56,337.99 1997 10 26,322,529.00 2,632,252.90 7,861,897.91 1998 18 4,946,491.00 274,805.06 635,835.08 1999 14 3,396,154.00 242,582.43 557,261.94 2000 25 2,925,476.00 117,019.04 230,259.44 2001 29 1,600,682.00 55,195.93 154,198.14 2002 41 24,133,913.00 588,632.02 2,981,869.68 2003 23 1,776,945.00 80,770.23 170,399.92 2004 14 1,509,515.00 107,822.50 153,897.00 2005 11 1,740,982.00 158,271.09 269,905.06 2006 16 5,021,153.00 313,822.06 1,069,512.08 2007 26 33,485,915.00 1,287,919.81 4,350,286.16 2008 20 150,549,632.00 8,363,868.44 33,720,083.09 2009 29 3,777,014.00 130,241.86 227,842.61 2010 26 24,561,504.00 944,673.23 3,092,721.06 2011 27 1,248,938.00 49,957.52 94,650.12 2012 12 124,777.00 10,398.08 15,055.14 2013 15 382,915.00 27,351.07 37,939.40 2014 9 11,253,356.00 1,250,372.89 2,870,911.44 2015 26 1,077,624.00 43,104.96 107,043.80 Table 1 Number and Volume of Beef Recalls per Year Year All recalls # of recalls Total (lbs.) Mean (lbs.) Std. Dev. (lbs.) 1995 6 963,956.00 160,659.33 176,190.02 1996 7 171,734.00 24,533.43 56,337.99 1997 10 26,322,529.00 2,632,252.90 7,861,897.91 1998 18 4,946,491.00 274,805.06 635,835.08 1999 14 3,396,154.00 242,582.43 557,261.94 2000 25 2,925,476.00 117,019.04 230,259.44 2001 29 1,600,682.00 55,195.93 154,198.14 2002 41 24,133,913.00 588,632.02 2,981,869.68 2003 23 1,776,945.00 80,770.23 170,399.92 2004 14 1,509,515.00 107,822.50 153,897.00 2005 11 1,740,982.00 158,271.09 269,905.06 2006 16 5,021,153.00 313,822.06 1,069,512.08 2007 26 33,485,915.00 1,287,919.81 4,350,286.16 2008 20 150,549,632.00 8,363,868.44 33,720,083.09 2009 29 3,777,014.00 130,241.86 227,842.61 2010 26 24,561,504.00 944,673.23 3,092,721.06 2011 27 1,248,938.00 49,957.52 94,650.12 2012 12 124,777.00 10,398.08 15,055.14 2013 15 382,915.00 27,351.07 37,939.40 2014 9 11,253,356.00 1,250,372.89 2,870,911.44 2015 26 1,077,624.00 43,104.96 107,043.80 Year All recalls # of recalls Total (lbs.) Mean (lbs.) Std. Dev. (lbs.) 1995 6 963,956.00 160,659.33 176,190.02 1996 7 171,734.00 24,533.43 56,337.99 1997 10 26,322,529.00 2,632,252.90 7,861,897.91 1998 18 4,946,491.00 274,805.06 635,835.08 1999 14 3,396,154.00 242,582.43 557,261.94 2000 25 2,925,476.00 117,019.04 230,259.44 2001 29 1,600,682.00 55,195.93 154,198.14 2002 41 24,133,913.00 588,632.02 2,981,869.68 2003 23 1,776,945.00 80,770.23 170,399.92 2004 14 1,509,515.00 107,822.50 153,897.00 2005 11 1,740,982.00 158,271.09 269,905.06 2006 16 5,021,153.00 313,822.06 1,069,512.08 2007 26 33,485,915.00 1,287,919.81 4,350,286.16 2008 20 150,549,632.00 8,363,868.44 33,720,083.09 2009 29 3,777,014.00 130,241.86 227,842.61 2010 26 24,561,504.00 944,673.23 3,092,721.06 2011 27 1,248,938.00 49,957.52 94,650.12 2012 12 124,777.00 10,398.08 15,055.14 2013 15 382,915.00 27,351.07 37,939.40 2014 9 11,253,356.00 1,250,372.89 2,870,911.44 2015 26 1,077,624.00 43,104.96 107,043.80 The reasons for recalls range from labeling to bacteria contamination. From 1995 through 2015, the leading cause of beef recalls was E.coli O157:H7. The second-leading reason for a recall fell in the category “Other”, which included misbranding, extraneous materials, or a non-life-threatening chemical/spoilage exposure. Therefore, the data reflect higher Class I and Class II/III recalls for the former and latter, respectively. Table 2 displays a chart with the various types of recall reasons and number of occurrences for each. Table 2 Type of Recalls 1995–2015 All recalls Year Listeria monocytogenes E.coli O157: H7 Allergen Other bacteria Other Processing 1995 0 5 0 0 1 1 1996 3 2 0 0 2 2 1997 0 6 0 0 4 0 1998 1 12 0 1 4 1 1999 2 9 0 0 3 2 2000 5 18 0 0 2 0 2001 5 21 1 1 1 1 2002 3 31 1 1 5 2 2003 3 10 1 1 8 1 2004 2 5 4 0 2 1 2005 3 5 1 1 1 0 2006 0 8 2 1 5 0 2007 1 22 0 0 3 2 2008 2 15 1 0 2 0 2009 4 13 1 3 8 0 2010 2 11 4 2 7 2 2011 2 12 5 2 6 2 2012 1 3 1 1 6 0 2013 1 5 3 2 4 1 2014 0 1 2 0 6 2 2015 1 7 9 0 9 0 Total 41 221 36 16 89 20 All recalls Year Listeria monocytogenes E.coli O157: H7 Allergen Other bacteria Other Processing 1995 0 5 0 0 1 1 1996 3 2 0 0 2 2 1997 0 6 0 0 4 0 1998 1 12 0 1 4 1 1999 2 9 0 0 3 2 2000 5 18 0 0 2 0 2001 5 21 1 1 1 1 2002 3 31 1 1 5 2 2003 3 10 1 1 8 1 2004 2 5 4 0 2 1 2005 3 5 1 1 1 0 2006 0 8 2 1 5 0 2007 1 22 0 0 3 2 2008 2 15 1 0 2 0 2009 4 13 1 3 8 0 2010 2 11 4 2 7 2 2011 2 12 5 2 6 2 2012 1 3 1 1 6 0 2013 1 5 3 2 4 1 2014 0 1 2 0 6 2 2015 1 7 9 0 9 0 Total 41 221 36 16 89 20 Table 2 Type of Recalls 1995–2015 All recalls Year Listeria monocytogenes E.coli O157: H7 Allergen Other bacteria Other Processing 1995 0 5 0 0 1 1 1996 3 2 0 0 2 2 1997 0 6 0 0 4 0 1998 1 12 0 1 4 1 1999 2 9 0 0 3 2 2000 5 18 0 0 2 0 2001 5 21 1 1 1 1 2002 3 31 1 1 5 2 2003 3 10 1 1 