Health Risk andtheDemand for Red Meat: EvidencefromFuturesMarkets RodneyG.Robenstein;Walter N.Thurman Reviewof AgriculturalEconomics,Vol.18,No.4.(Oct.,1996),pp.629-641. StableURL: http://links.jstor.org/sici?sici=1058-7195%28199610%2918%3A4%3C629%3AHRATDF%3E2.0.CO%3B2-7 ReviewofAgriculturalEconomicsiscurrentlypublishedbyAmericanAgriculturalEconomicsAssociation. YouruseoftheJSTORarchiveindicatesyouracceptanceofJSTOR'sTermsandConditionsofUse,availableat http://www.jstor.org/about/terms.html.JSTOR'sTermsandConditionsofUseprovides,inpart,thatunlessyouhaveobtained priorpermission,youmaynotdownloadanentireissueofajournalormultiplecopiesofarticles,andyoumayusecontentin theJSTORarchiveonlyforyourpersonal,non-commercialuse. Pleasecontactthepublisherregardinganyfurtheruseofthiswork.Publishercontactinformationmaybeobtainedat http://www.jstor.org/journals/aaea.html. EachcopyofanypartofaJSTORtransmissionmustcontainthesamecopyrightnoticethatappearsonthescreenorprinted pageofsuchtransmission. TheJSTORArchiveisatrusteddigitalrepositoryprovidingforlong-termpreservationandaccesstoleadingacademic journalsandscholarlyliteraturefromaroundtheworld.TheArchiveissupportedbylibraries,scholarlysocieties,publishers, andfoundations.ItisaninitiativeofJSTOR,anot-for-profitorganizationwithamissiontohelpthescholarlycommunitytake advantageofadvancesintechnology.FormoreinformationregardingJSTOR,[email protected]. http://www.jstor.org WedFeb2713:59:282008 HEALTH RISK AND THE DEMAND FOR RED MEAT: EVIDENCE FROM FUTURES MARKETS Rodney G. Robenstein and Walter N. Thurman The literature on demand shifts for meat This interpretation is at least plausible: medi- is extensive. Since 1977, a large number of cal science has accumulated evidence that diets studies have documented instabilities in aggre- high in cholesterol and fat, both found in red gate meat demand equations estimated from meat, increase the risk of heart disease. How- time series data. (See Braschler; Chavas; Choi ever, the timing of the public's cholesterol and Sosin; Dahlgran; Moschini; Moschini and concern is not obviously the same as the timing Meilke [1984, 19891; Nyankori and Miller; and of the shifts found in empirical meat demand Thurman.) The modal conclusion from these studies. While most demand studies find shifts studies is that per-capita demands for red meats, in the mid-1970s, Shekelle and Liu reported beef, and pork have shifted inwards over time that, as late as 1978, only 13 percent of a and that per-capita demands for poultry and fish sample of Chicago consumers were aware that have shifted out. In studies attempting to pin- too much cholesterol or fat in the diet might point the timing of shifts, the shifts seem to increase the risk of heart attacks. Evidence pre- have occurred, or to have begun, in the mid-to- sented here shows that firm scientific conclu- late 1970s. sions as to the link between heart disease and These studies are not without their critics. cholesterol or fat were not available until the Chalfant and Alston found fault with the para- early 1980s. metric conclusions of the structural shift litera- Annual time series data reveals only so ture; they demonstrated that neither American much about consumer response to health infor- nor Australian aggregate meat consumption data mation. This study explores an alternative, and violate the weak axiom of revealed preference. potentially rich, source of information on Both Wohlgenant and, later, Alston and consumer meat demands: red meat futures Chalfant argued that perceived structural insta- markets. Contracts for live and feeder cattle and bility may be due to the use of inappropriate for live hogs and pork bellies afford the oppor- empirical functional forms. tunity and incentive for traders to register their Almost no structural shift studies have beliefs about consumer demand. Daily futures analyzed causes of the shifts.' However, most market data are analyzed to answer the ques- authors are willing to attribute the shifts to tion: do traders, whose profits depend upon changes over time in the way that consumers accurately forecasting meat demand, revise their view the health implications of eating red meat. demand forecasts when significant information is released on the negative health implications of eating red meat? This research finds that they Robenstein is a farm business consultant with the do not. Illinois Farm Business Farm Management Association. Thurman is a Professor of Agricultural and Resource Economics and Professor of Economics at North Carolina State University. The authors would like to thank Charles R. Knoeber and participants in the Agricultural Economics Workshop at North Carolina State University for helpful comments. 