0‘. .'-X\Av ._| :a fi‘ [0’ . «Id: (5. .11 S .1 - .XSRI 1| . .l.....:. t il.‘|\0 .. tot: . A. . ‘3‘ {in l...’ u I 1 u. . L ‘1 D1 1“ fl. m1...11.u..w.»s..... .3... -, .l . LII .. y. 2:4. Mimi. 5&1 . n 1 ’ . I l I u _ I(‘ . 11L G). {I [IFK .. ,\I .M» . > . |- U. l K' . l n 1“! 6.23.0-1: .1 svr-a? )5. 1.. 1 ......]L .p.. I). 41 'In I." ’t ’1‘“ W. 3“! 1 4.!)1vh9 . iguana u). b E... ,. 1.1...- w -— 1) \ , u 13”- 's L all I“ . . 2:311... . 1. .VISZEK 5 3.7.,1' L.) a. 1 ls... ;. .l‘ . 1 «, a). ‘ I: “(pl t.\ . F n’. :“‘."-‘-t. ;{;t"'.g ' ' 3‘ . to." . '4 Piv- 31-h. \..\l\b I: 11... q . fir . ~ . t n .3 2 y . : . .1 1» u. h, 4|‘b- II . 1 . . “Mm... hi9» H51. 1.1;. . . .t . 1, 41‘ 3159...‘ «L? :11“ 1...... 1 . w-.. .h .Jhnfi 1 .unotc‘1oncskit . II \ 1w!-I_ J“v..ur,1,1.b . 3.3.07.1}.00) ’99.): 3|!!! A .. .1 . 1 1 1 .51.!![5 .sb v’lnr..| u .{.Iirl 2-2.1; . . . .11 init. 1 1 A .. , va1o.v«l¢ 1.1 . . , 4 on ‘u. 4 .‘ . .. . . . Kb DrH|.avl.n\a1H.:l z \z u . . 1. . 1 .r . R, ‘ .. 145.3. .lnnvua..\...ur>h.Y-1..192 L- - LEERAEY 5 Michigan 5 ate University This is to certify that the thesis entitled CONSUMER RESPONSE TO RISK INFORMATION: A CASE STUDY OF THE IMPACT OF THE ALAR SCARE ON NEw YORK CITY FRESH APPLEfDEMAND presented by William Preston Guyton has been accepted towards fulfillment of the requirements for Masters degree in Science (Ag. Econ.) \az ° g' " 'yi Major professor Date January 31, 1990 07639 MS U i: an Affirmative Action/Equal Opportunity Institution PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or betore date due. DATE DUE DATE DUE DATE DUE 1'1 1.. J‘. 256T’IUi-LMO Eamon J! u w 1 we 1 - c ’71 rm 3 M- MSU Is An Affirmative Action/EquaI Opportunity Institution CONSUMER RESPONSE TO RISK INFORMATION: A CASE STUDY OF THE IMPACT OF THE ALAR SCARE ON NEW YORK CITY FRESH APPLE DEMAND BY William Preston Guyton A Thesis Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Agricultural Economics 1990 ABSTRACT CONSUMER RESPONSE TO RISK INFORMATION: A CASE STUDY OF THE IMPACT OF THE ALAR SCARE ON NEW YORK CITY FRESH APPLE DEMAND BY William Preston Guyton This study examined the impact of the U.S. Alar scare on the New York City retail fresh apple market. The principal objective of the research was to develop a framework for estimating consumer response to risk information. The effect over time of increasing amounts of risk information, the lagged effect of media coverage, and one time shifts in the demand curve were examined. Other objectives included estimating revenue losses to retailers in New York City and providing policy implications and suggestions for dealing with food scare incidents that could occur in the future. An econometric model of demand for fresh apples in New York City was constructed and monthly data collected from January 1980 to July 1989. Variables used in the model included own price, price of substitutes, seasonality, and a measure of risk information available to consumers about the Alar scare. The information variable was the monthly number of articles written in the New ‘1gr31 Times during the observation period. This variable was used as a proxy for media coverage of Alar. William Preston Guyton The major finding of the study was that the Alar incident was found to have influenced demand for fresh apples in New York City. It appears that demand fell as early as July of 1984 representing a one time sustained shift in demand during the observation period. ACKNOWLEDGEMENTS There are many people who helped make this thesis possible. First and foremost, I would like to thank my major professor Dr. Eileen 0. van Ravenswaay for her support and inspiration in developing this study. She, along with my thesis committee professors Dr. John Hoehn and Dr. Richard Baillie, provided econometric and statistical advice on how to conduct a consumer response study to risk information. I would also like to extend my gratitude to Lih-Chyun Sun for his suggestions on improving and modifying econometric models and for his patience in explaining distributed lag models. His encouragement during this study was appreciated. Sedef Birkan provided assistance with data collection. The funding for this project was provided by a cooperative research agreement (58-3J23-1-0334X) between the Economic Research Service USDA and.Michigan State University. Their financial support is greatly appreciated. I would also like to thank Doug Edwards and Lynne Ericksen of the USDA, Market News Service for providing data and advice on apple shipments and production in the United States. ii TABLE OF CONTENTS LIST OF TABLES O O O O O O O O O O O O O O O O O O O 0 iv LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . v Chapter 1. INTRODUCTION . . . . . . . . . . . . . l 1.1 Background and Chronology . . . . . . . 2 1. 2 Summary . . . . . . . . . . . . . . . . 8 2. THEORY . . . . . . . . . . . . . . . . . . . 11 2.1 Conceptual Framework of Consumer Response 11 2.2 Existing Empirical Evidence . . . . . . 14 2.2.1 Summary . . . . . . . . . . . . . 20 2.3 Hypotheses . . . . . . . . . . . . . . . 20 2.4 Revenue Losses to New York Retailers . . 21 3. DATA AND METHODS . . . . . . . . . . . . . . 24 3.1 DATA . . . . . . . . . . . . . . . . . . 24 3.1.1 Consumption . . . . . . . . . . . 24 3.1.2 Population . . . . . . . . . . . 26 3.1.3 Prices . . . . . . . . . . . . . 27 3.1.4 Income . . . . . . . . . . . . . 29 3.1.5 Consumer Awareness . . . . . . . 30 3.1.6 Other Variables . . . . . . . . . 32 3.2 METHODS . . . . . . . . . . . . . . . . 32 4. ECONOMETRIC FINDINGS . . . . . . . . 35 4.1 Selection of Fresh Apple Substitute . . 36 4.2 Treatment of Income . . . . . . . . . 36 4.3 Selection of Media Coverage Variables. . 37 4.4 Model Estimation . . . . . . . . . . . . 38 4.5 Model Estimation Results . . . . . . . . 39 4.6 Estimation of Lost Revenue . . . . . . . 42 5. SUMMARY AND IMPLICATIONS . . . . . . . . . . 50 5.1 Methods . . . . . . . . . . . . . . . . 50 5.2 Estimates of Revenue Losses . . . . . . 52 5.3 Policy Implications . . . . . . . . . . 52 5.4 Further Research . . . . . . . . . . . . 53 APPENDIX ONE . . . . . . . . . . . . . . . . . . . . . 58 APPENDIX TWO . . . . . . . . . . . . . . . . . . . . . 71 REFERENCES . . . . . . . . . . . . . . . . . . . . . . 76 TABLE 4.1 TABLE 4.2 TABLE 4.3 TABLE 4.4 LIST OF TABLES iv Page 37 39 40 43 LIST OF FIGURES Page FIGURE 1 . l . . . . . . . . . . . . . . . . . . . . . 10 FIGURE 2 . 1 . . . . . . . . . . . . . . . . . . . . . 23 FIGURE 3.1 . . . . . . . . . . . . . . . . . . . . . 25 FIGURE 4.1 . . . . . . . . . . . . . . . . . . . . . 45 FIGURE 4.2 . . . . . . . . . . . . . . . . . . . . . 46 FIGURE 4 . 3 . . . . . . . . . . . . . . . . . . . . . 47 FIGURE 4 . 4 . . . . . . . . . . . . . . . . . . . . . 48 FIGURE 4.5 O O O O O O O O O O O O O O O O O O O O O 49 CHAPTER ONE INTRODUCTION In recent years, consumer's lack of confidence in both government and the agricultural sector to set and comply with guidelines for "acceptable" levels of pesticide residues has resulted in economic losses to agricultural producers and consumers. Examples include the EDB contamination of grain products in 1983, heptachlor contamination of fluid milk in Hawaii and Arkansas, and most recently, the Alar scare in the U.S. apple market. Understanding why such food safety scares occur and the extent of their economic consequences will help prevent similar losses in the future. Consequently, this study examines the case of Alar and its impact on apple sales. Specifically, the thesis examines the extent and duration of consumer response to the Alar scare using an econometric model of market demand for fresh apples in New York City. New York City was chosen as a test site due to the availability of the most complete and comprehensive pricing and media coverage data. Another reason was that New York City represents a large and diverse cross section of con- sumers. A chronology of the Alar incident based primarily on analysis of newspaper coverage about Alar follows in Section 1.1. Chapter Two presents the theoretical approach for estimating and examining the Changes in consumer demand for 1 2 fresh apples. In Chapter Three, the econometric methods and data used in the study are presented. Estimation results from the econometric model of apple demand are presented in Chapter Four. Chapter Five summarizes the study, examines policy implications and offers suggestions for future research. as o d C re 0 o The Environmental Protection Agency requires chemicals used to enhance agricultural production to pass stringent toxicological tests before being permitted onto the market. Although newly registered agricultural chemicals have been carefully examined and evaluated for both acute and chronic toxicity, many of the older pesticides were registered prior to the availability of chronic toxicity data (e.g., on carcinogenicity). To date, only a third of 600 older active ingredients used to produce nearly 50,000 commercial pes- ticides have been completely reviewed by the agency (Lag Angels; Times, May 22, 1988). Daminozide, or Alar as it is called by its trade name, is one of these older chemical products that was registered as a pesticide in 1963 under the Federal Insecticide, Fun- gicide, and Redenticide Act (Cooperative Extension Service Bulletin, Michigan State University, May 1989) . In 1968, Alar was first marketed by Uniroyal Chemical Company as a growth regulator in apple production but has also been used to a lesser extent on other agricultural products including cherries, nectarines, peaches, Concord grapes, and peanuts. 3 Apple growers claim that there are major economic benefits from using the growth regulator which include better fruit set and maturity, fruit firmness and coloring, avoidance of preharvest drop, and increased shelf-life during storage (Cooperative Extension Service Bulletin, Michigan State University, May 1989). Alar is a systemic spray which is usually applied three times a year; in the spring, early summer, and six to eight weeks before harvest. Since the chemical penetrates the fruit surface, washing will not cleanse apples of Alar residues. Before 1985, one million pounds of the substance was sprayed annually on an estimated 40 percent of the U.S. apple crop (Lgs Anggles Times, May 22, 1988). Since then, the E.P.A. reports that apple growers have substantially reduced Alar use. The spray has been used primarily on red varieties of apples such as Red Delicious, McIntosh, and Stayman although it has also been applied to Golden Delicious apples as well (New 312;); Times. Aug. 30, 1985). E.P.A. authorities were prompted to reassess Alar in July of 1984 after numerous studies, conducted in the 1970's, linked daminozide and its derivative UDMH to cancer in laboratory animals. Both daminozide and UDMH were labeled as "possible carcinogens” following a risk assessment made by the agency's Office of Pesticide Programs (0.P.P.) and the E.P.A. recommended cancelling Alar's registration effective at the beginning of the 1986 growing season. The proposal to ban Alar was rejected, however, in 1985 by the E.P.A. 