8 1 2004 2 5 4 0 2 1 2005 3 5 1 1 1 0 2006 0 8 2 1 5 0 2007 1 22 0 0 3 2 2008 2 15 1 0 2 0 2009 4 13 1 3 8 0 2010 2 11 4 2 7 2 2011 2 12 5 2 6 2 2012 1 3 1 1 6 0 2013 1 5 3 2 4 1 2014 0 1 2 0 6 2 2015 1 7 9 0 9 0 Total 41 221 36 16 89 20 All recalls Year Listeria monocytogenes E.coli O157: H7 Allergen Other bacteria Other Processing 1995 0 5 0 0 1 1 1996 3 2 0 0 2 2 1997 0 6 0 0 4 0 1998 1 12 0 1 4 1 1999 2 9 0 0 3 2 2000 5 18 0 0 2 0 2001 5 21 1 1 1 1 2002 3 31 1 1 5 2 2003 3 10 1 1 8 1 2004 2 5 4 0 2 1 2005 3 5 1 1 1 0 2006 0 8 2 1 5 0 2007 1 22 0 0 3 2 2008 2 15 1 0 2 0 2009 4 13 1 3 8 0 2010 2 11 4 2 7 2 2011 2 12 5 2 6 2 2012 1 3 1 1 6 0 2013 1 5 3 2 4 1 2014 0 1 2 0 6 2 2015 1 7 9 0 9 0 Total 41 221 36 16 89 20 Index Construction Significant consideration and research was taken in developing our index. Lusk and Schroeder (2002) determined that seriousness and size of recall determine market response. Other papers also identified recall class and size as key variables in price response to food safety events. Pozo and Schroeder (2016) found recall size to be significant, causing abnormal stock returns to become more negative when increased by 1%. The literature consistently finds that recalls involving serious health consequences or death have significant impact on financial variables (Thomsen and McKenzie 2001; Pozo and Schroeder 2016). Therefore, this study treats recall size and severity as primary measures for the index construction, yielding a higher index value for more serious recalls. We might have tried to incorporate media coverage into the index construction, but decided that converting media mentions into a quantitative measure was too subjective. Following Lusk and Schroeder (2002), this paper includes size categories as follows: size1 = less than 1,162 lbs., size2 = between 1,162 and 4,516 lbs., size3 = between 4,516 and 32,000 lbs., size4 = between 32,000 and 175,288 lbs., and size5 = greater than 175,288 lbs. For each of these categories, we assign an index of odd numbers 1 through 9, with 9 being assigned to the category with the most pounds recalled. Next, we index the classes by 3, 2, and 1 for Class I, II and III, respectively, giving a larger weight to events more likely to cause a serious illness. These two numbers are then summed to arrive at the contribution to the index for each recall event. While most previous research focused on the short-term impact of recalls and food safety information, we are taking a long-term view, hypothesizing that consumers build up an impression on the safety of foods from cumulative exposure to media coverage of food safety incidents such as recalls. Piggott and Marsh (2004) did test for such an effect in a meat demand system, and found no impact of food safety information on prices beyond a calendar quarter. Thus, if our hypothesis is correct, our results will differ from their findings. Clearly, any such longer-term, cumulative index is ad hoc in nature. Therefore, sensitivity analysis will be conducted on the length and shape that should be imposed on the index that reflect the rate of decay for past information’s effect on cattle prices. We explain the baseline index specification now and then report sensitivity analysis findings in the empirical results section. The two-year frame for our baseline index was divided into four subperiods that consist of six months of recall data for each group. All of the recalls in a six-month period were summed together to find the total index measure per six months. Each subperiod is then weighted decreasingly by distance from the current week. The most recent six months have a weight of one, the following six months have a weight of 0.75, and the remaining two six-month subperiods receive weights of 0.5 and 0.25, respectively. Then all the weighted values are summed to arrive at the final index value for that week. In turn, this window will move through the entire dataset of recalls, re-indexing every week, creating an index based on a rolling window of two years of recalls. We calculated the index from December 31, 1996 through December 22, 2015, totaling 989 weekly observations. Empirical Methodology Based on demand theory, we treat the food safety index as a demand shifter.2 This method was successfully used by Marsh, Schroeder, and Mintert (2004) and Attavanich, McCarl, and Bessler (2011). We use a simple autoregressive model of live cattle futures prices with the addition of several exogenous variables. The explanatory variables included in this study were chosen from previous literature and economic theory, specifically as it relates to food recall events. The main exogenous variable consists of a