'Two useful exceptions are articles by Brown and conclusion was reached despite the estimation of an Schrader and by Capps and Schmitz. Brown and Schrader autonomous and negative egg demand trend. Capps and developed an index of the frequency of occurrence of Schmitz used the same cholesterol information index in a cholesterol discussion in professional medical journals time series analysis of meat demand. They found that the and. in a time-series demand model, found that the index was correlated with a negative trend in meat demand demand for eggs was negatively related to the index. This as well. Review of Agricultural Economics 18(1996):629-641 Copyright 1996 North Central Administrative Committee 630 REVIEW OF AGRICULTURAL ECONOMICS, Vol. 18>N o. 4. October 1996 Announcement Effects in Red Meat prices that should be affected by red meat Futures Markets health-related information is identified. Finally, in a series of event study regressions, effects of A simple story motivates our study and the information releases are analyzed. These connects consumers' beliefs to observable data: three components are discussed in turn. the demand for a food product depends, in part, upon its perceived health benefits. The health Health-related Research Announcements consequences of eating red meat have been of The Wall Street Journal is the source of concern for the last 20 years and, to the extent cholesterol announcement information. This that beliefs concerning those consequences have daily national publication pervades U.S. busi- changed, one would expect demand shifts to ness and, because of its extensive coverage of have reflected those changes. Press reports futures markets, is likely to be read by futures about health consequences of red meat refer to traders.' Relevant articles were located using scientific studies. Medical researchers are The Wall Street Journal Index from 1971 to continually engaged in projects designed to 1987 and the National Newspaper Index from establish causes of heart disease and the link to 1988 to 1990. In searching for public eating red meat. Consumers learn of scientific announcement dates, key words such as "choles- results in this area in at least two ways. They terol," "health," "beef," "pork," "cattle," and learn about the results of research by reading "hogs" were used. popular media reports on research projects. In The article titles and summaries identified addition, they learn through the advice of their by the indexes were studied to identify those doctors. Doctors may change their own beliefs that discuss dietary cholesterol, blood serum about the red meat-heart disease connection by cholesterol, or heart disease, in connection with reading medical journals but they, too, may red meat consumption. Relevant articles found learn from the popular press. prior to 1983 were sparse and less newsworthy Whenever there is a release of new infor- than those after 1983. In fact, no articles were mation from a professionally credible source, it found for the years 1973, 1974, 1975, 1977, will have some impact on the beliefs of consum- 1979, and 1981. ers and health care professional^.^ If consumers' After culling irrelevant articles, 52 articles beliefs are affected by the new information then remained. Each article was categorized either as their consumption patterns will be affected as strengthening the links connecting red meat, well. If new information serves to strengthen dietary cholesterol, blood serum cholesterol, and consumers' belief in the causal relation between heart disease or weakening one or more of those red meat consumption and heart disease, then links. Thirty-three articles strengthen the red the demand for red meats will shift inward. meat and coronary disease link and 19 weaken Such a shift in consumer demand would also be the link, covering the period 1971 through reflected in an inward shift in the derived, farm- 1990. level demand for red meats. To the extent that A further classification identified each supply curves for red meats are upward sloping, article as strong, moderate, or weak. Four there should be near-term (if not long-term) characteristics were used to rank the articles. reductions in prices. The hypothesized