's Scientific 4 Advisory Panel (S.A.P.), a group of expert scientists who are assigned to evaluate the agency's regulations. The S.A.P. contended that there was insufficient data and.methodological problems with the Alar toxicological tests. As an alternative to a ban, the E.P.A. chose to lower the tolerance level of Alar in apples from 30 to 20 parts per million and required , Uniroyal to conduct private long-term studies for carcinogeni- city. The chemical manufacturer was also obligated to measure concentrations of both daminozide and UDMH in the apple supply (Consumer Reports, May 1989). Dissatisfied by the E.P.A.'s decision to permit the use of Alar, several states including Massachusetts, New York, and Maine, petitioned in the spring of 1986 to prohibit farmers from spraying daminozide on their crops. State officials, supported by the National Resource Defense Council, prepared to file a suit if the E.P.A. refused to comply with their requests. The E.P.A. denied demands to change its 1985 decision on Alar until there was sufficient proof that the growth regulator posed an imminent health risk to humans. The agency contested that full test results, expected in 1990, would determine whether or not daminozide should be banned. In mid—1986, four of the nation's largest supermarket chains responded to growing pressure from state, consumer, and environmental groups by refusing to purchase fresh apples that had been treated with Alar. Retail food stores began advertising "Alar free" apples and conducting tests to measure Alar residues in an attempt to restore consumer confidence in 5 their fresh produce. As the boycott spread to grocery stores and processors nationwide, growers who continued to spray their orchards with daminozide found it increasingly difficult to market their apples. Uniroyal reported that daminozide sales in 1987 were down 75 percent from the previous two years indicating that the industry boycott had been effective in _ discouraging Alar use. The voluntary ban on Alar resulted in economic losses to many industry participants, particularly in certain states. Despite an abundance of apples on the market during the 1986 and 1987 harvest season, apple growers throughout New York and New England reported significant harvest losses. Apples matured earlier than expected, rain disrupted picking, and because many growers refrained from spraying Alar, preharvest drop became a critical problem. The New York apple crop was 10 percent below average in October of 1987 which many attributed, at least in part, to the Alar boycott (New 2913 Times. Oct. 5, 1987). In February of 1988, the Alar controversy was fueled by allegations that Safeway supermarkets in California were violating promises to consumers by selling apples with Alar residues. An independent laboratory in Oakland detected traces of daminozide in apples sold in Los Angeles and Sacramento that were supposedly "Alar free". Safeway responded to the accusations by assuring the public that "the levels measured in apples is very small and there is no imminent threat to health" (New York Times, Feb. 3, 1988) . 6 Disagreements among federal and state officials over the health risks of Alar continued to attract national attention for the next year. Massachusetts, dissatisfied with E.P.A. standards, adapted its own guidelines of no more than one part per million for daminozide residues. Other states also considered imposing stricter standards on the growth regula- , tor. The government continued to receive harsh criticism from consumer and environmental groups on the slow process taken to evaluate and restrict the use of the potentially harmful agricultural chemical. In February of 1989, the National Resource Defense Council (N.R.D.C.), a nonprofit environmental and health group, released its "Intolerable Risk" report which warned of the health dangers from exposure to Alar and seven other possible carcinogenic chemicals commonly sprayed on fresh produce. The N.R.D.C. study concluded that preschool children have a greater risk than adults of contracting cancer from chemicals in their diets. This was attributed to the facts that children are more susceptible to the effects of toxic chemicals and relative to their weight, they consume greater quantities of fruits and vegetables. Of the carcinogenic chemicals examined in the study, the council found daminozide to pose the greatest health risk to children (New York Times, Feb. 25, 1989). During the same month, "60 Minutes", a nationally televised show on CBS with a viewing audience of 30 million people, aired a news segment warning the public of the dangers from 7 exposure to daminozide. Based on information from the N.R.D.C. study, "60 Minutes" targeted Alar as "the most potent cancer-causing agent in our food supply" (ghigaqg Tribune, March 26, 1989) . Extensive press coverage in local and national newspapers, consumer reports, and news magazines followed the "60 Minutes" report. In addition to the Alar . dispute, the concurrent story of cyanide contamination of Chilean grapes prompted both 11M and Newsweek to feature cover stories on the food safety controversy in the final week of February. The E.P.A., Food and Drug Administration, and the U.S. Department of Agriculture responded publicly, disputing the news reports that Alar was a health risk and claiming that there was no conclusive evidence to date that the growth regulator was an imminent health hazard. In March of 1989, concerned by the reports of possible health risks to children, many public school systems includ- ing New York City, Los Angeles, San Francisco, Chicago, and Atlanta banned apples and apple products from school lunch programs. By the end of the month, however, apples returned to school cafeterias after fruit growers and distributors assured school officials that they would not sell Alar-treated apples. Although confidence in apples was restored in public schools, shippers and consumers were increasingly reluctant to purchase apples and apple products. Apple growers in Washington reported a drop in wholesale prices from $15 to $12 for a 42 pound box of Red Delicious in the last two weeks 8 of March and the International Apple Institute (I.A.I.), an industry trade association, reported that apple sales were down 20 percent from the previous year. The apple industry responded at first with advertisement campaigns in defense of the use of Alar, but later withdrew the ads. By May, the Washington Apple Commission predicted that .expected losses to growers from the Alar scare would exceed $100 million in lost sales and lower prices (The Backer, May 1989). In order to allay public concerns and curb further losses, the apple industry announced plans to voluntarily bar the use of Alar. The I.A.I. predicted that as a result Of this ban, Alar residues in fresh apples would be virtually zero by the fall of 1989. Uniroyal Chemical Company agreed to comply with the E.P.A.'s request to stop sales of all Alar products and to recall existing stocks. Terms of the volun- tary ban included reimbursements to all growers who returned their daminozide products to dealers and distributors (Cooper- ative Extension Service, Michigan State University, June 20, 1989). The E.P.A. announced final steps to ban the use of Alar in May of 1989, but no action was likely to take effect until early 1991. Meanwhile, in an attempt to aid the apple industry, the federal government agreed to purchase an estimated $9.5 million worth of fresh apples. .2 umma From the chronology, important conclusions may be drawn about the Alar scare. First of all, the controversy covered 9 a period of years, beginning in 1984 and continuing through 1989. During this period, news coverage grew and peaked in February of 1989 when the National Resource Defense Council published their "Intolerable Risk" report. Figure 1.1 shows the number of articles written per month about Alar in the NE! 12;; Iimss. Monthly number of articles in the figs 19;; . Iimss are listed in Appendix One. Second, consumers appear to have responded to risk information reported in 1989, but the impact of Alar on consumption prior to this time is unclear. Third, three important events characterized the Alar scare: l. the July 1984 E.P.A. reassessment of Alar, 2. the July 1986 grocery store boycott of Alar-treated apples, 3. and the February 1989 "60 Minutes" report. Each of these events potentially had an effect on the consumption of apples in the U.S. 10 FIGURE 1 . 1 Number of Articles Per Month in the New York Times About Alar 38 imes 25'} —-—:—-— 201 15 10 ”n 5 0L ‘LHHL‘L‘L A A” 1984 I 1985 1986 1987 1988 1989 Months Number of Articles in New York T CHAPTER TWO THEORY This chapter develops the theoretical framework used to examine the impact of the Alar incident on fresh apple demand and for estimating the resulting sales losses. Previous studies of the effects of health warnings on consumer demand for a product are reviewed in the next two sections. Section 2.2 then conceptualizes sales losses from a food scare. 1 one tua ewo 0 sum 0 se Neoclassical economic theory states that demand for goods is derived from individuals' utility functions. Consumers are assumed to maximize their utility for different commodities subject to income and price constraints. At a given point in time, quantity of a good demanded is a function of its own price, holding other factors constant. Over time, however, other factors may vary and thus, quantity demanded of a good becomes a function of its own price (Po) , price of substitutes (P.) , income (I), changes in tastes and preferences, and changes in information about the product (e.g. advertising). At the market level, population and the process of information diffusion are additional factors that will influence quantity demanded and so the demand equation may be defined as follows: Q = f(P°, P I, Population, Information) 'I 11 12 For the individual, utility received from a particular good depends on the goods' characteristics. When information about a good changes, a resulting change in the products' characteristics also occurs. This in turn will affect an individual's utility function and subsequently, purchases of the commodity. At the market level, this process becomes , more complex since individuals receive and respond to informa- tion about the good at different times. Thus, in order to incorporate information in a market demand model, it is necessary to have an understanding of both the characteristics of information and how it is diffused to the public. In studies of consumer response to a product warning, the effect of changes in risk information on changes in consump- tion of an adulterated good is analyzed. The length of the risk exposure, duration of the risk information, and the duration of the effect of information should all be considered when developing a conceptual framework for consumer response to food safety uncertainty. Both length of risk exposure and duration of risk information may be categorized as temporary or persistent. For example, in the case of heptachlor contamination of Oahu milk in 1982 (Smith, van Ravenswaay, and Thompson, 1989), consumers were exposed to a banned agricultural chemical over a period of several months. When residues of heptachlor were detected in the milk supply, government and industry responded by dumping the contaminated milk and reassuring the public that existing supplies were safe to drink. Thus, the health 13 risk was temporary. News coverage of the incident, however, persisted for sixteen months following the contamination, so . although the risk: was eliminated, the duration. of risk information continued. In this case, an initial drop in consumption would be expected followed by a gradual restora- tion of pre-contamination sales levels as media coverage . eventually declined and consumers regained confidence in the adulterated good. Conversely, before and during the Alar scare, daminozide was legally registered and widely used by apple growers. Over a period of years, consumers of apples and apple products were continually exposed to a suspected carcinogenic chemical. The health risks and media coverage of Alar persisted over five years as state and federal health officials debated whether daminozide should be banned. In such a case, a reduction in apple consumption would be expected to occur during the entire period. Restoration to normal consumption would not be ex- pected until the health risk was eliminated. Another factor to consider when analyzing risk information is for how long a period information affects consumers in the market. Following news coverage of a food scare, risk in- formation may have a permanent or cumulative effect. In other words, consumers continue to respond to increasing amounts of risk information. Another response to risk information may be a lagged media effect which may occur if information about the health risk is gradually forgotten over time. A related question is whether consumers respond to the amount of risk 14 information available or are simply sensitive to its presence or absence. For Alar, risk information was transmitted over a period beginning in July of 1984 and peaking nearly five years later in March of 1989. As an increasing number of articles were written about Alar, a growing segment of the public became . informed of the health risks of exposure to the growth regulator. Consequently, sales of apples declined. Thus, one would expect that the cumulative effects of news coverage about daminozide would reduce apple purchases over time as consumers responded to increasing amounts of risk information. As media coverage declined, consumers would be expected to forget about the health risks, thus creating a lagged media effect. Past studies of consumer response to product warnings have 'considered these characteristics about risk information when determining sales or welfare losses. The following section presents existing empirical evidence to support this theoreti- cal framework. 