two-year accumulated recall index. Since beef, like most agricultural commodities, demonstrates seasonality, the empirical model also includes dummy variables for months, and yearly fixed effects are included to account for inflation and trends in tastes and preferences. The logarithm of the price series was used as the dependent variable and when tested for nonstationarity with an augmented Dickey-Fuller test was found to be stationary. Two lagged values of logged futures prices were included as regressors to account for past prices influencing present prices. Modeled autocorrelation and partial autocorrelation graphs justify the specification. Thus, the equation used to model prices is given by lnPt=β0+β1Indext+β2ln⁡Pt-1+β3ln⁡Pt-2+∑j=212θjMonthjt+∑k=219δkYearkt+ɛt (1) where P is the nearby futures price with t = 1,…, 886 denoting the week in our sample, Index denotes the accumulated index, Monthjt is a set of monthly dummy variables, the Yearkt are the year dummies to provide yearly fixed effects, and ɛ is the stochastic error term. Empirical Results Equation (1) is estimated using OLS. The main variable of interest, Index, is statistically significant at the 5% level and the coefficient for the index variable is negative (−0.0000893), which supports our hypothesis that an accumulation of recalls will put downward pressure on cattle prices. Table 3 provides the full regression results for our model. Table 3 Regression Results Log (cattle price) Coefficient Standard error p values Intercept 0.47414*** 0.072 <0.001 Index −0.00009** 0.000 0.011 Log(P)t-1 0.86462*** 0.035 <0.001 Log(P)t-2 0.02359 0.033 0.475 Year  1998 −0.00323 0.004 0.455  1999 0.01002** 0.005 0.034  2000 0.01865*** 0.006 0.001  2001 0.02420*** 0.007 0.001  2002 0.02318** 0.008 0.003  2003 0.03662*** 0.010 <0.001  2004 0.04039*** 0.008 <0.001  2005 0.03700*** 0.007 <0.001  2006 0.03269*** 0.007 <0.001  2007 0.04833*** 0.008 <0.001  2008 0.05361*** 0.010 <0.001  2009 0.04062*** 0.008 <0.001  2010 0.06293*** 0.011 <0.001  2011 0.07920*** 0.013 <0.001  2012 0.08032*** 0.012 <0.001  2013 0.07764*** 0.012 <0.001  2014 0.09908*** 0.014 <0.001  2015 0.09028*** 0.015 <0.001 Month  February 0.00519 0.004 0.155  March 0.00124 0.003 0.702  April −0.01381*** 0.004 <0.001  May −0.00379 0.003 0.260  June −0.00089 0.003 0.798  July 0.00002 0.003 0.995  August 0.00653* 0.004 0.068  September 0.00419 0.003 0.187  October 0.00935*** 0.003 0.004  November 0.00442 0.004 0.211  December 0.00607 0.004 0.124 R2 0.9927 Log (cattle price) Coefficient Standard error p values Intercept 0.47414*** 0.072 <0.001 Index −0.00009** 0.000 0.011 Log(P)t-1 0.86462*** 0.035 <0.001 Log(P)t-2 0.02359 0.033 0.475 Year  1998 −0.00323 0.004 0.455  1999 0.01002** 0.005 0.034  2000 0.01865*** 0.006 0.001  2001 0.02420*** 0.007 0.001  2002 0.02318** 0.008 0.003  2003 0.03662*** 0.010 <0.001  2004 0.04039*** 0.008 <0.001  2005 0.03700*** 0.007 <0.001  2006 0.03269*** 0.007 <0.001  2007 0.04833*** 0.008 <0.001  2008 0.05361*** 0.010 <0.001  2009 0.04062*** 0.008 <0.001  2010 0.06293*** 0.011 <0.001  2011 0.07920*** 0.013 <0.001  2012 0.08032*** 0.012 <0.001  2013 0.07764*** 0.012 <0.001  2014 0.09908*** 0.014 <0.001  2015 0.09028*** 0.015 <0.001 Month  February 0.00519 0.004 0.155  March 0.00124 0.003 0.702  April −0.01381*** 0.004 <0.001  May −0.00379 0.003 0.260  June −0.00089 0.003 0.798  July 0.00002 0.003 0.995  August 0.00653* 0.004 0.068  September 0.00419 0.003 0.187  October 0.00935*** 0.003 0.004  November 0.00442 0.004 0.211  December 0.00607 0.004 0.124 R2 0.9927 Note: *significant at 10%, **significant at 05%, ***significant at 01% Table 3 Regression Results Log (cattle price) Coefficient Standard error p values Intercept 0.47414*** 0.072 <0.001 Index −0.00009** 0.000 0.011 Log(P)t-1 0.86462*** 0.035 <0.001 Log(P)t-2 0.02359 0.033 0.475 Year  1998 −0.00323 0.004 0.455  1999 0.01002** 0.005 0.034  2000 0.01865*** 0.006 0.001  2001 0.02420*** 0.007 0.001  2002 0.02318** 0.008 0.003  2003 0.03662*** 0.010 <0.001  2004 0.04039*** 0.008 <0.001  2005 0.03700*** 0.007 <0.001  2006 0.03269*** 0.007 <0.001  2007 0.04833*** 0.008 <0.001  2008 0.05361*** 0.010 <0.001  2009 0.04062*** 0.008 <0.001  2010 0.06293*** 0.011 <0.001  2011 0.07920*** 0.013 <0.001  2012 0.08032*** 0.012 <0.001  2013 0.07764*** 0.012 <0.001  2014 0.09908*** 0.014 <0.001  2015 0.09028*** 0.015 <0.001 Month  February 0.00519 0.004 0.155  March 0.00124 0.003 0.702  April −0.01381*** 0.004 <0.001  May −0.00379 0.003 0.260  June −0.00089 0.003 0.798  July 0.00002 0.003 0.995  August 0.00653* 0.004 0.068  September 0.00419 0.003 0.187  October 0.00935*** 0.003 0.004  November 0.00442 0.004 0.211  December 0.00607 