release The four criteria for "strong" articles were that of information that shifts back demand for red they: (1) contained actual news; (2) came from meat over a future period also implies a current a reliable source; 3) received both a prominent lowering of red meat futures prices. size and location in the paper; and (4) took a The methodology consists of three parts. firm stand, providing either positive or negative First, credible new publicly-released information and the dates near which it became public are identified' Next' a of futures 'While 7%e WallStreetJ ou.r nal is certainly .not the only media outlet to which consumers oav attention. it orovides us , ' with aconsistent daily observation on announcements which is 'Consumers may respond to information from non- indexed. One could expand the analysis in the current article by credible sources as well. including announcements from other media. HEALTH RISK AND THE DEMAND FOR RED MEAT Robenstein, Thurman 631 information tied directly to one of the choles- from January 1983 through December 1990 for terol links in question, especially dietary recom- the live cattle (LC), feeder cattle (FC), pork mendations referring to red meat. "Moderate" bellies (PB), and live hogs (LH) contracts. To articles contain information from respectable allow for later testing of long-term versus short- sources, but failed to maintain one of the other term reactions in the market, prices from both three ranking characteristics. Examples of the nearby contract and from the contract "moderate" articles included editorials (not closest to maturing six months from the current actual news), articles of only one or two para- date were collected. For example, in January, graphs and not readily visible (possibly over- the nearby month for live hogs is February and looked because of size and location), and the six-month contract is July. articles which may mention dietary cholesterol To keep a continuous daily data series, the and red meats, but concentrate more on drug price from the nearby contract month is rolled use to lower cholesterol (leaving the signifi- over to the next trading contract month as the cance of dietary cholesterol in question). nearby contract expires. Since the nearby "Weak" articles contained at least one of the contract month becomes more noisy as it con- four ranking characteristics, but tended to be verges to the cash price in the month of expira- less newsworthy, such as from sources with tion, the nearby contract month is rolled over on possible vested interests, and were written the first notice day. First notice occurs on the without a firm positive or negative stand toward last trading day of the month before the expira- the cholesterol links. tion month for FC, PB, and LH and on the Of the 52 articles, nine "strong" articles second to last trading day of the previous month were found, each positively supporting the for LC. For example, the June 1992 contract for cholesterol links, 14 "moderate" articles with LH would roll over Friday, May 29 and the nine positive and five negative links, and 30 June 1992 contract for LC on Thursday, May "weak" articles with 16 positive and 14 negative 28. The six-month contract simply rolls over on links. Since all the "strong" articles and most of the last trading day when the next trading the "moderate" and "weak" articles occurred contract becomes a closer approximation to six after 1982, only those articles between 1983 and months from the current date. For example, on 1990 were considered. Among these 36 articles January 31, 1992 the July 1992 six-month were nine "strong" articles which positively contract for LC rolls over to the August 1992 supported red meat-heart disease links, eight of ~ontract.~ the 10 "moderate" articles were positive, and Possible differences in short-run and long- the nine of the 17 "weak" were positive. Be- run effects are accounted for with different cause none of the "strong" and only two of the versions of the red meat index. One version "moderate" articles took a negative stand to- includes all four red meat contracts in both the ward the cholesterol links, negative articles nearby and six-month contracts. Another in- were not used. The final sample comprised 26 cludes all four red meat contracts, but only in articles supporting the link between either red the nearby contracts. A third version includes meat consumption and blood cholesterol, blood the four red meat contracts, but only