3.; zgisting Espirical figidence In 1969, Joseph Brown analyzed the effect of consumer knowledge of potentially dangerous pesticide residues in cranberries on changes in price elasticity of demand. Brown hypothesized that advertising and awareness of the food scare would have opposite effects on price elasticity. The case study involved the 1959 announcement by the Department of 15 Health, Education, and Welfare that cranberries contained residues of amino triazole, a suspected carcinogen. A linear model was estimated regressing cranberry consump- tion on price, income, and dummy variables that represented homemakers' age categories. Dummy variables were incorporated to represent risk information in the model. Results indicated . that.there‘was no statistical evidence that the 1959 cranberry scare had any significant impact on elasticity of demand for processed cranberries. There were temporary reductions in cranberry purchases observed in 1959, but consumption was restored to pre-product warning levels by 1960. This caSe study illustrates an example of both temporary risk exposure to pesticides in cranberries and temporary risk information. Shulstad and Stoevener's (1978) case study analyzed Oregon pheasant hunters' response to reports that pheasants contained excessive concentrations of mercury that could pose health risks to humans. From 1950 through .1971, the effects of information about the mercury contamination on both the average number of pheasant-hunting days per hunter and number of pheasant hunters in Oregon were analyzed. Four different measures of information were tested in the study, including column inches from newspaper articles about mercury contamina- tion in the preceding season, cumulative column inches from the past two seasons, number of articles in the past season, and cumulative number of articles in the previous two seasons. Cumulative number of articles was selected as the best measure since it produced the highest R-square. 16 Results from.an ordinary least squares regression estima- tion showed that although the information variable was not statistically significant to the average number of pheasant hunting days per hunter per season, there was a strong relationship between information and the number of hunters per season. In other words, the possibility of mercury contamina- _ tion had reduced the number of pheasant hunters during the sample period rather than merely reducing their consumption of the potentially dangerous product. Avoidance cost of mercury contamination to Oregon pheasant hunting was estimated to be $1.35 million based on the weighted loss of consumer surplus per season. This study is an example of persistent risk exposure accompanied by temporary risk information dissemination. Response to increasing amounts of information was also examined and measured by the cumulative effect of articles reported in the local newspaper, representing a shift in the demand function, and was found to be statistically signifi- cant. Swartz and Strand (1981) were also interested in determin- ing the effects of information about a product perceived as a health risk on consumer's purchasing behavior. Their study analyzed the changes in expected consumer and producer surpluses in the Baltimore oyster market following December 1985 reports of kepone contamination in Virginia's James River. Public concerns over kepone, a suspected carcinogen, had prompted health authorities to prohibit the harvest of l7 oysters in the James River. Oysters entering the Baltimore wholesale market nearly 200 miles to the north were supposedly uncontaminated. The lagged effect of media coverage was examined using a polynomial distributed lag model for the information variable. Information was measured by weighing articles in . four Washington and Baltimore newspapers based on their likelihood to effect oyster purchases and their likelihood of being read. Results showed a significant relationship between the reduction in oyster sales and a two month media lag, despite the fact that oysters sold in Baltimore were uncon- taminated. Consumer and producer losses due to imperfect information were then calculated and estimated to exceed $13,000 in 1969 dollars. Swartz and Strand also studied the impact of the timing of media coverage on the Baltimore oyster demand model. They estimated that if news of the contamination had been released at the onset of the oyster harvest season, economic losses could have been 25 percent greater than currently observed. Conversely, if news reports had been released at the end of the harvest season, predicted losses would have been 25 per- cent lower. This study could be classified as a model where there was not a danger of risk exposure, yet consumers reacted to risk information reported concerning a contamination that occurred in a different location. The duration of risk information lasted ten months, so it could be considered persistent. 18 Smith, van Ravenswaay, and Thompson (1984, 1988) also studied consumer response to a product warning in a case of the March of 1982 heptachlor contamination incident of Oahu milk. Heptachlor, a suspected carcinogen, had entered the milk supply when pineapple leaves sprayed with the pesticide were harvested and fed to dairy cattle. Despite government . efforts to assure consumers that milk remaining on store shelves after the incident was safe to drink, sales dropped substantially. A market demand model for Oahu fresh milk was estimated and findings indicated that both the positive and negative media coverage variables negatively effected consumption. Information about heptachlor was measured by weighing articles in Honolulu newspapers based on Budd's Attention Score (1964) . Articles were then categorized for content to determine if they were negative or positive. A model was estimated to determine the lagged effect of negative media coverage on sales and a three month lag was found to be statistically significant. Results indicated that a 1 percent increase in media coverage would reduce sales by 0.06 percent. The estimate of uncompensated lost sales to Oahu producers of Class I-A milk.was $422,000 or over $26,000 per producer. As stated earlier, this study could be categorized as a case where risk exposure was temporary, yet media coverage per- sisted for sixteen months after the incident. Reed Johnson (1987) conducted another study that quan- tified risk information and determined its impact on 19 consumption. The particular case he examined was the con- troversy over fumigant ethylene dibromide (EDB) in December of 1983 when the highly toxic pesticide was detected in grain products in Florida. Information was quantified using three different measures; cumulative column inches, change in total column inches, and an interaction variable between the change . in column inches and price. Six different models were com- pared and in each case, the information variable was found to be significantly related to the reduction in purchases of grain products. This case study is an example where risk exposure was persistent, yet the duration of intense national media coverage only lasted for ‘three months. Consumer response to increasing amounts of risk information was measured by the cumulative column inches variable. Estimated sales losses as a result of consumers' reactions to the EDB contamination were between $18 and $30 million. Johnson stressed that these estimates were not an actual reduction in consumer or producer losses since consumers could substitute other goods with higher prices for the contaminated product. The estimates do, however, give an indication of the relative magnitudes and economic impacts of a food scare. Johnson suggested that future welfare losses during a food scare could be minimized by disseminating appropriate risk information to the public earlier. When accurate risk in- formation is not available, however, or if misinformation does not allow consumers to avoid negative outcomes by changing their behavior, welfare losses are inevitable. Be concludes 20 that in this situation, the unrestricted flow of information may actually be to the public's detriment. 232.1 gummagg Although approaches may differ, the preceding studies all demonstrate that health warnings play a prominent role in . altering purchases of affected commodities. In each of the studies, the length of risk exposure and duration of news coverage, as well as article content, were important. As increasing amounts of risk information becomes available to the public, a cumulative media effect is observed in several of the studies. An additional finding is that negative and positive media coverage increase consumer awareness of a food contamination and, therefore, both need to be considered when conducting a consumer response study. In several of the studies, the cumulative effect of risk information was found to significantly reduce purchases. 2,; fingtheses Given the conceptual framework of consumer response and existing empirical findings, four hypotheses were formulated for the Alar study. First of all, the researchers were interested to determine if risk information had an impact on consumer's purchasing behavior; The second hypothesis was to determine whether all past information about the food scare would continue to effect consumer behavior. This would be tested by looking at the cumulative media effect of risk information. The third hypothesis was to test if a lagged 21 media effect was present. It was predicted that as risk information declined and consumers began to forget about the health risk from the growth regulator, a lagged media effect would become significant. The fourth hypothesis was to determine if once risk information was released, consumers would continue to believe the risk persisted regardless of how . much or how little media coverage occurred. This would be tested by a dummy variable beginning at the time when new information was released about the health risks of exposure to daminozide. The dummy variable would measure a one time shift in the demand curve. A comparison of positive and negative media coverage was not analyzed since past studies of consumer response indicated that all types of risk informa- tion increase consumer awareness. 