0.004 0.124 R2 0.9927 Log (cattle price) Coefficient Standard error p values Intercept 0.47414*** 0.072 <0.001 Index −0.00009** 0.000 0.011 Log(P)t-1 0.86462*** 0.035 <0.001 Log(P)t-2 0.02359 0.033 0.475 Year  1998 −0.00323 0.004 0.455  1999 0.01002** 0.005 0.034  2000 0.01865*** 0.006 0.001  2001 0.02420*** 0.007 0.001  2002 0.02318** 0.008 0.003  2003 0.03662*** 0.010 <0.001  2004 0.04039*** 0.008 <0.001  2005 0.03700*** 0.007 <0.001  2006 0.03269*** 0.007 <0.001  2007 0.04833*** 0.008 <0.001  2008 0.05361*** 0.010 <0.001  2009 0.04062*** 0.008 <0.001  2010 0.06293*** 0.011 <0.001  2011 0.07920*** 0.013 <0.001  2012 0.08032*** 0.012 <0.001  2013 0.07764*** 0.012 <0.001  2014 0.09908*** 0.014 <0.001  2015 0.09028*** 0.015 <0.001 Month  February 0.00519 0.004 0.155  March 0.00124 0.003 0.702  April −0.01381*** 0.004 <0.001  May −0.00379 0.003 0.260  June −0.00089 0.003 0.798  July 0.00002 0.003 0.995  August 0.00653* 0.004 0.068  September 0.00419 0.003 0.187  October 0.00935*** 0.003 0.004  November 0.00442 0.004 0.211  December 0.00607 0.004 0.124 R2 0.9927 Note: *significant at 10%, **significant at 05%, ***significant at 01% The results indicate that a single recall, even of sizable and serious proportions, cannot increase the index enough to push farm-level prices downward by an economically significant amount. However, the accumulation of recalls does add up to a price effect, transmitted through the supply chain, that is large enough to matter to farmers, processors, and retailers. To show that beef industry actors are being impacted enough to induce action, we need to put these results more clearly in context. Economic Impact Fewer recalls means increased farm revenue, according to our results, but how much? As one way to put the numbers in context, we computed the annual change in the index each year of our sample. The annual index change in table 4 below gives the cumulative change for one year of recall information. The recall index increased the most in 1998 (+80.75), while its biggest drop was in 2012 (-100.75). Using these index change values, the estimated coefficients of our model, average settlement prices for those two years, and accounting for the effect of lagged prices in our model, we computed rough estimates of the expected change in equilibrium futures prices (since these index changes are quite large, these figures are only approximate). That amounted, for example, to −$0.87/cwt and $2.07/cwt changes in price for 1998 and 2012, respectively. Applying these numbers to USDA total disappearance statistics, and factoring in the average yield of meat per animal, elevated levels of recalls cost cattle farmers roughly $454 million in 1998.3 On the opposite side, thanks to less frequent and serious recalls in 2012, farm-level revenues are estimated to have increased $1.07 billion nationwide. Such gains and losses, once transmitted to processors and retailers through higher or lower prices suggest the potential for more investment in food safety being profitable. Table 4 Economic Gains and Losses From Food Safety Date Annual index change % price change Avg yearly loss ($/cwt) Tot. beef cons. (Billion lbs.) Loss in dollars 12/30/1997 31 −0.27435 −0.18 25.61 −$93,480,435.31 12/29/1998 80.75 −0.71464 −0.46 26.31 −$240,211,084.78 12/28/1999 −8.75 0.077438 0.05 26.94 $27,519,866.97 12/26/2000 62 −0.5487 −0.39 27.34 −$210,661,302.34 12/26/2001 24.75 −0.21904 −0.16 27.03 −$86,377,785.50 12/31/2002 63 −0.55755 −0.39 27.88 −$215,281,809.15 12/30/2003 −40 0.354 0.28 27 $152,647,693.34 12/21/2004 −81.5 0.721275 0.6 27.75 $333,276,244.77 12/27/2005 −50.75 0.449138 0.39 27.75 $216,479,512.76 12/26/2006 −7.5 0.066375 0.06 28.14 $32,158,613.69 12/26/2007 7.5 −0.06638 −0.06 28.14 −$35,346,902.54 12/30/2008 −9 0.07965 0.08 27.19 $41,143,339.27 12/22/2009 33.75 −0.29869 −0.25 26.84 −$135,605,483.35 12/28/2010 5 −0.04425 −0.04 26.39 −$22,223,049.58 12/27/2011 −11.5 0.101775 0.12 25.54 $59,786,730.04 12/26/2012 −100.75 0.891638 1.1 25.75 $567,097,572.72 12/31/2013 −25.75 0.227888 0.29 25.48 $147,298,381.29 12/30/2014 −32.25 0.285413 0.43 24.68 $211,896,288.16 12/22/2015 33.25 −0.29426 −0.43 24.77 −$212,294,668.12 Date Annual index change % price change Avg yearly loss ($/cwt) Tot. beef cons. (Billion lbs.) Loss in dollars 12/30/1997 31 −0.27435 −0.18 25.61 −$93,480,435.31 12/29/1998 80.75 −0.71464 −0.46 26.31 −$240,211,084.78 12/28/1999 −8.75 0.077438 0.05 26.94 $27,519,866.97 12/26/2000 62 −0.5487 −0.39 27.34 −$210,661,302.34 12/26/2001 24.75 −0.21904 −0.16 27.03 −$86,377,785.50 12/31/2002 63 −0.55755 −0.39 27.88 −$215,281,809.15 12/30/2003 −40 0.354 0.28 27 $152,647,693.34 12/21/2004 −81.5 0.721275 0.6 27.75 $333,276,244.77 12/27/2005 −50.75 0.449138 0.39 27.75 $216,479,512.76 12/26/2006 −7.5 0.066375 0.06 28.14 $32,158,613.69 12/26/2007 7.5 −0.06638 −0.06 28.14 −$35,346,902.54 12/30/2008 −9 0.07965 0.08 27.19 $41,143,339.27 12/22/2009 33.75 −0.29869 −0.25 26.84 −$135,605,483.35 12/28/2010 5 −0.04425 −0.04 26.39 −$22,223,049.58 12/27/2011 −11.5 0.101775 0.12 25.54 $59,786,730.04 12/26/2012 −100.75 0.891638 1.1 25.75 $567,097,572.72 12/31/2013 −25.75 0.227888 0.29 25.48 $147,298,381.29 12/30/2014 −32.25 0.285413 0.43 24.68 $211,896,288.16 12/22/2015 33.25 −0.29426 −0.43 24.77 −$212,294,668.12 Table 4 Economic Gains and Losses From Food Safety Date Annual index change % price change Avg yearly loss ($/cwt) Tot. beef cons. (Billion lbs.) Loss in dollars 12/30/1997 31 −0.27435 −0.18 25.61 −$93,480,435.31 12/29/1998 80.75 −0.71464 −0.46 26.31 −$240,211,084.78 12/28/1999 −8.75 0.077438 0.05 26.94 $27,519,866.97 12/26/2000 62 −0.5487 −0.39 27.34 −$210,661,302.34 12/26/2001 24.75 −0.21904 −0.16 27.03 −$86,377,785.50 12/31/2002 63 −0.55755 −0.39 27.88 −$215,281,809.15 12/30/2003 −40 0.354 0.28 27 $152,647,693.34 12/21/2004 −81.5 0.721275 0.6 27.75 $333,276,244.77 12/27/2005 −50.75 0.449138 0.39 27.75 $216,479,512.76 12/26/2006 −7.5 0.066375 0.06 28.14 $32,158,613.69 12/26/2007 7.5 −0.06638 −0.06 28.14 −$35,346,902.54 12/30/2008 −9 0.07965 0.08 27.19 $41,143,339.27 12/22/2009 33.75 −0.29869 −0.25 26.84 −$135,605,483.35 12/28/2010 5 −0.04425 −0.04 26.39 −$22,223,049.58 12/27/2011 −11.5 0.101775 0.12 25.54 $59,786,730.04 12/26/2012 −100.75 0.891638 1.1 25.75 $567,097,572.72 12/31/2013 −25.75 0.227888 0.29 25.48 $147,298,381.29 12/30/2014 −32.25 0.285413 0.43 24.68 $211,896,288.16 12/22/2015 33.25 −0.29426 −0.43 24.77 −$212,294,668.12 Date Annual index change % price change Avg yearly loss ($/cwt) Tot. beef cons. (Billion lbs.) Loss in dollars 12/30/1997 31 −0.27435 −0.18 25.61 −$93,480,435.31 12/29/1998 80.75 −0.71464 −0.46 26.31 −$240,211,084.78 12/28/1999 −8.75 0.077438 0.05 26.94 $27,519,866.97 12/26/2000 62 −0.5487 −0.39 27.34 −$210,661,302.34 12/26/2001 24.75 −0.21904 −0.16 27.03 −$86,377,785.50 12/31/2002 63 −0.55755 −0.39 27.88 −$215,281,809.15 12/30/2003 −40 0.354 0.28 27 $152,647,693.34 12/21/2004 −81.5 0.721275 0.6 27.75 $333,276,244.77 12/27/2005 −50.75 0.449138 0.39 27.75 $216,479,512.76 12/26/2006 −7.5 0.066375 0.06 28.14 $32,158,613.69 12/26/2007 7.5 −0.06638 −0.06 28.14 −$35,346,902.54 12/30/2008 −9 0.07965 0.08 27.19 $41,143,339.27 12/22/2009 33.75 −0.29869 −0.25 26.84 −$135,605,483.35 12/28/2010 5 −0.04425 −0.04 26.39 −$22,223,049.58 12/27/2011 −11.5 0.101775 0.12 25.54 $59,786,730.04 12/26/2012 −100.75 0.891638 1.1 25.75 $567,097,572.72 12/31/2013 −25.75 0.227888 0.29 25.48 $147,298,381.29 12/30/2014 −32.25 0.285413 0.43 24.68 $211,896,288.16 12/22/2015 33.25 −0.29426 −0.43 24.77 −$212,294,668.12 For the nineteen years, changes in the level of recalls were responsible for decreasing revenue for nine years and increasing it for ten. Table 1 indicated that recalls soared from 2000 to 2002. It was also in these years that E. Coli O157: H7 was attributed to more recalls and contaminated more food products than any other time frame on record. Therefore, these results reflect the significant losses that farmers faced when the beef industry was experiencing distress during this time. In 2000, there were nineteen confirmed cases, nineteen likely cases, and forty-nine suspected cases of E. coli O157: H7 originating from several Wendy’s stores in Oregon (Knowlton 2000). In the same year a young girl died and 65 people fell ill from an outbreak stemming from Sizzler restaurants in Wisconsin. In 2002, the third largest recall in history occurred when ConAgra distributed ground beef contaminated with E. coli O157: H7. The outbreak sickened about nineteen people in six different states, forcing ConAgra to recall over 19 million pounds of ground beef. Another company the same year recalled over 2.8 million pounds of ground beef after fifty-seven people in seven states fell ill to E. coli O157: H7 contaminated ground beef. By examining the data, one can see that the only other years to result in losses comparable to 2000–2002 are the years that also experienced abnormally high E. coli O157: H7 outbreaks, such as 1998 and 2007. In 2007, the second-largest beef recall in history (21.7 million) occurred due to E. coli O157: H7 contamination in ground beef. Interestingly, the year with the most pounds recalled, 2008, benefitted economically at $41 million. This indicates that recall severity plays as fundamental a role in prices as the size and number of the recalls. Specifically, the results support the findings of