in the six- cholesterol and heart disease, or red meat month contracts. If an announcement effect has consumption and heart disease directly. Because both current and long-term effects, reactions the categorization of articles is somewhat should be detected in all four versions of the subjective, the larger sample of 52 articles is index. Otherwise, the longer-term six-month summarized in the Appendix. index should pick up announcement reactions that the nearby index does not, or vice versa. A Portfolios of Red Meat Futures Contracts and fourth version of the red meat index includes a Market Index both live cattle and feeder cattle in the nearby and six-month contracts but excludes the pork Daily futures price data were collected for red meats traded on the Chicago Mercantile 4When measuring price changes, it is inappropriate to Exchange. The data consisted of daily closes include roll-over dates. so they are dropped from the sample 632 REVIEW OF AGRICULTURAL ECONOMICS, Vol. 18, No. 4, October 1996 contracts. Because beef consumption falls more dramatically in the 1980s than pork consump- tion, this portfolio is used to study specific effects that announcements may have on beef where ri,=rate of return on the ith contract on demand. day t, In pi,-natural log of the price of the ith contract on day t, and In p,,,=natural log of the Event Study Regressions price of the ith contract on day t-1. The rate of return on the red meat index is then calculated The methodology in this study was first as: applied by Fama et al. (FFJR). FFJR maintained the following market model: where k is the number of contracts included in where R,,-rate of return on security i at time t, the red meat index. M,=rate of return on the market index at time t, To test effects of announcements on red and e,,srandom error for security i at time t. meat futures, a regression model based on the In this study, the Commodity Research structure proposed by FFJR is formed which Beureau (CRB) index serves as the market includes dummy variables to indicate announce- index.' The role of the individual security is ment dates and the 20-day periods before and played by a portfolio of red meat futures con- after: tracts. The market model is adapted to control for uninteresting market-wide influences on red meat futures prices. The sensitivity of the return of security (futures contract) i to the return on the market index reflects co-movements among the commodities. Correlated movements among where AD, equals 1 on the 26 announcement the commodities may indicate normal adjust- dates; BEF,, equals 1 on the ithd ay before the ments to macroeconomic influences such as announcement; and AFT,, equals 1 on the ithd ay interest rates, inflation, strength of the U.S. after the announcement. Rates of return com- dollar, exports, and supply shocks. Conditioning puted from the red meat and CRB indices are on the CRB index increases the power of tests used for Rt and MI. by holding constant futures price variation If the announcement date, when AD, clearly unrelated to health announcements. equals 1, was the only date on which an effect The rate of return on the red meat index occurred, then p, would be expected to be is constructed such that an equal dollar amount negative and all the y,'s and 0,'s to equal zero. is invested in each of the four red meat futures However, information may be leaked to the contracts: live cattle, feeder cattle, pork bellies, market before the announcement date or mar- and live hogs. The portfolio is rebalanced daily. kets may not digest the announcement instantly. The return on this portfolio is calculated as For these reasons, a 20-day window on either follows. First, a daily return on a particular side of the announcement is introduced. The contract is defined as: sum of p,, the y,'s, and the 0,'s is the regression estimate of the cumulative impact on returns from a single announcement. 