2.4 Revenue Losses to New York Retailers To estimate revenue losses to retailers in New York, the change in the equilibrium price and quantity induced by Alar needs to be estimated. Normally, the equilibrium quantity and price per time period are jointly determined by demand and supply curves. However, in this case, only the New York City market is examined, and supply can be viewed as being perfect- ly elastic. (Retail prices were in fact stable during the period examined” See Figure 3.1.) Thus, New York City retail apple prices would be determined by the supply curve and quantities purchased by the demand curve. The change in 22 revenue for a given time period (e.g. one month) is the shaded area in Figure 2.1. The sum over time periods following the initial release of information is the total retail revenue loss. 23 FIGURE 2 . 1 Demand Shifted in Apple Consumption As a Result of Risk Information in New York ($/lb) in a Particular Month Retail Price of Apples A Per Capita Consumption of Apples Per Pound in New York in a Particular Month CHAPTER THREE DATA AND METHODS 3 ate Since apple supply to the New York City metropolitan area is assumed to be perfectly elastic, only the demand for fresh apples need be estimated. This requires data on apple pur- - chases, prices, income, risk information, and population for the New York City metropolitan area. Monthly data spanning from 1980 to mid-year 1989 was obtained so that the effects of the Alar incident could be observed over time. The sample period included 42 observations before the beginning of the controversy in July 1984, 108 observations before the height of the Alar scare in February of 1989, and five months that followed. Data described in the sections below can be found in Appendix One. 3.1,; Qonsumptigg In this study, arrivals of fresh apples in the New York City-Newark metropolitan area were used as a proxy for fresh apple consumption. The assumption is that apples received or warehoused for the New York City-Newark market were consumed within the metropolitan area. This assumption was supported by Doug Edwards of the Market News Service in Washington, D.C. who explained that shippers sell bulk quantities of apples to retail chain stores for local consumption. Supermarkets do 24 FIGURE 3 . 1 Monthly Per Capita Apple Consumption and Retail Price in New York City g 1.75 i 1.50 r; 1.25 £ ‘0'. I.” 5'. I: “:75 Y... a m g 0.25 g m 1 a: so 91 32 83 84 as as a? 98 99 Months Actual per Capita Apple Consumption in New York City CMSA --------- Deflated Retail Price of Apples in New York City 25 26 not in turn resell truckloads of fresh apples to grocery stores outside the metropolitan area. Monthly arrivals of fresh apples for 22 cities in the United States and Canada are published annually by the USDA Market News Service in "Fresh Fruit and Vegetable Arrival Totals". The document reports monthly apple arrivals in 1,000 . cwt units subdivided by state of origin and mode of transpor- tation. Of the 22 cities reported in the document, New York City accounted for approximately 14 percent of all fresh apples arrivals in 1989. Fresh apples unloaded in the various metropolitan areas are not desegregated by variety, hence, it was not possible to examine consumption changes over time of specific apple varieties in New York City. According to Mark Arney, Manager of the Michigan Apple Committee, apples have a relatively long shelf-life in grocery stores of three to four weeks, thus no adjustments were made to account for apple spoilage in the grocery stores. Apple consumption data for New York City is found in Appendix One. 32;.g gogulatiog In order to calculate per capita apple consumption in New York City, population estimates for the metropolitan area were required. "Local Area Personal Income", published by the Department of Commerce, Bureau of Economic Analysis (BEA) reports population estimates for metropolitan statistical areas (MSA) in the United States. ‘Monthly population figures of the New York City-Northern New Jersey-Long Island Con- solidated Metropolitan Statistical Area (CMSA) were calculated from annual data. Since population estimates‘for 1988 and 1989 had not yet been released by the BEA, projections were made based on population growth rates from 1980 to 1987 using the following formula: 1980 population estimate(1 + r)”"°"""=1987 population Solving for r, the growth rate was calculated to be 0.0002924. This figure was multiplied by the last monthly observation in 1987 and then added to subsequent monthly observations to determine 1988 and 1989 population estimates. Population data for the New York City CMSA may be found in Appendix One. 3 ice The U.S. Department of Commerce, Bureau of Labor Statis- tics (BLS) reports average monthly retail food prices on a U.S. and regional basis in their 221 nsgsilsd BERQII- Al- though the regional average price is an average of prices collected from metropolitan areas within the region, food prices for specific cities are not reported by or available from the agency. Moreover, BLS reports retail prices only for red delicious apples. Thus, it was necessary to seek other sources that would provide retail prices of fresh apples and apple substitutes by city from 1980 to 1989. 27 28 The New York City Department of Consumer Affairs provided average monthly retail price data for fresh fruits and other food commodities in New York City. Personnel in the department collect retail prices of 41 selected food items bi-weekly from approximately 150 food stores for the New York City Market—Basket Survey. Average retail prices of all apple - varieties collected from the second half of each month and reported in cents per pound were used for the apple market demand model. Average retail price of bananas, collected from the second half of each month, was chosen to represent a substitute for fresh apples. Both apple and banana prices were then deflated by the New York City-Newark Consumer Price Index for all urban consumers on a 1982-84 base period, reported by the U.S. Department of Labor, BLS (U.S. Department of Commerce, BIS, Q21 W M). In collecting the retail price data, the New York Department of Consumer Affairs randomly chooses stores from each of the five boroughs located in the metropolitan area which include both large supermarkets and small specialty shops. The number of stores selected from each borough reflects the approximate borough size. Different sizes and types of grocery stores were assigned weights to assure that the stores represented their approximate proportion of the retail food market. Although the stores were chosen randomly, efforts were made to select store locations that represented a cross-section of the metropolitan area's different income levels and ethnic neighborhoods. The survey is conducted during the work week with no regularity intended and thus storekeepers and owners can not anticipate when prices will be collected from their stores. Oranges have been used as substitutes in other apple market demand studies (Baumes and Conway, 1985), but unfor- tunately, retail orange prices are not collected by the New York Department of Consumer Affairs. Northeastern regional average prices of oranges, reported monthly by the Bureau of Labor Statistics, were used as a proxy in the New York City- Newark area and deflated by the Consumer Price Index for all urban consumers in the Northeastern region of the U.S. (U.S. Department of Commerce, BLS, 921 nstsilsg Bspgzg). Average retail prices of apples and oranges in the New York City metropolitan area are found in Appendix One. 11111__IR£2!£ Personal income for the New York City-Newark CMSA is reported annually in the Department of Commerce, Bureau of Economic Analysis "Local Area Personal Income" publication. However, personal income flows during a given year do not have constant distribution patterns across months and thus monthly estimates calculated from annual income figures are poor proxies. Consequently, quarterly U.S. disposable income estimates reported in the "Survey of Current Business" by the U.S. Department of Commerce, Bureau of Economic Analysis, were used. In other words, monthly per capita personal income was not used, but instead, the quarterly figure was substituted. 29 The quarterly income estimate used in each month was then deflated by the New York City Consumer Price Index for all urban consumers on a 82-84 base. 3.1 s e o t 0 Previous studies have demonstrated that consumer aware- ness of a contamination or food scare incident may signifi- cantly affect sales. Thus, a proxy for consumer information about the Alar scare was needed. Two.approaches were used to measure consumer information of the incident. The first approach involved incorporating a dummy variable for each of the key events in the Alar controversy: the July 1984 E.P.A. reassessment of Alar, the July 1986 grocery store boycott of apples sprayed with Alar, and the February 1989 "60 Minutes" television segment. These dummy variables measure one time shifts in the apple demand curve that may have occurred as a result of the key events. The second approach was to develop a proxy variable for the-amount of media coverage of Alar. Based on the assumption that newspaper coverage was representative of variations in total media coverage over time, articles about the Alar event were collected from a newspaper in the New York City area. The dissemination of risk information to the public was modeled by the cumulative number of articles about Alar appearing in this newspaper. The lagged effect of media coverage on fresh apple consumption was modeled by including the number of articles reported in prior months. 30 31 Selecting a newspaper to represent media coverage in New York City required several considerations. First of all, it was desirable to choose a newspaper that could be electroni- cally searched from a data base. Secondly, a newspaper with a large circulation size was sought. Based on these prere— quisites, the 11%! 19;); Tm, with an approximate circulation . of over a million, was chosen as a proxy for newspaper cov- erage of Alar. Other local newspapers such as Nswsday, ng YerkIribuns. 1442212329542. andNeYYerknailxNestere not covered or only partially covered on electronic data bases from 1984 through 1989. The number of articles per month in the Nsw 1915 Timss containing information about Alar was discovered to be strongly related to coverage of Alar from other national newspapers, including the B95593 M3 (85 percent correlation with the M 31.23 Tmss) and the §a_r; Fransisco QEIQDIQIQ (92 percent correlation with the NEE.YQI£ Timss). Thus, it seemed reasonable to assume that the Egg Ygxk Timss would be representative of newspaper coverage of Alar in New York City. A Nexis data base search of the full text of all leg 19115 Tings articles from July 1984 to July 1989 was performed by Charles Wolf at the Michigan State Law Library in Lansing. Using' the key"words "Alar“ and "daminozide" the search revealed 74 citations of articles related to the Alar inci- dent. A coding method similar to Budd's Attention Score (1964) and Radar Hayes (1989) was initially devised to weight the articles based on their prominence in the paper, but was later 32 eliminated since the weighted media variable proved to be insignificant in the models. The coding scheme involved weighing newspaper articles based on headline size, article length, location in the paper, and position on the page. 3.1.6 cher VariabTes Data on advertising expenditures for the New York City metropolitan area was unavailable, hence the possible effect of advertising by the apple industry is omitted from the model. Seasonality was accounted for using an autoregressive component (AR) in MicroTSP that identified the relationship between error terms in different time periods. First order serial correlation was observed and corrected using the AR(1) term. Once introduced, AR( 1) corrected for trend and seasonal patterns in the market demand models for fresh apples in New York City. §I£__!2£222§ Supply was assumed to be fixed in each period, thus single-equation demand models were estimated. Each of the demand