Moghadam (2013), who finds E. coli O157: H7 recalls responsible for adverse returns in live cattle markets. Moreover, this study adds to his work by including some very large recalls that were excluded in the years studied in the previous literature. To ensure that the empirical results are not overly dependent on our method of index construction, we carried out sensitivity analysis on the weights used when aggregating recalls in the four six-month subperiods of our two-year rolling recall window. Results for the coefficient on the recall Index variable with alternative weighting schemes are shown in table 5, and appear suitably robust: all are statistically significant, negative, and within one-half of the estimated standard error for the original coefficient estimate. Table 5 Index Weight Robustness Check Results Weights Estimated coefficient t-value (1, 0.75, 0.50, 0.25) baseline −0.0000893 −2.53 (1, 0.50, 0.25, 0.125) −0.0000737 −2.21 (1, 0.67, 0.33, 0.17) −0.0000799 −2.32 (1, 1, 1, 1) −0.0000947 −3.40 Weights Estimated coefficient t-value (1, 0.75, 0.50, 0.25) baseline −0.0000893 −2.53 (1, 0.50, 0.25, 0.125) −0.0000737 −2.21 (1, 0.67, 0.33, 0.17) −0.0000799 −2.32 (1, 1, 1, 1) −0.0000947 −3.40 Table 5 Index Weight Robustness Check Results Weights Estimated coefficient t-value (1, 0.75, 0.50, 0.25) baseline −0.0000893 −2.53 (1, 0.50, 0.25, 0.125) −0.0000737 −2.21 (1, 0.67, 0.33, 0.17) −0.0000799 −2.32 (1, 1, 1, 1) −0.0000947 −3.40 Weights Estimated coefficient t-value (1, 0.75, 0.50, 0.25) baseline −0.0000893 −2.53 (1, 0.50, 0.25, 0.125) −0.0000737 −2.21 (1, 0.67, 0.33, 0.17) −0.0000799 −2.32 (1, 1, 1, 1) −0.0000947 −3.40 Further, we reran the model with the four six-month subperiods entered as separate indices, allowing the estimation algorithm to determine the optimal weights. Taking the resulting four estimated coefficients and normalizing by the first period estimated coefficient, we arrive at the weights that would maximize our model’s R2. Those weights are (1, 0.75, 0.64, and 0.67). The t-values on the estimated coefficients are −2.70, −2.23, −1.50, and −2.23, respectively, on the four subperiods from most recent to the longest lags. Thus, we conclude that the long-term, cumulative effects hypothesized in our original index are, in fact, real and statistically significant. This is in contrast to the finding of Piggott and Marsh (2004), who performed a similar test and found that effects of food safety information did not persist past one calendar quarter. Additional robustness checks were performed on the components of the index itself: the size and severity of the recalls. We constructed indices with only the severity of recalls component, only the size of recalls, with the weights on severity changed from (3, 2, 1) in the baseline index to (5, 3, 1), thus placing more weight on more severe recalls, and utilizing the pounds of recalls rather than the size classes in the baseline specification. The resulting coefficient estimates on the index are displayed in table 6. While the magnitude changes (because the mean index value is changing), the coefficient estimates remain negative throughout all these alternatives with roughly the same statistical precision. Table 6 Index Component Robustness Check Results Index component Estimated coefficient t-value Severity only −0.00029 −3.12 Size class only −0.00011 −1.93 Severity weights (5, 3, 1) −0.0000731 −3.16 Size in pounds, not size class −0.00030 −3.26 Index component Estimated coefficient t-value Severity only −0.00029 −3.12 Size class only −0.00011 −1.93 Severity weights (5, 3, 1) −0.0000731 −3.16 Size in pounds, not size class −0.00030 −3.26 Table 6 Index Component Robustness Check Results Index component Estimated coefficient t-value Severity only −0.00029 −3.12 Size class only −0.00011 −1.93 Severity weights (5, 3, 1) −0.0000731 −3.16 Size in pounds, not size class −0.00030 −3.26 Index component Estimated coefficient t-value Severity only −0.00029 −3.12 Size class only −0.00011 −1.93 Severity weights (5, 3, 1) −0.0000731 −3.16 Size in pounds, not size class −0.00030 −3.26 Having checked the sensitivity of our results to different specifications of the index in terms of shape, length of time, precise way of quantifying size and severity of recalls, and relative weighting of severity and poundage of the recall, we find the empirical results comfortingly robust. All the alternative specifications have qualitatively and quantitatively similar results, and when separated out even the final six month period (18–24 months) had a statistically significant and negative price impact. We still prefer our baseline index weights due to the more theoretically