'The index "is an unweighted geometric mean of the To test announcement leaks or delays individual price relatives of 27 (changed to 21 after 1986) commodity futures prices. ... The current price value in the measured in months rather than days, a monthly index for each commodity is found by averaging all futures futures returns model also is analyzed. The contracts up to, but not including, those that mature twelve monthly returns model estimated is: months in advance ofthe present date" (Commodity research Bureau, 1984. p. 38). The ratio of current prices to the 1967 annual average price (1967=100) ofeach commodity completes the CRB calculation. See the Appendix for a list of contracts included in the CRB index. HEALTH RISK AND THE DEMAND FOR RED MEAT Robenstein, Thurman 633 where R, and M, now represent monthly rates of change in the red meat index when RRM,, return on the appropriate indexes; AM, refers to RNB,, and R6Mt are used and a 0.25 percent an announcement month dummy variable; and increase if RCAT, is used. Since the CRB index BEF, and AFT, signal months that are one is less representative of cattle contracts alone, month before and after months in which an- its parameter estimate is lower when RCAT, is nouncements are made. The daily market returns used. model is estimated first. Without dummy vari- The estimated announcement date dummy ables representing announcement effects, the parameter (not shown) has an insignificant t- estimate of the market returns model from statistic, so the collective effect of the days on January 1983 to December 1990 is: and around the announcement date are exam- ined for evidence of abnormal market behavior. R, = 0.0057 + 0.5199M, + u,, Testing for a collective effect using the t-statis- (6) (0.0207) (0.0334) tics of the individual dummy coefficients (which with an adjusted R2of 0.1 15 1, a Durbin-Watson represent days on and around the announce- Statistic of 2.016, and 1,858 observations. The ment) is unreliable. In fact, under the null hy- standard errors for the regression parameters are pothesis, one would expect to find one or two shown in parentheses. The red meat index, R,, of the 41 dummy coefficient estimates signifi- cant when such a test is performed at the 5 used as the dependent variable includes all eight percent leveL6 Therefore, an F-test is performed red meat contracts, live cattle, feeder cattle, pork bellies, and live hogs in both the nearby to determine if the dummy variables are jointly significant. At the 5 percent level, none of the and six-month contracts. The parameter estimate f9r the CRB index is significantly different from red meat index models show a significant joint zero. It can partially be explained by the fact F-statistic for the nine strongest announcement that three of the red meat contracts are included dates.' in the index. However, it primarily reflects the An F-test to determine if the sum of the reaction of red meats to other related commodi- dummy estimates is significantly different from ties included in the index, especially grains. zero is also performed. For the nine strongest Significance of the CRB index persists through- announcement dates, none of the models exhibit out all of the market returns testing. a sum of coefficients that is significantly differ- The event study regressions, estimates of ent from zero at the 5 percent level equation (4), are reported in Table 1 for the Beyond testing for statistical significance, economic significance of the results is consid- four versions of the red meat futures index. ered. A 90 percent confidence interval for the Table 2 defines four indices used to measure red meat price changes. Initial regressions that sum of the dummy coefficients is computed for used all 26 announcements (strong, moderate, the following reasons. If zero is contained and weak) showed no significance, individually or joint, of the announcement day dummies. 6Dummy variables include the announcement date. 20 days preceding, and 20 days following. The inclusion of Therefore, only regressions using "strong" dummies for several announcement dates ensures that each announcement dates are presented in Table 1. dummy variable lands on anon-rolluver date, validating all 41 As expected, the CRB market index dummy variables. However, when testing a single announce- parameter estimate is highly significant in the ment date, three contract roll-over dates will occur within this window, leaving three dummy variables invalid. four regressions with different measures of red 'Results reported in Table 1 measure red meat return meat price changes. The four measures are: with the four portfolios described in Table 2. The four RRM, (the change in nearby and six-month beef portfolios represent different combinations of the eight pork and beef contracts. Numerous other portfolios could be and pork contracts); RNB, (the change in nearby constructed from other subsets of the eight contracts. We beef and pork contracts); R6Mt (the change in investigated the same empirical