equations included prices of apples and bananas, disposable income, a first-order autoregressive term, and various combinations of information variables. The informa- tion variables included were dummy variables in different time periods and quantitative media variables such as CUMNYT and NYT with lagged values. The generalized form of each demand equation is as follows: 33 LPCQ = f(LDRPA, LDRPS, LDINC, Information Variable(s), AR(1)) where: LPCQ = log of per capita fresh apple consumption in New York City, LDRPA log of the retail price of fresh apples per pound deflated by the New York City-Northeastern N.J. CPI-U (July l983=1.00), LDRPS = log of the retail price of fresh apple substitutes similarly deflated, LDINC - log of disposable per capita income deflated by the U.S. CPI-U (July l983=1.00), AR(1) = a first order autoregressive error specification. Information variables tested were: NYT = monthly number of articles about Alar in the New lggk Tings and lagged values, CUMNYT: cumulated number of articles about Alar in the New 12:3 Timss reported monthly, DV = a dummy variable, zero before the first news article about Alar in July 1984, and unity after, DV2 - a dummy variable, zero before the July 1986 grocery store boycott of Alar-treated apples, and unity after, DV3 = a dummy variable, zero before the February 1989 "60 Minutes" report, and unity after. Choosing an appropriate model to represent consumer response to the Alar scare was based on several criteria. First of all, the variables in the equations needed to comply with economic rationale; signs on coefficients were checked and elasticities tested to determine if they were consistent 34 with theory. Secondly, likelihood tests were performed to distinguish which models were statistically superior. Lastly, the absence of autocorrelation in the models was analyzed. To calculate New York City revenue loss to retailers resulting from Alar, predictions of revenue with the Alar scare were estimated and subtracted from estimates of what .sales would have been in the absence of the Alar incident. The estimated loss was then calculated on a percentage basis by dividing the estimated loss by the estimated loss added to the actual revenue of fresh apples in New York City from July 1984 to July 1989. CHAPTER FOUR ECONOMETRIC FINDINGS With an understanding of the theoretical approach and methods used, an econometric demand model of fresh apple consumption for New York City is analyzed before, during, and . for a short period following the Alar scare. The sections of this chapter explain the rationale and procedures used in selecting appropriate combinations of exogenous variables for the market demand model. Results from the model estimations are discussed in Section 4.5. 4.1 Selection 0 es 1 te Both bananas and oranges were individually tested as possible substitutes to be included in the fresh apple demand model. To select appropriate substitutes, deflated retail price of bananas per pound (DRPBNY) and deflated retail price of oranges per pound (DRPONY) were tested separately and together in each of the demand models. Although both sub- stitutes yielded insignificant coefficients, deflated banana price was chosen to represent the substitute since it had a positive coefficient and yielded a higher R2 than deflated orange price. Banana prices had been collected from the same agency and same locations as apple prices and are a more ac- curate data source. Orange prices were collected on a regional basis and regression results produced negative 35 36 coefficients. There were also gaps in the data set of orange prices where missing observations had to be interpolated. . atme ncom Per capita disposable personal income was originally included in the market demand model for fresh apples but was . later eliminated since its coefficient was insignificant and did not contribute to the model's ability to explain varia- tions in consumption. 1Another reason for excluding income is because apple purchases comprise such a small portion of consumers' budgets. 4. Se ectio A o iate Med a Covera e Va 'ables News coverage of Alar in New York City began in July of 1984 and continued sporadically for the next five years. To examine the impact of the amount and type of risk information on apple sales, seven different models were estimated. The media effect was captured in the models by the dummy variable (DV) for July of 1984, the cumulative number of articles reported monthly (CUMNYT), and the current and lagged number of articles written in the New York Times (NYT). A dummy variable for the July 1986 grocery store boycott (DV2) and a dummy variable for the February 1989 "60 Minutes" report (DV3) were tested but found to be insignificant regardless of 37 below‘. The dummy variable in 1984 was used to detect a step shift in the demand function following the E.P.A.'s reassessment of Alar. The cumulative number of articles variable, on the other hand, measured subsequent shifts in demand as consumers responded to the total amount of all risk information reported about Alar. Legged variables of news articles represented the effect of prior risk information on current purchasing behavior. 4 M t Six models with various combinations of media variables and one model excluding any measure of media effect were converted to log form and identified using MicroTSP statisti- cal computer package to examine the autocorrelation properties of each equation. All of the models exhibited first order serial correlation, indicating dependency among error terms in the current and previous time period, so an autoregressive error component.AR(1) was introduced to correct this problem. The estimation results of each of the five1models is displayed in Table 4.1. Several similarities among the seven models' statistical results were observed. For all models, the Durbin-Watson and ‘ LPCQNY = -0.787 - 2.01 LDRPANY + 0.32 LDRPANY - 0.30 DV3 (-1.95) (-5.70) (0.80) (-1.28) +0.65 AR (1). Numbers in parentheses are t-statistics. (8.47) 38 TABLE 4. 1 Alar Regression Results NOBEL: 1 2 3 4 5 6 7 CONSTANT -0.63 '0.61 -0.71 ~0.61 -0.68 -0.78 -0.88 (0.39) (0.39) (0.38)*** (0.40) (0.39)*** (0.39)*** (0.41)** LDIPAIY -2.05 -2.04 ~2.02 -2.07 ’2.07 -1.95 ~1.99 ' (0.34)* (0.33) (0.33)* (0.35)* (0.34)‘ (0.34)* (0.35)’ LDIPIIY 0.30 0.32 0.20 0.50 0.36 0.11 0.24 (0.40) (0.40) (0.39) (0.40) (0.38) (0.39) (0.39) 0Vi7-04 '0.26 '0.29 -0.24 -0.36 (0.14)*** (0.13)** (0.14)*** (0.13)‘ CUMNYT -0.003 '0.007 -0.01 (0.006) (0.005)*** (0.004)‘ IYT -0.001 -0.003 -0.006 (0.013) (0.011) (0.011) IYT(-1) '0.016 '0.019 -0.022 (0.013) (0.012)*** (0.012)‘** IYT(’2) '0.009 ~0.013 -0.017 (0.015) (0.012) (0.012) IYT('3) -0.017 -0.019 -0.022 (0.013) (0.012)*** (0.011)**’ AR(1) 0.587 0.587 0.504 0.640 0.611 0.616 0.703 (0.002). (0.082)’ (0.080)* (0.079)* (0.079)' (0.073)* (0.070)* sdj. :2 0.62 0.62 0.62 0.61 0.61 0.61 0.59 0.0. 1.72 1.70 1.71 1.72 1.73 0.71 1.76 LL Stet -6.84 -6.96 -9.05 ~9.S4 -10.53 -10.49 -13.70 Dependent variable e 1% "Significant at the e g 0.01 level. “Significant st the e s 0.05 level. ***Signtficsnt st the e g 0.10 level. Seaple period I May 1980 - July 1989 Figures In parentheses are stsndsrd errors. 39 Q-statistics indicated a low probability of serial correla- tion after the AR(1) term was incorporated and in each case, the independent variables accounted for approximately 60 percent of the variation of apple purchases in New York City. Deflated apple price and the autoregressive specification, AR(1) , were consistently significant in each regression. . Although the coefficient on deflated banana price was positive as expected, it was insignificant in all models after the autoregressive term was introduced. Since the equations are reported in log form, the coeffi- cients on price represent estimated elasticities in each mod- el. The range of retail price elasticities, -l.95 to -2.07, is similar among the models, showing an elastic demand for fresh apples at the market level. Own-price elasticities from previous studies of fresh apple demand in the U.S. are sum- marized in Table 4.2. Although results vary considerably among the studies, more recent estimates by Edman (1972) and Baumes and Conway (1985) tend to support this study's findings of an elastic demand relationship for fresh apples at the retail level. 4.5 Model Estimation Results All of the estimated models in Table 4.1 exhibited similar statistical properties of explanatory power and t- statistics, so in order to determine if risk information made 40 TABLE 4.2 Estimated Blasticities of Retail Fresh Apple Demand Source Period Own-Price Tomek(1968) 1946-67 -l.2 to -O.105 Waugh(l964) 1948-62 -1.239 - Brandow(l956) ? -O.6 George and King ? -O.72 (1971) Edman(1972) 1963-69 elastic in Interseasonal each season Baumes and Conway 1952-81 -2.288 (1985) a difference in the models, further testing was required. Likelihood ratio tests were performed on the models and results are displayed in Table 4.3. These tests involved comparing models using a Chi Square distribution to determine whether or not to reject the null hypothesis that the re- stricted was superior to the unrestricted model. Likelihood statistics used in the calculation of the ratios are found in Table 4.1. The result from the likelihood tests of Model 1 versus Model 7 (Table 4.3) indicates that risk information should be included in the models of fresh apple demand in New York City during the Alar scare. Model 7 which excludes any measure of 41 TABLE 4.3 Likelihood Test for Estimated Equations Unrestricted Restricted Likelihood Degrees :33;re Reject Model Model Ratio Freedom at 5% Ho? 1 2 0.24 1 3.841 NO 1 3 4.42 4 9.488 NO 1 7 13.72 6 12.590 YES 3 6 2.88 1 3.841 NO 2 6 7.06 4 9.488 N0 risk information was not found to be statistically superior to Model 1 which includes all measures of risk information. Testing whether different forms of information variables would significantly effect market demand for fresh apples in New York City was also examined. Neither Model 2, which included a dummy variable and lagged media effect, nor Model 3, which measured the cumulative media and dummy effect, were superior to Model 1. Testing Model 6 against Model 2 and Model 3 was to determine whether there is a one time shift in the demand curve rather than the amount of risk information making a difference. Both of these tests were found not to be statistically superior at the 95 percent level. It should be noted, however, that at the 90 percent confidence level, Model 6 was superior to Model 3, indicating that a one time shift in demand is a reasonable way to describe the Alar incident. 