pleasing shape (recall memories fade within a few years), but adjusting those admittedly ad hoc assumptions has little effect on our results. Conclusions and Policy Implications This study investigates the gradual impact of food safety events and their surrounding negative publicity (measured through recalls) over an extended period of time, as was suggested by but not done in previous food safety literature. Our results offer important information for the cattle industry. Whereas before the impact of recalls on daily cattle prices was known and believed to be temporary, this research uncovers a more prolonged impact. Here, the results indicate not only a significant impact on weekly commodity prices, but they show that accumulated recalls act together. Thus, traders’ responses to a recall event are based on present and past knowledge of food safety events for the same product. Consumer theory applied to the interaction between traders and consumers suggests that consumers are substituting meat for other products in the short and medium term, or the market is experiencing a long-run structural change in meat caused by consumers’ perceptions of meat due to recalls. The structural change conclusion is consistent with Moschini and Meilke (1989). Previous literature used longer-term indices in meat demand models, focused on retail consumption, while effects on futures prices were modeled assuming only short-term effects. We have married these two bodies of literature by uncovering long-term effects at the commodity price level. Since futures prices reflect derived demand for future retail consumption, it makes sense that long-term consumption changes should be transmitted to futures markets. We provide results for the effect of beef recalls on the cattle industry over the past twenty years. Based on the results, recalls are costing cattle producers economically significant amounts of money when the losses to the industry as a whole are considered. Disaggregating these losses across individual firms may result in only marginal changes in farm revenue. Therefore, it is uncertain whether recall costs financially incentivize producers to invest in food safety safeguards, although it would certainly be an appropriate subject for research supported by check-off funds. Recall costs, both private and public, influence recall policy. By better understanding the costs of a recall, business owners can more accurately make investment decisions for food safety technologies and protocols. When you view cattle as an input in the production of beef, it seems safe to assume the input (cattle) price is lower because of a decline in consumer demand for beef. This leads to a downward shift in the derived demand for wholesale (processed) beef and for both live and feeder cattle. If processors and retailers operate on margins that are percentage mark-ups (as is common), while lower input costs might sound good, they will also lead to shrinking margins. That means that processors and retailers will share some of the cost of recalls in the form of lower prices, margins, and volumes. The new knowledge that recalls can adversely impact prices for as long as two years should encourage processors, in particular, to invest more heavily in food safety efforts to prevent recalls from happening. The fact that recall costs are shared across all industry participants and are long-lasting should encourage the industry to support efforts by the USDA to more stringently regulate food safety in the meat industry. Finally, because larger size recalls come with larger price impacts, the industry should support efforts toward traceability so that when food safety issues do arise, any resulting recalls will be as small as possible. More importantly, if this information influences companies to make even minor investments in food safety or changes to safety protocol, it could move the meat industry toward higher food safety standards, thus decreasing illnesses and death from contaminated beef. Footnotes 1 The previous literature (e.g., McKenzie and Thomsen 2001; Moghadam et al. 2013) typically used event study methods for estimation. To have an accurate measure of normal returns, surrounding recall events were dropped so that the abnormal return of one event did not influence the normal return estimation of another. This guaranteed the independence required for the test statistics; however, it potentially led to losing meaningful data from the analysis. 2 While it is certainly true, as pointed out by an associate editor, that recalls shift supply inward, the previous empirical research surveyed above has found negative price impacts to food safety incidents. This suggests that while we are really capturing a reduced form, or net, effect, the downward shift in demand dominates the inward shift in supply to produce a negative price impact overall. 