relationships using other six-month beef and pork contracts); and RCAT, portfolios, including portfolios consisting ofjust one of the (the change in nearby and six-month beef eight futures contracts. The key result described in the text, the non-significance of the cumulative announcement effect. is contracts). A 1 percent change in the market robust to choice of portfolio. index is associated with about a 0.5 percent 634 REVIEW OF AGRICULTURAL ECONOMICS. Vol. 18, No. 4. October 1996 Table 1. Daily Market Returns Model Results, Nine Strong Announcement Dates RRM, RNB, R6M, RCAT, Intercept 0.0037 0.0032 0.0062 0.0272 (0.0225)" (0.0246) (0.0215) (0.0189j M, 0.5 130" 0.5050" 0.5219" 0.2407" (0.0336) (0.0367) (0.0321) (0.0282) Sum of Dummy Coefficients 0.1505 0.3718 -0.0708 -0.3100 F-stat (joint) 0.9220 1.0139 0.8426 0.7952 (0.6131) (0.4470) (0.7468) (0.8171) F-stat (sum) 0.0058 (0.9394) 90 percent C.I. (sum) (-3.09. 3.39) Average Returnc 3.9187 Adjusted R' 0.1136 Mean Squared Error 0.800 Durbin-Watson 2.012 n 1,858 Dependent variables are measured in daily percentage changes, not annualized Wumbers in parentheses represent standard errors for coefficients and p-values for F-statistics bIndicates coefficient significantly different from zero at the 5 percent level 'The average of absolute percentage changes of the red meat index over all 41-day intervals in the sample Table 2. Futures Contracts Included in the Four Red Meat Futures Indices Contract RRM, RNB, R6M, RCAT, LC-nearby J J J LC-sixth month J J J FC-nearby J J J FC-sixth month J J J PB-nearby J J PB-sixth month J J LH-nearby J J LH-sixth month J J LC = live cattle, FC = feeder cattle, PB = pork bellies, LH = live hogs. HEALTH RISK AND THE DEMAND FOR RED MEAT Robenstein, Thurman 635 within the interval (as is found), an argument excess returns. Similar results hold for regres- for abnormal returns around the announcement sions that use all positively ranked announce- date is weak. This settles the issue of statistical ment dates: strong, moderate, and weak. Fur- significance. However, a failure to reject a null ther, each of the nine strong dates is analyzed hypothesis can be interpreted in two ways. One in its own regression. All the announcement could conclude either that there is no dates, except one, exhibit similar insignificant announcement effect or that the data are too results. The one significant date is October 5, weak to detect it. But, if the 90 percent confi- 1987 and its regressions are reported in Table dence interval contains only returns that are 3. On that date, an article in The Wall Street deemed to be economically small then the Journal appeared on the first page of the second regression estimates the announcement effect to section, "Doctors' Orders: Cholesterol Study be economically insignificant, and this is found. Calls for Broad Treatment Change." Included The question arises: what is a reasonable were recommendations by the National Heart, measure of the economic significance of a daily Lung, and Blood Institute that "doctors monitor return? Because the sum of the dummy vari- every American adult's cholesterol level and ables in the event study regressions represents prescribe specific diets ...with detailed lists of the cumulative effect, over 41 days, of a single unacceptable and acceptable foods, mostly announcement, it is compared to the uncondi- recommending that patients avoid high-fat tional average of the market's movements over foods... and limit their intake of fatty beef and all 41-day periods. The 41-day average return veal" (p. 13 1). For this announcement date, the is computed by summing the daily returns over F-statistic for the sum of the dummy coeffi- a 41-day interval, taking its absolute value, and cients was significant at the 5 percent level. averaging all of the possible 41 -day returns for Events surrounding the specific date were the entire sample period. The resulting figure further investigated. Cattle prices dropped can be interpreted as the average amount by precipitously in October 1987.' According to which the market goes up or down over a 41- reports in The Wall Street Journal, traders day period. began worrying about falling cattle demand and All of the confidence intervals in Table 1 excess fat cattle inventories as cattle futures are zero, establishing the lack of statistical began a two-week slump starting in the last significance of the announcement effect. As to week of September. The week of