42 . m tes ew 0 et ue For five of the models, estimated sales loss to retailers was calculated from July of 1984 to July of 1989 to determine the impact of the Alar incident on the New York City apple market. The estimated loss in each case reflects the reduc- tion in revenue attributed to consumer awareness of Alar while . permitting changes to occur over time in apple prices, banana prices, population, and seasonality. Estimated total New York retail revenue loss was calcu- lated by subtracting the estimated equation with information variables from the same equation without information vari- ables. This was accomplished using the edit mode of MicroTSP computer package. For example, Model 5 which included the cumulative media variable was entered into the edit mode. A second equation was then replicated using the same variables and coefficients, but eliminating the cumulative media variable term (see example below). LPCQNY =b.678-2.065*LDRPANY+.357*LDRPBNY-1.027D-02*CUMNYT +AR(1) LPCQNY =-.678-2.065*LDRPANY+.357*LDRPBNY+AR(1) Since both functions were in log form, the antilog of each equation was taken. The two equations were then subtracted from one another and the results of each of the 61 monthly observations from July 1984 to July 1989 were multiplied by monthly population estimates and monthly deflated retail price of apples. These figures were then summed to arrive at the 43 retail loss attributed to the media variable. In the case of Model 5, the retail loss from the cumulative media variable was estimated to be $93 million. Percentage loss in revenue for’ each. model was then determined by the following procedure: Percentage loss in revenue = z - Actual revenue 8 price * quantity (in each period) = x Estimated loss in revenue = y z = 1 * 100 X+y Total actual apple revenue during the five year period were $468,822,363. An example of the revenue loss for Model Two from July 1984 to July 1989 follows: Estimated loss in revenue (y) = $187,251,133 Actual revenue (x) = $468,822,363 Percentage loss in revenue (2) = $125,167,215 $195,167,215 + $468,822,363 z = 29.4 percent loss in revenue to retailers. Results of total estimated revenue loss to retailers and percentage losses from each of the five models are summarized in Table 4.4. In addition to the five models listed above, the estimated sales loss of the equation that included only the dummy variable for February of 1989 (see footnote one) was calcu- lated to be $15,719,188. The percentage revenue loss to 44 TABLE 4.4 Total Estimated Sales Losses and Percentage Losses Total Estimated Percentage Sales Loss Sales Loss (millions of dollars) Model: 2 $195.2 29.4% 3 $194.8 29.3% 4 $ 38.4 7.6% 5 $ 93.2 16.6% 6 $187.3 28.5% retailers from risk information reported during this time period was 3 percent. All of the models that included the dummy variable for July 1984 (Models Two, Three and Six) had similar sales loss percentages. This indicates that the dummy variable captures the same effect in the different models, even when other information variables are included. Per Capita Apple Consumption (lbs) 45 FIGURE 4 . 1 Model 2 Estimated and Actual Per Capita Apple Consumption in New York City Using a Dummy Variable in 1989 and a Cumulative Media Variable 1.25 1.00 0.50 0.25 0.00 , 1904 1905 1906 1907 1900 1909 Actual per Capita Apple Consumption --------- Estimated per capita apple consumption Using a Dummy Variable and Cumulative Media Per Capita Apple Consumption (lbs) 46 FIGURE 4.2 Model 3 Estimated and Actual Per Vapita Apple Consumption in New York City Using a Dummy Variable in 1984 and Logged Media Variable 1.25 1.00 0.75 e. 50 .. .9 as eeeee 0.00 1904 1905 1906 190? 1900 1909 Actual per Capita Apple Consumption --------- Estimated per capita apple consumption Using a Dummy Variable and Logged Media Variable Per Capita Apple Consumption (lbs) 47 FIGURE 4.3 Model 4 Estimated and Actual Per Capita Apple Consumption in New York City Using Logged Media Variables 1.25 1% r... 0.75 '- .I 3 e". 0.50 .. 0.25 0.00 , . 1904 1905 1906 1907 1900 1909 Actual per Capita Apple Consumption --------- Estimated Per Capita Apple Consumption Using Logged Media Variables Per Capita Apple Consumption (lbs) 48 FIGURE 4 . 4 Model 5 Estimated and Actual Per Capita Apple Consumption in New York City Using a Cumulative Media Variable 1.25 1.00 1.75 ' 0.50 0.25 0.00 1904 1915 ‘ 1911 1917 ' 1911' 1919 Actual per Capita Apple Consumption --------- Estimated per capita apple consumption Using A cumulative Media Variable Per Capita Apple Consumption (lbs) 49 FIGURE 4.5 Model 6 Estimated and Actual Per Capita Apple Consumption in New York City Using a Dummy Variable Beginning in July 1984 0,58 '- g: ....... . :3- ~. o I. '.’ .0 0.25 0.01 . . 1114 1915 1911 1917 1911 1919 Actual per Capita Apple Consumption --------- Estimated per Capita Apple consumption Using a Dummy Variable in July 1984. CHLPTER.FIVE SUMMARY AND IMPLICATIONS This chapter summarizes the methods and major findings of this study. A discussion of policy implications and sug- gestions for future research follow in Sections 5.4 and 5.5. .§;L__!2£LQQ§ The consumer response models developed in this study considered the effects of risk information on consumption of fresh apples during the Alar scare. The effect of increasing amounts of information over time as well as the lagged effect of media coverage was examined. By allowing apple prices and prices of substitutes to fluctuate over time, the model cap- tures the dynamics of the apple market during the observation period. In contrast to previous studies [e.g. Schulstad and Stoevener (1978), Swartz and Strand (1982), and Johnson (1987)], all of the demand equations presented in Chapter Four were individually tested to identify trends in consumption over time and to correct for dependency among error terms in different time periods. These adjustments made the models less restrictive and thus improved the accuracy of sales loss estimates. During the observation period, the Alar incident was characterized by a downward shift in the market demand curve of apples beginning in July of 1984 and subsequent shifts in 50 51 the months that followed due to the increasing amount of risk information to consumers in New York City. Regression results proved that risk information was significant in changing con- sumers' purchasing behavior as demonstrated by the likelihood test comparing Model 7 and Model 1. Despite the apparent strengths, the methods used in this . study also have weaknesses. Various combinations of informa- tion variables summarized in Table 4.1 all produced virtually the same explanatory power in each of the models. This indicates that the variation in apple consumption was no more accurately explained by one equation over another. Newspaper coverage from only one major newspaper was used as a proxy for all media attention of the Alar incident. Radio and tele- vision coverage were not included as risk information sources in the study. Other local newspapers in New York City may have presented more or less coverage to the Alar scare, although a separate test conducted in this study found that newspaper coverage in other national newspapers was highly correlated with coverage in the New 191;); Times. The methods developed in this study will be of interest to government officials as well as participants in the food industry. Determining methods to measure media coverage and its subsequent affect on consumption during a food scare enables government and industry to develop strategies for dealing with any future food scares that may occur. Quantify- ing sales losses helps producers, marketers, and retailers to realize the costs incurred during a food scare when they 52 present estimates of economic losses to government officials who provide financial assistance. 5.2 E timates o v n o ses to Reta ers Since the July 1984 EPA reassessment announcement, par- ticipants in the apple industry have been accruing monthly .sales losses due to consumer awareness of Alar. Estimated revenue loss to retailers in New York City due to media cover- age in this study only indicate losses that occurred during the period of observation. Thus, determining the entire estimates of revenue loss from risk information will require extending the data set to the end of the Alar incident. Estimates from the models that included the dummy vari- able for July 1984 consistently reported that apple sales were 30 percent lower than they would.have been over the five year observation period had Alar not occurred. These percentages, however, only represent a portion of the sales losses incurred by participants in the apple industry. .Additional costs such as grocery store testing of Alar in apples and extra holding costs of apples in cold storage and controlled atmosphere have yet to be determined. .3 o m ations As stated in Chapter One, many of the older commercial pesticides used on agricultural produce have not been fully tested for chronic toxicity. Undoubtedly, conflicts will arise in the future as consumers demand a safer food supply, 53 free of potentially harmful pesticide residues. lessons learned from the Alar scare will help government health officials deal with similar product warning cases that may arise in the years to come. In July of 1984, the E.P.A. first recommended cancelling Alar's registration but later the decision was reversed due . to insufficient data on the carcinogenicity of Alar. Incon- sistencies in the agency's policy on Alar may have lead to a loss in credibility. Consumer interest groups conducted their own tests that implicated Alar as a health threat to the public. Confused by conflicting reports of the dangers of exposure to the growth regulator, consumers responded by reducing their consumption and consequently, participants in the apple industry experienced sizeable sales losses over the next five years. Had government recalled existing stocks of Alar and instigated a temporary ban on Alar pending further tests results at the onset of the controversy in July 1984, revenue losses to industry may have been minimized. A temporary ban would have eliminated media attention of Alar and concerns among state officials and consumer interest groups of the health risk of exposure to daminozide. .4 he esear h Many lessons remain to be learned from the Alar scare. The true magnitude of the incident will not be realized until apple consumption returns to pre-product warning levels. This 54 will require extending the period of observation past July 1989 so that marketing data from the 1989 autumn apple crop and beyond may be included in the analysis. The extension of the time series data may enable researchers to determine if Uniroyal's June 1989 voluntary ban on Alar and recall of existing stocks was effective in regaining consumer confidence .in fresh apples, thus restoring sales to pre-Alar scare levels. A lagged effect of media coverage may be detected in the future as consumers begin to forget the incident and resume their normal purchasing behavior. The impact of Alar on other subsectors of the apple in- dustry remains unknown. Growers, manufacturers, and retailers in the processed apple market undoubtedly experienced losses attributed to the Alar incident. The additional fear among consumers of exposure to UDMH, a suspected carcinogenic by- product of Alar found in processed apple products, may have resulted in even larger sales losses to this segment of the apple industry. Due to data constraints, the impact of Alar on the processed apple market in New York City was not studied, but provided that relevant data on apple juice, apple sauce, apple pie, and other processed apple products was available, an additional market demand model could be con- structed and compared to the fresh apple market. Another subsector that was not studied was the organic apple market. Amid concerns of exposure to Alar, many con- sumers may have converted to purchasing organic apples as a substitute for apples grown conventionally. .Although organic 55 produce is usually more expensive, some consumers may have been willing to pay the additional cost to avoid the risk of exposure to the growth regulator. Consumers may have also converted to purchasing other substitutes other than the ones estimated in this study. Within the conventional apple market, substitution - effects may also have occurred during the Alar incident. A number of the newspaper articles in the New XQIB 11mg: and news magazines reported that in general, red varieties of apples were treated with Alar, while most green apples such as Granny Smith, were not” Providing that.data was available, it would be interesting to observe if consumers substituted green for red varieties of apples during the observation period. The effects of Alar on apple markets in other metro- politan areas is also of interest. A pooled analysis, similar to the study conducted by Johnson (1987) would allow research- ers to observe consumer response to a product warning in dif- ferent regions of the United States. Although newspaper coverage of Alar from. major' newspapers in Boston, San Francisco, and Los