3 Since beef disappearance is in pounds of meat and futures prices are in pounds of cattle, we need to convert between these two using an estimated yield for beef. This number is variable, but generally somewhere in the 40s or 50s in terms of percent (e.g., a heifer or steer yields 45% of its weight in retail beef pounds (Nold, 2013). We use 50% here to make our loss estimates somewhat conservative. References Attavanich W. , McCarl B.A. , Bessler D. . 2011 . The Effect of H1N1 (Swine Flu) Media Coverage on Agricultural Commodity Markets . Applied Economic Perspectives & Policy 33 2 : 241 – 59 . Google Scholar CrossRef Search ADS Burton M. , Young T. . 1997 . Measuring Meat Consumers’ Response to the Perceived Risks of BSE in Great Britain . Risk Decision and Policy 2 1 : 19 – 28 . Google Scholar CrossRef Search ADS Center for Disease Control and Prevention. 2016 . Estimates of Foodborne Illness in the United States. Available at: http://www.cdc.gov/foodborneburden/2011-foodborne-estimates.html. Accessed May 11, 2018. Henson S. , Mazzocchi M. . 2002 . Impact of Bovine Spongiform Encephalopathy on Agribusiness in the United Kingdom: Results of an Event Study of Equity Prices . American Journal of Agricultural Economics 84 2 : 241 – 58 . Google Scholar CrossRef Search ADS Kinnucan H.W. , Xiao H. , Hsia C. , Jackson J.D. . 1997 . Effects of Health Information and Generic Advertising on U.S. Meat Demand . American Journal of Agricultural Economics 79 1 : 13 – 23 . Google Scholar CrossRef Search ADS Knowlton S. 2000 . Meat Thought to be Coli Culprit: A Case at a Second Wendy’s in Tualatin has been Linked to the Outbreak. Available at: http://www.about-ecoli.com/ecoli_outbreaks/news/meat-thought-to-be-e-coli-culprit/. Accessed May 11, 2018. Lusk J.L. , Schroeder T.C. . 2002 . Effects of Meat Recalls on Futures Market Prices . Agricultural and Resource Economics Review 31 1 : 47 – 58 . Google Scholar CrossRef Search ADS Marsh T.L. , Schroeder T.C. , Mintert J. . 2004 . Impacts of Meat Product Recalls on Consumer Demand in the USA . Applied Economics 36 9 : 897 – 909 . Google Scholar CrossRef Search ADS McKenzie A.M. , Thomsen M.R. . 2001 . The Effect of E. Coli O157: H7 on Beef Prices . Journal of Agriculture and Resource Economics 26 2 : 431 – 44 . Moghadam A.K. , Schmidt C. , Grier K. . 2013 . The Impact of E. Coli O157: H7 Recalls on Live Cattle Futures Prices: Revisited . Food Policy 42 : 81 – 7 . Google Scholar CrossRef Search ADS Moschini G. , Meilke K.D. . 1989 . Modeling the Pattern of Structural Change in U.S. Meat Demand . American Journal of Agricultural Economics 71 2 : 253 – 61 . Google Scholar CrossRef Search ADS Nold R. 2013 . How Much Meat Can You Expect From a Fed Steer? South Dakota State University Extension. Available at: http://igrow.org/livestock/beef/how-much-meat-can-you-expect-from-a-fed-steer/. Accessed May 11, 2018. Patterson P. , Flake O. . 1999 . Health, Food Safety and Meat Demand . American Journal of Agricultural Economics 81 ( 5) : 1304 . Piggott N.E. , Marsh T.L. . 2004 . Does Food Safety Information Impact U.S. Meat Demand? American Journal of Agricultural Economics 86 1 : 154 – 74 . Google Scholar CrossRef Search ADS Pozo V.F. , Schroeder T.C. . 2016 . Evaluating the costs of meat and poultry recalls to food firms using stock returns . Food Policy 59 : 66 – 77 . Google Scholar CrossRef Search ADS Robenstein R.G. , Thurman W.N. . 1996 . Health Risk and the Demand for Red Meat: Evidence from Futures Markets . Review of Agricultural Economics 18 4 : 629 – 41 . Thomsen M.R. , McKenzie A.M. . 2001 . Market Incentives for Safe Foods: An Examination of Shareholder Losses from Meat and Poultry Recalls . American Journal of Agricultural Economics 82 3 : 526 – 38 . Google Scholar CrossRef Search ADS U.S. Department of Agriculture. 2015 . FSIS Food Recalls. Available at: https://www.fsis.usda.gov/wps/portal/fsis/topics/food-safety-education/get-answers/food-safety-fact-sheets/production-and-inspection/fsis-food-recalls/fsis-food-recalls. Accessed May 11, 2018. Schlenker W. , Villas-Boas S.B. . 2009 . Consumer and Market Responses to Mad Cow Disease . American Journal of Agricultural Economics 91 4 : 1140 – 52 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of the Agricultural and Applied Economics Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Applied Economic Perspectives and PolicyOxford University Press

Published: May 24, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off