October 19 economic significance, the lower (negative) end through 22 saw cattle futures drop by $4.50 per of the confidence interval is compared with the hundredweight, making for "the worst week average absolute 41-day return. For example, since spring 1986, when the U.S. announced the 90 percent confidence interval for RRM, is plans to pay dairy operators to slaughter their bounded by -3.09 percent and 3.39 percent. herds" (Siconolfi, p. 40). On October 26, cattle Relative to the 41-day average movement of and hog contracts fell by the permissible limit 4.92 percent, the confidence interval bounds of $1.50 per hundredweight and pork bellies by support with 90 percent confidence that excess $2.00 per hundredweight. Analysts blamed a returns due to announcements are economically bearish cattle-inventory report from the United insignificant for the nine strongest announce- States Department of Agriculture and stock ment dates. The 41-day average movement is market declines. Given the extraordinary cir compared with the low, and negative, end of the cumstances in the cattle market during this confidence interval because of interest are period and that no Wall Street Journal commen- negative effects of cholesterol announcements tary at the time attributed the meat futures drop as demonstrated by negative returns in the red to the October 5 health announcement, the meat market. abnormal returns were unlikely to have been The 4 1-day average movement in the red caused by cholesterol announcement effects. meat index is reported in Table 1 for each of the four index versions. For the nine announcements, each of the four models exhib- 'Coincidentally or not, October 19, 1987 was the date its statistically and economically insignificant of the stock market crash. 636 REVIEW OF AGRICULTURAL ECONOMICS: Vol. 18: No. 4. October 1996 Table 3. Daily Market Returns Model Results, Individual Announcement Date, October 5, 1987 RRM, RNB, R6M, RCAT, Intercept 0.0122 0.01 13 0.0130 0.03 12 (0.0209)" (0.0229) (0.0199) (0.0 173) M, 0.5 182b 0.5107b 0.5257b 0.2367b (0.0337) (0.0369) (0.0321) (0.0280) Sum of Dummy Coefficients -1 1.998 -11.538 -12.458 -9.8128 F-stat (joint) 1.2978 1.1103 1.4389 2.0301 (0.1065) (0.2971) (0.041 1) (0.0002) F-stat (sum) 4.675 1 3.6104 5.5566 4.5367 (0.0307) (0.0576) (0.0185) (0.0333) 90 percent C.I. (sum) (-21.10, (-21.50, (-21.13, (-17.37: -2.90) -1.58) -3.79) -2.26) Average Return' 4.9187 5.7158 4.3322 3.6120 Adjused R2 0.1205 0.0978 0.1360 0.0588 Mean Squared Error 0.7938 0.9506 0.7200 0.5471 Durbin-Watson 2.007 1.993 1.994 2.023 n 1,858 1,858 1,858 1.858 Dependent variables are measured in daily percentage changes, not annualized. Wumbers in parentheses represent standard errors for coefficients and p-values for F-statistics bIndicates coefficient significantly different from zero at the 5 percent level. 'The average of absolute percentage changes of the red meat index over all 41-day intervals in the sample The monthly market returns model is Conclusions tested next. Without dummy variables represent- ing the announcement month or the month The most obvious times that public before and after, the empirical estimate of the knowledge of the links between red meat con- market model is: sumption and heart disease could be said to have changed were identified. On those dates, RRM, = 0.0072 + 0.3501Mt + v,, reactions in futures markets, the most obvious (7) place for immediate adjustments to occur, were (0.021 1) (0.1393) examined. No evidence was found of any such with an adjusted R2of 0.0530, a Durbin-Watson adjustment. Further, failure to find evidence of Statistic of 1.8626, and 96 observations. a reaction is not due to weak data. Probable The announcement effect regressions are bounds on the size of a futures price reaction to repeated with the monthly average data, but health information were identified and were again the null hypothesis of no announcement small in economically relevant terms. effect cannot be rejected. Further, the confi- The link between health perceptions and dence intervals for the estimated effect are red meat demand has been discussed extensively always small relative to actual average returns. and previous empirical work has shed useful, For individual announcement date regressions, albeit indirect, light on the question. What only the October 1987 announcement proves seems clear is that structural change analysis of significant. Only the announcement month is time series demand