Angeles was highly correlated with coverage in the New 39;); lines, the effects of risk information on con- sumption was only studied in New York City. The impact of Alar on the international apple market would also be of interest. Another topic worthy of investigation is the consumer "saturation point" of awareness. An information saturation 56 point refers to a level reached by consumers in which they have been so inundated with risk information that they no longer try to avoid products that are reported as potentially health threatening. At this point, consumers may believe that "everything gives you cancer" and will consequently ignore risk information, refusing to adjust their purchasing be- havior. In February of 1989, the Alar incident, compounded by news reports of cyanide contamination in Chilean grapes, may have caused many consumers to reach such a level. Model Eight (see footnote one), measured the effects of the February 1989 media blitz on Alar and the information variable was found to be insignificant. In other words, the DV3 variable may be interpreted as an indication that consumers were no longer responding to risk information about Alar. After the data set is expanded, this phenomena may be explored with greater certainty. A final topic that merits further research is a study of consumer perceptions of reliable sources of risk information. Consumers receive risk information from a variety of different types of media sources including newspapers, news magazines, television, and radio. News coverage among different forms of media tend to be highly correlated. For example, during both months when the ”60 Minutes" reports were aired on national television, there was a large increase in the number of news articles reported in the Hex X913 Iimee. Consumers perceptions of reliable sources of risk information, however, is unknown and would be worthy of investigating since it 57 might allow policy makers to understand the medium that has the greatest influence on consumer perceptions of risk information. APPENDIX ONE DATA Months QNY' POPNYb pcouv‘ NYTd 1980 Jan. 189.0000 17434.60 1.084051 0.000000 Feb. 196.0000 17433.20 1.124292 0.000000 Mar. 198.0000 17431.80 1.135855 0.000000 Apr. 150.0000 17430.40 0.860565 0.000000 May 152.0000 17429.00 0.872110 0.000000 June 132.0000 17427.60 0.757419 0.000000 July 83.0000 17429.70 0.476199 0.000000 Aug. 47.0000 17431.70 0.269624 0.000000 Sept 103.0000 17433.80 0.590806 0.000000 Oct. 201.0000 17435.90 1.152794 0.000000 Nov. 181.0000 17437.90 1.037969 0.000000 Dec. 220.0000 17440.00 1.261468 0.000000 1981 Jan. 248.0000 17442.10 1.421847 0.000000 Feb. 246.0000 17444.10 1.410219 0.000000 Mar. 237.0000 17446.20 1.358462 0.000000 Apr. 175.0000 17448.30 1.002963 0.000000 For description of variables, see footnotes on page 70. 58 59 Months QNY POPNY PCQNY NYT May 172.0000 17450.30 0.985656 0.000000 June 119.0000 17452.40 0.681855 0.000000 July 92.0000 17455.10 0.527067 0.000000 Aug. 102.0000 17457.90 0.584263 0.000000 Sept 113.0000 17460.60 0.647171 0.000000 Oct. 177.0000 17463.30 1.013554 0.000000 Nov. 258.0000 17466.10 1.477147 0.000000 Dec. 212.0000 17468.80 1.213592 0.000000 1982 Jan. 161.0000 17471.50 0.921501 0.000000 Feb. 199.0000 17474.30 1.138815 0.000000 Mar. 221.0000 17476.90 1.264526 0.000000 Apr. 191.0000 17479.70 1.092696 0.000000 May 219.0000 17482.50 1.252681 0.000000 June 136.0000 17485.20 0.777801 0.000000 July 77.0000 17495.70 0.440108 0.000000 Aug. 78.0000 17506.20 0.445556 0.000000 Sept 79.0000 17516.70 0.450998 0.000000 Oct. 137.0000 17527.20 0.781642 0.000000 Nov. 174.0000 17537.70 0.992148 0.000000 Dec. 185.0000 17548.20 1.054239 0.000000 1983 Jan. 144.0000 17558.70 0.820106 0.000000 Feb. 187.0000 17569.20 1.064363 0.000000 60 Months QNY POPNY PCQNY NYT Mar. 213.0000 17579.70 1.211625 0.000000 Apr. 196.0000 17590.20 1.114257 0.000000 May 180.0000 17600.70 1.022687 0.000000 June 135.0000 17611.20 0.766558 0.000000 July 94.0000 17619.70 .0.533494 0.000000 Aug. 73.0000 17628.20 0.414109 0.000000 Sept 103.0000 17636.70 0.584010 0.000000 Oct. 157.0000 17645.20 0.889760 0.000000 NOV. 237.0000 17653.70 1.342495 0.000000 Dec. 160.0000 17662.20 0.905890 0.000000 1984 Jan. 186.0000 17670.70 1.052590 0.000000 Feb. 182.0000 17679.20 1.029458 0.000000 Mar. 193.0000 17687.70 1.091154 0.000000 Apr. 154.0000 17696.20 0.870243 0.000000 May 155.0000 17704.70 0.875474 0.000000 June 85.0000 17713.20 0.479868 0.000000 July 66.0000 17719.30 0.372475 1.000000 Aug. 69.0000 17725.50 0.389270 0.000000 Sept 72.0000 17731.60 0.406055 0.000000 Oct. 110.0000 17737.80 0.620145 0.000000 NOV. 148.0000 17743.90 0.834090 0.000000 Dec. 118.0000 17750.00 0.664789 0.000000 61 Months QNY POPNY PCQNY NYT 1985 Jan. 129.0000 17756.20 0.726507 0.000000 Feb. 127.0000 17762.30 0.714997 0.000000 Mar. 143.0000 17768.50 0.804795 0.000000 Apr. 160.0000 17774.60 0.900161 0.000000 May 150.0000 17780.80 0.843606 0.000000 June 83.0000 17786.90 0.466636 0.000000 July 71.0000 17795.30 0.398982 0.000000 Aug. 59.0000 17803.70 0.331392 2.000000 Sept 70.0000 17812.10 0.392991 1.000000 Oct. 118.0000 17820.50 0.662159 0.000000 Nov. 137.0000 17828.90 0.768415 0.000000 Dec. 149.0000 17837.30 0.835328 0.000000 1986 Jan. 132.0000 17845.70 0.739674 3.000000 Feb. 134.0000 17854.10 0.750528 0.000000 Mar. 134.0000 17862.50 0.750175 1.000000 Apr. 135.0000 17870.80 0.755422 0.000000 May 107.0000 17879.20 0.598461 2.000000 June 65.0000 17887.60 0.363380 1.000000 July 41.0000 17890.40 0.229173 2.000000 Aug. 21.0000 17893.10 0.117364 0.000000 Sept 95.0000 17895.90 0.530848 0.000000 Oct. 187.0000 17898.60 1.044775 0.000000 Nov. 147.0000 17901.30 0.821169 1.000000 62 Months QNY POPNY PCQNY NYT Dec. 201.0000 17904.10 1.122648 0.000000 1987 Jan. 107.0000 17906.80 0.597538 3.000000 Feb. 85.0000 17909.60 0.474606 0.000000 Mar. 149.0000 17912.30 0.831831 0.000000 Apr. 112.0000 17915.10 0.625171 0.000000 May 96.0000 17917.80 0.535780 2.000000 June 175.0000 17920.50 0.976535 1.000000 July 150.0000 17923.20 0.836904 0.000000 Aug. 78.0000 17928.40 0.435064 0.000000 Sept 84.0000 17933.90 0.468387 1.000000 Oct. 157.0000 17939.10 0.875183 1.000000 Nov. 132.0000 17944.30 0.735610 0.000000 Dec. 120.0000 17949.50 0.668542 0.000000 1988 Jan. 167.0000 17954.70 0.930119 0.000000 Feb. 164.0000 17959.90 0.913145 1.000000 Mar. 160.0000 17965.20 0.890611 1.000000 Apr. 154.0000 17970.50 0.856960 0.000000 May 119.0000 17975.80 0.662001 0.000000 June 89.0000 17981.10 0.494964 0.000000 July 81.0000 17986.40 0.450340 0.000000 Aug. 63.0000 17991.70 0.350161 0.000000 Sept. 62.0000 17996.90 0.344504 1.000000 63 Months QNY POPNY PCQNY NYT Oct. 86.0000 18002.20 0.477719 0.000000 Nov. 98.0000 18007.50 0.544218 0.000000 Dec. 97.0000 18012.70 0.538509 0.000000 1989 Jan. 91.0000 18018.00 0.505050 1.000000 Feb. 98.0000 18023.30 0.543741 7.000000 Mar. 103.0000 18028.60 0.571315 25.000000 Apr. 81.0000 18033.90 0.449154 5.000000 May 88.0000 18039.20 0.487826 11.000000 June 70.0000 18044.50 0.387930 0.000000 July 94.0000 18049.80 0.520781 0.000000 64 Months DRPANY’ DRFBNYf CPINY" 1980 Jan. 0.728900 0.409207 0.782000 Feb. 0.760456 0.456274 0.789000 Mar. 0.787500 0.475000 0.800000 Apr. 0.818859 0.459057 0.806000 May 0.838471 0.468557 0.811000 June 0.852619 0.426309 0.821000 July 0.883777 0.411622 0.826000 Aug. 0.996399 0.432173 0.833000 Sept 0.933014 0.430622 0.836000 Oct. 0.713436 0.404281 0.841000 Nov. 0.673759 0.413712 0.846000 Dec. 0.678363 0.421053 0.855000 1981 Jan. 0.637312 0.428737 0.863000 Feb. 0.640732 0.423341 0.874000 Mar. 0.660592 0.444191 0.878000 Apr. 0.656852 0.430351 0.883000 May 0.641892 0.450451 0.888000 June 0.681564 0.413408 0.895000 July 0.715859 0.396476 0.908000 Aug. 0.796943 0.393013 0.916000 Sept 0.763441 0.419355 0.930000 Oct. 0.679612 0.399137 0.927000 Nov. 0.680346 0.399568 0.926000 For description of variables, see footnotes on page 70. 65 Months DRPANY DRPBNY CPINY - Dec. 0.679612 0.399137 0.927000 1982 Jan. 0.678149 0.409042 0.929000 Feb. 0.676692 0.397422 0.931000 Mar. 0.745946 0.421622, 0.925000 Apr. 0.700431 0.431035 0.928000 May 0.725720 0.405550 0.937000 June 0.721003 0.397074 0.957000 July 0.761210 0.354536 0.959000 Aug. 0.737279 0.353063 0.963000 Sept 0.710608 0.370752 0.971000 Oct. 0.660569 0.335366 0.984000 Nov. 0.652395 0.346585 0.981000 Dec. 0.656410 0.338462 0.975000 1983 Jan. 0.644172 0.378323 0.978000 Feb. 0.622449 0.377551 0.980000 Mar. 0.621814 0.377166 0.981000 Apr. 0.615540 0.454087 0.991000 May 0.623743 0.482897 0.994000 June 0.661986 0.471414 0.997000 July 0.700000 0.450000 1.000000 Aug. 0.729271 0.439560 1.001000 Sept. 0.772277 0.425743 1.010000 66 Months DRPANY DRPBNY CPINY Oct. 0.651530 0.394867 1.013000 Nov. 0.648968 0.363815 1.017000 Dec. 0.648330 0.353635 1.018000 1984 Jan. 0.642023 0.359922 1.028000 Feb. 0.628627 0.377176 1.034000 Mar. 0.626808 0.366442 1.037000 Apr. 0.691643 0.384246 1.041000 May 0.691643 0.374640 1.041000 June 0.699904 0.383509 1.043000 July 0.696565 0.381679 1.048000 Aug. 0.710900 0.369668 1.055000 Sept 0.725047 0.329567 1.062000 Oct. 0.725730 0.367578 1.061000 Nov. 0.629108 0.309859 1.065000 Dec. 0.657277 0.328638 1.065000 1985 Jan. 0.609185 0.365511 1.067000 Feb. 0.643057 0.382106 1.073000 Mar. 0.679070 0.390698 1.075000 Apr. 0.658017 0.417053 1.079000 May 0.638298 0.388529 1.081000 June 0.683287 0.378578 1.083000 July 0.710332 0.359779 1.084000 67 Months DRPANY DRPBNY CPINY Aug. 0.677656 0.357143 1.092000 Sept. 0.729927 0.374088 1.096000 Oct. 0.673953 0.346084 1.098000 Nov. 0.623306 0.316170 1.107000 Dec. 0.639640 0.396396 1.110000 1986 Jan. 0.724508 0.357782 1.118000 Feb. 0.735426 0.367713 1.115000 Mar. 0.735426 0.376682 1.115000 Apr. 0.746403 0.458633 1.112000 May 0.775473 0.477908 1.109000 June 0.823635 0.384960 1.117000 July 0.826667 0.355556 1.125000 Aug. 1.029281 0.346052 1.127000 Sept 1.000000 0.362832 1.130000 Oct. 0.749559 0.361552 1.134000 Nov. 0.750221 0.370697 1.133000 Dec. 0.711775 0.333919 1.138000 1987 Jan. 0.706190 0.348736 1.147000 Feb. 0.711188 0.364267 1.153000 Mar. 0.708117 0.371330 1.158000 Apr. 0.720412 0.351629 1.166000 May 0.852515 0.383632 1.173000 68 Months DRPANY DRPBNY CPINY ‘ June 0.780985 0.373514 1.178000 July 0.814249 0.347752 1.179000 Aug. 0.782170 0.336417 1.189000 Sept 0.742905 0.358932 1.198000 Oct. 0.623960 0.332779 1.202000 Nov. 0.564315 0.340249 1.205000 Dec. 0.563847 0.339967 1.206000 1988 Jan. 0.623330 0.356189 1.123000 Feb. 0.660125 0.401427 1.121000 Mar. 0.633745 0.370370 1.215000 Apr. 0.611745 0.358891 1.226000 May 0.619397 0.374898 1.227000 June 0.641755 0.446791 1.231000 July 0.736246 0.380259 1.236000 Aug. 0.909823 0.338164 1.242000 Sept 0.896825 0.341270 1.260000 Oct. 0.649762 0.340729 1.262000 Nov. 0.611597 0.357426 1.259000 Dec. 0.619048 0.365079 1.260000 1989 Jan. 0.629921 0.346457 1.270000 Feb. 0.650470 0.352665 1.276000 Mar. 0.636152 0.403414 1.289000 69 Months DRPANY DRPBNY CPINY ’ Apr. 0.594595 0.440154 1.295000 May 0.591398 0.460829 1.302000 June 0.605364 0.413793 1.305000 July 0.604900 0.367534 1.306000 4. 70 FOOTNOTES Total fresh apple arrivals in New York City in 16,000 cwt units. Source: Printouts from the USDA Market News Service's "Fresh Fruit and Vegetable Arrivals in Cities." Doug Edwards and Lynne Erickson (202) 447- 3343. CMSA population estimates fro New York City in 1,000 people. Source: U.S. Department of Commerce, Bureau of Economic Analysis. Survey 9: 011;;th fleeiness. Various Issues. Per capita fresh apple consumption in New York City CMSA in pounds. QNY/POPNY Number of articles written about Alar appearing in the New 1916.11.89.51 Deflated retail price of fresh apples in New York City (in pounds). Deflated retail price of bananas in New York city (in pounds). Source: Cathie Konopa at the New York City Department of Consumer Affairs (212) 566-5046. General Food Basket Report. Consumer Price Index for New York City. Source: U.S. Department of Commerce, Bureau of Labor Statistics, e21 pe;eile§_3epe;;. Various issues.. 71 APPENDIX TWO CERONOLOGICAL REFERENCE OF NEWSPAPER ARTICLES ABOUT ALAR IN THE NEW YORK TIMES "Pesticide Inquiry Seeks Possible Cancer Link." UPI, Washing- ton, New 1e;k.11mee, 22 July 1984, p.A30. "U.S. Agency Urges Banning Chemical Used with Apples." UPI, Washington, Neg Xezk Linea, 29 August 1985, sec. II p.2. Jenkins, Nancy. "Fruit-Chemical Ban Weighed." New Nork limes, 30 August 1985, sec. 11 p.4. Saft, Marcia. "In Apple Orchards, Hard-Won Victory." New Xegk Iimee, 22 September 1985, sec. 23, p.l and 8. Schneider, Keith. "Tiny Traces of Suspected Chemical Found in Apple Juice and Sauce." Neg Xezk Iimee, 14 January 1986, p.A15. Shabecoff, Philip. "E.P.A. Won't Ban Use of Chemical on Apples." Neg Xerk,11me§, 23 January 1986, p.A18. News Summary: New Kerk Iimee, 23 January 1986, p.81. Shabecoff, Philip. "Pesticides Finally Top the Problem List of E.P.A." Neg Kerk Iimee, 6 March 1986, p.B12. Schneider, Keith. "Additives in Food a Concern." Neg leek Iimee, 24 May 1986, p.I48. Williams, Winston. "Polishing the Apple's Image." Neg Xezk Iieee, 25 May 1986, sec. 3, p.4. Battista, Carolyn. "Chemical a Concern to Apple Growers." Neg Kerk Timee, 15 June 1986, sec. llCN, p.12. Schneider, Keith. "Immediate Ban Sought on Use of a Farm Chemical." Neg Xezk Times, 3 July 1986, p.A22. Molotsky, Irvin. "Consumer’ Saturday; A Ban on. Treated Apples." Neg lexk Iigee, 26 July 1986, sec. 1, p.10. Fabor, Harold. "State Apple Crop Down 15% and Prices Rise." Neg Xerk Timee, 2 November 1986, sec. 1, p.41. 72 Fabricant, Florence. "Food Notes." Neg Xeek Iimee, 7 January 1987, sec. C, p.7. "U.S. Says it Won't Ban Use of Apple Pesticide." Reuters, Washington, Neg leek Iimee, 7 January 1987, p.A23. Burros, Marian. "The Fresh Apple Appeal of Foods Grown Organically." Neg Xeek Iimee, 28 January 1987, p.C1. Narus, Bob. "The Environment." Neg leek Iimee, 10 May 1987, sec. 11NJ, p.26. Myerhoff, Albert. "Want the Pesticide Industry in Your Milk?" Neg leek Iimee, 26 May 1987, sec. p.A23. Burros, Marian. "U.S. Food Regulation: Tales From a Twigh- light Zone." Neg leek Iimee, 10 June 1987, p.c1. Battista, Carolyn. "Apple Growers See Smaller Harvest." Neg leek Iimee, sec. llCN, p.10. "Fallen Apples Reduce Hudson Valley's Yield." Neg Xerk,1ieee, 5 October 1987, p.D6. Schneider, Keith. "Supermarket Chain Accused of Breaking Vow on Pesticide." New Yerk Iimee, 3 February 1988, p.86. "Washington Talk: Briefing: Forbidden Fruit." Neg:1eek Times, 30 March 1988, p.A22. Shabecoff, Philip. ”House Backs Bill Speeding Removal of Some Pesticides." Neg leek Iimee, 21 September 1988, p.Al. Volk, Patricia. "Bless You Marty." Neg Xeek lines, 1 January 1989, sec. 6, p.20. Shabecoff, Philip. "Hazard Reported in Apple Chemical." Neg leek Iimee, 2 February 1989, p.Al. Shabecoff, Philip. "100 Chemicals for Apples Add Up to Enigma on Safety.“ Neg xeek Iieee, 5 February 1989, sec. 1, p022. "Even the Apple Has Fallen." Neg leek 11mg, 6 February 1989, p.A14. Shabecoff, Philip. "E.P.A. Restricts Fungicide's Use on 42 Products." Neg leek Iimee, 17 February 1989, p.A14. "Fruit of That Forbidden Tree an Apricot?" To the Editor, New leek Iimee, 21 February 1989, p.A22. Burros, Marian. "Eating Well." Neg Xeek Iimee, 22 February 1989, p.C8. 73 "Pesticides Termed High Cancer Risk for Children." AP, New leek Iimee, 25 February 1989, sec. 1, p.30. Rierden, Andi. "Consumers Prefer Apples and Squash Pure, Not Pretty." Neg leek Iimee, sec. 12CN, p.1. Leary, Warren. "Fear of Aflatoxin: The Debate About the Car- cinogens That Man Didn't Make." Neg leek 11m, 5 March 1989, sec. 4, p.24. Leary, Warren. ”U.S. Urges Consumers Not to Eat Fruit From Chile." Neg leek Iimee, 14 March 1989, p.A15. "How to Reduce Some Nutritional Risks." Neg leek Iimee, 15 March 1989, p.C4. Burros, Marian. "Eating Well; Is That Food Really Safe." New leek Times, 15 March 1989, p.C1. "Health Official Rebukes Schools Over Apple Bans." AP, Los Angeles, Neg leek Iimee, 16 March 1989, p.Blo. Shabecoff, Philip. "3 U.S. Agencies, to Allay Public Fears, Declare Apples Safe." Neg leek Iimee, 17 March 1989, p.A16. News Summary, Lead: International. Neg leek Iimee, 17 March 1989, p.A2. "U.S.-Chile Diplomacy on Grape Scare." Neg leek Iieee, 17 March 1989, p.Al. "Apples Pass Tests for Return to Los Angeles School Menu." AP, Los Angeles, Neg leek nmee, 18 March 1989, sec. 1, p.50. Libov, Charlotte. "We're Concerned About Our Children." New leek Iimee, 19 March 1989, sec. lZCN, p.3. Schneider, Keith. "Pesticide Regulation, Slow and Unsteady." Neg leek Iimee, 19 March 1989, sec. 4, p.7. Dionne, J. Jr. "The Limits of Risk: From Apples to Ter- rorism." Neg leek limee, 19 March 1989, sec. 1, p.3. "Board Returns Apples to New York Schools." Neg leek Iimee, 21 March 1989, p.83. "Organic Produce Preferred." AP, Washington, Neg leek Iimee, 21 March 1989, p.A16. Brody, Jane. "Health: Personal Health." Neg leek Ilmee, 23 March 1989, p.B12. 74 Egans, Timothy. "Farming Without Alar, Suffering With the Rest." Neg leek Iieee, 24 March 1989, p.All. Hamilton, Robert. "They Have to See the Whole Picture." New leek Iimee, 26 March 1989, sec. 12CN, p.3. Stevens, William. "Officials Call Microbes Most Urgent Food Threat." Neg leek Iimee, 28 March 1989, p.C1. Goldman-Posluns, Marcy. "Foods That Play April Fool's Joke on Palate and Eye.” Neg leek Linea, 29 March 1989, p.C3. Burros, Marian. "A Growing Harvest of Organic Produce." New leek Iimee, 29 March 1989, p.C1. Oakes, John. "A Silent Spring for Kids." Neg leek Ilmee, 30 March 1989, p.A25. Leary, Warren. "Tests Data Differ on Ripening Agent." New leek Iimee, 30 March 1989, p.Al7. "Apple-Juice Makers in Layoffs." AP, Milton, N.Y. Neg leek Iimee, 5 April 1989, p.82. McGill, Douglas. "Making Mashed Peas Pay Off." Neg leek Times, 9 April 1989, sec. 3, p.4. "Of Course We Worry About What's in Our Food." To the Editor. Neg leek Iimee, 11 April 1989, p.A30. Shabecoff, Philip. ”Moderation Becomes Mainstream: The Nation is Getting Ready to Cut Its Dose of Pesticides." Neg leek Iimee, 16 April 1989, sec. 4, p.6. "U.S. to Review Food-Safety Efforts." Reuters, WaShington, Neg leek Iimee, 26 April 1989, p.C12. Schneider, Keith. "Fears of Pesticides.ThreatenMAmerican Way of Farming." Neg leek Iieee, 1 May 1989, p.Al. “Fruit Growers Pull Commercials to Protest Report by CBS on Alar." AP, Yakima, Washington, Neg leek Iimee, 7 May 1989, sec. 1, p.1. Passell, Peter» "The American Sense of Peril: A.Stifling Cost of Modern Life.” Neg leek Iimee, 8 May 1989, p.Al. "Alar Ban is Coming." AP, Washington, Neg leek Iimee, 12 May 1989, p.88. Henry, Sondra. ""I Know Where I Am: But Why?" Neg leek Limee, 14 May 1989, sec. 12LI, p.32. Daugherty, Karen. "The Muddle in Eating Right." New leek Iimee, 14 May 1989, sec. 12CN, p.32. 75 "Steps to Ban Alar Announced." AP, Washington, Neg leek limes, 14 May 1989, sec. 1, p.1. News Summary, Lead: International. m leek 11m, 16 May 1989, p.A2. Shabecoff, Philip. "Apple Industry Says It Will End Use of Chemical." Neg leek Iimee, 16 May 1989, p.Al. "A Safer Spin on Sweeteners." Editorial, fig leek Meg, 19 May 1989, p.A34. Bellaforte, Stephen. "Danger: Keep Off the Grass." Neg Mk flmee, 21 May 1989, sec. l2LI, p.22. Fabor, Harold. "Celebration in 3-State Region." New leek Times, 26 May 1989, p.C28. Business Digest: Neg leek Linea, 3 June 1989, sec. 1, p.35. Shabecoff, Philip. "Apple Chemical Being Removed In U.S. Market." Neg leek flmee, 3 June 1989, sec. 1, p.1. REFERENCES Ames, Bruce N. and Gold, Lois S. "Pesticides, Risk, and Applesauce." Letter to the Editor. fielenee, 19 May IE» p. 755. Brown, Joseph D. "Effect of a Health Hazard 'Scare' on Consumer Demand." Meagan 1192:1151 9.: 89121911119111 Beenemlee 51 (August 1969): 676-78. Budd, Joseph H. WAttention Score: A Device for'Measuring News 'PlaY-'" 19111211311501 Quarterly (Spring 1964): 259-62- Buxton, Boyd M. "Economic Impact of Consumer Health Concerns About Alar on Apples." limit and Tree Nuts. fiitllatien end euelook,leaeeook. TFS 250 Economic Research Service, USDA, (August 1989): pp 85-88. Clarke, Darral G. "Econometric Measurement of the Duration of Adverstising Effect on Sales." legenel ee Neekeeing Beeeeeen 13 (November 1976): 345-57. Cony, Ann. "Stricter Pesticide Controls Urged." fieeeemegee flee, 28 February 1989, sec. D, p. 1. Derr, Darl I. and Miklavic, Maureen. "Apple Industry Takes Action on Alar." heeieeee,Niee, 9 February 1989, sec. 1, p. 1. Edman. Victor G. Retell 03.113111! for £22511 899.191. Marketing Research Report 952, 1972 Economic Research Service, USDA. Environmental Protection Agency, "Guidelines for Carcinogen Assessment." Eeeeeel 8291§L§£ 51 No. 185 (September 26, 1986): 33992-34003. Foster, W. and Just, R. E. "Consumer Valuation of Health Risk: The Case of Heptachlor Contamination of Milk in Hawaii." Unpublished, June 1984. 76 77 George, P.S. and King, G.A. "Consumer Demand for Food Commodities in the United States with Projections for 1980." Giannini Foundation Monograph no. 26, California Agricultural Experiment Station, 1971. Gibson, Richard. "Apples with Alar Frightened Grocers More than Buyers." 3111.1 semee leeenel, 7 August 1989, Po 3-30 Groth, Edward III. "Alar in Apples." Letter to the Editor, fieieeee, 19 May 1989, p. 755. Hale, Tom. "More to Food Safety than Report Poses." lee geekee, 16 August 1989, p. 20A. Hayes, Radar. ”Determining the Consumer Information Content of Newspapars: A Proposed. Analytical Framework and Illustrative Application-" 119111031 9.: 92115111121; Affairs 23 (Summer 1989): 127-144. Jasanoff, Sheila. "EPA's Regulation of Daminozide: Unscrambl- ing the Messages of Risk." m, Teetmeleey, end Neeee leleee 12 (Summer/Fall 1987): pp 116-124. Johnson, Reed F. "Economic Costs of Misinforming About Risk: The EDB Scare and the Media." Biek Anelgeie Vol 8, No 2. 1988: 261-69. Karst, Tom. "Apple Growers Poker-Faced as Deal's Cards are Played." The Beekee, 20 May 1989, p. 2A. Kmenta, Jan. Elemenee efi Neenemeteiee. New York: The Macmillan Company, 1971. Merritt, Mike. "Core of a Problem: Apple Industry Freas Outcry Over Pesticides May Destroy State Crops." me Seeetee Lines, 5 March 1989, sec. D, p.1. ' Price, David W. and Mittelhammer, Ronald c. "A Matrix of Demand Elasticities for Fresh Fruits." Eeeeen {[eenenel 9.: 89.111511111211131 32911211518 1979: 69-86. Savage, J.A. "Consumer Compel Agricultural Chemical Reation. " W and 59.21.2121: Eel/.1911 60 (Winter 1987): 24-26- Sherrell, Daniel, Reidenbach, R. Eric, Moore, Ellen, Wagle, Jay, and Spartlin, Thaddeus. "Exploring Consumer Response to Negative Publicity." E32119 Beleeiene Begieg 11 (Spring 1988): 13-28. Shulstad, Robert N., and Stoevener, Herbert H. ”The Effects of Mercury Contamination in Pheasants on the Value of 78 Pheasant Hunting in Oregon. " Leer; Beenemiee 54 (February 1978): 39-49. Smith, Mark E. "Consumer Response to Food Contamination: A Case Study of Heptachlor Contamination of Oahu Milk." M.S. thesis, Michigan State University, 1984. Smith, Mark E., van Ravenswaay, Eileen 0., and Thompson, Stanley R. "Sales Loss Determination in Food Contamina- tion Incidents: An Application to Milk Bans in Hawaii." Anerisan Journal of Agrieulturel Egonemies 70 (August 1988): 513 - 520. Sun, Lena H. "Major Area Markets Say Apples Free of Alar." Ihe NeekiegLen Beet, 4 March 1989, sec. D, p. 11. Swartz, David G., and Strand, Ivar E., Jr. "Avoidance Costs Associated with Imperfect Information: The Case of Kepone." Lane Meg 57 (May 1981): 139-50. Taylor, Ronald B. ”Pesticide Battle Lines." 1% Angeles limee, 22 May 1988, sec. 1, p.3. Tomek, William G., and Robinson, Kenneth L. Eeeeeee Eeieee. 2nd ed. Ithaca: Cornell University Press, 1981. U.S. Department of Commerce, Bureau of Economic Analysis. 591221 91 current Busines_. Various issues. U.S. Department of Commerce, Bureau of Labor Statistics. ePI eeeeilee BEECH.- Various issues. HICH GRN in! TRTE U s NIV. LIBRARIES lll‘ Ill IN” ”WWI” 111111111111 9 12589 312 30061