models has not settled the significant, not the month before or the month issue of whether or not red meat demands have after. This result strengthens the argument that shifted. The current article examines a different both the monthly and daily tests for this period sort of information and expands the set of tested are affected by extraneous forces. implications of a health-linked demand shift. HEALTH RISK AND THE DEMAND FOR RED MEAT Robenstein, Thurman 637 A different interpretation of the results is Chalfant; J.A, and J..M.Alston. "Accounting for Changes in that futures markets are not, in fact, likely Taste.'' Journal Political Economics 96(1988):391-410. places to look for evidence of shifts in demand. Chavas, J.P. "Structural Change in the Demand for Meat.'' Futures prices are volatile and their movements American Journal of Agricultural Economics 65(1983):148-53. reflect many short-term phenomena. One could Choi, S. and K. Sosin. "Testing for Structural Change: The argue that the changes in red meat demand are Demand For Meat." American Journal of Agricultural gradual and that they are unlikely to register in Economics 72(1990):227-36. futures prices, even if they are cumulatively Commodity Research Bureau, Inc. Commodity Year Book. large. Jersey City, NJ: Commodity Research Bureau. Inc., There is a difference between taking a 1984. position in a live cattle contract for delivery two Dahlgran. R.A. "Complete Flexibility Systems and the Sta- months hence and forecasting demand for beef tionarity of U.S. Meat Demand." Western Journal of one or two years hence. However, long-term Agricultural Economics 12(1987): 152-63. demand shifts, whatever their cause, show up at Fama, E.F.. L. Fisher, M.C. Jensen. and R. Roll, "The some time in futures prices and traders who Adjustment of Stock Prices to New Information." International Economics Revue 1O (1969): 1-2 1 anticipate those shifts will profit. If identifiably important public releases of information have no Moschini, G. "Testing Preference Change in Consumer Demand: An Indirectly Separable Semiparametric discernible reaction in the futures market, then Model." Journal of Business and Economic Statistics two c~nclusionsa re possible. Either reasonable 9(1991):111-17. forecasts of meat demand should be unchanged Moschini, G. and K.D. Meilke. "Parameter Stability and the by scientific evidence on the health conse- U.S. Demand for Beef." Western Journal of Agricul- quences of eating red meat or futures traders are tural Economics 9(1984):271-82. ignoring this information. We do not believe . and K.D. Meilke. "Modelling the Pattern of futures traders consistently and persistently fail Structural Change in U.S. Meat Demand." American to react to important information. Rather, a lack Journal ofAgricultura1 Economics 71(1989):253-61 of reaction from the futures market calls into i"iationa1 "Vewspaper Index. Foster City. CA: Information question the link between the scientific pro- Access Company. 1988-1990. nouncements and consumer behavior. Nyankori, Y.C.O. and G.H. Miller. "Some Evidence on Applications of Structural Change in Retail Demand [Received June 1994. Final version received for Meats.: Southern Journal ofAgricultura1 Econom- July 1996.1 ics 14(1982):65-70. Shekelle, R.B. and S. Liu. "Public Beliefs About Causes References and Prevention of Heart Attacks." Journal American Medical Association 240(1978):756-58. Alston, J.M. and J.A. Chalfant. "Can We Take the Con Out of Meat Demand Studies?" Western Journal of Agri- Siconolfi, M. "Commodities." The Wall Street Journal. cultural Economrcs 16(1991):36-48. (1987, October 23). p. 40. Braschler, C. "The Changing Demand Structure for Pork Thuman. W.N. "The Poultry Market: Demand Stability and and Beef in the 1970s: Implications for the 1980s." Industry Structure." American Journal of Agricultural Southern Journal of Agricultural Economics Economics 69(1987):30-37. 15(1983): 105-10. Wall Street Journal Index, The. Ann Arbor, MI: University Brown, J.B. and L.F. Schrader. "Cholesterol Information Microfilms International, 197 1-87. and Shell Egg Consumption." American Journal of Wohlgenant. M.K. "Estimating Cross Elasticities of Demand Agrrcultural Economics 72(1990):548-55. for Beef." Western Journal of Agricultural Economics Capps. Jr., 0,and J.D. Schmitz. "A Recognition of Health 10(1985):322-29. and Nutrition Factors in Food Demand Analysis." Western Journal of Agricultural Economics 16(1991):21-35.
Description: