ABSTRACT This paper hypothesis es that the Mad Cow Disease (MCD) scare in Europe brought on by the ban on exports of British beef has had a negative impact on beef consumption in the United States. This is in light of the fact that MCD has no direct impact on food safety in the US. Using monthly data an Almost Ideal Demand System containing an intercept dummy capturing developments in the MCD media developments is estimated. While concerns over the results generated by the model are raised, the estimated structural change variable is indeed found to be significant for the beef share equation suggesting that media in the case of MCD has had a negative impact on consumer demand for beef. TABLE OF CONTENTS Abstract i Table of Contents ii Introduction: 1 Literature Review 2 Methodology 3 Data 6 Estimation Procedure 9 Results 11 Discussion and Critique of results 15 Conclusions 16 References 18 INTRODUCTION: In March of 1996 the British beef industry was dealt an incredible blow when a European Union commission imposed a worldwide ban on the export of UK beef. The ban was imposed after an outbreak of mad cow disease (MCD) in Britain.
MCD is the laymen term for Bovine Spongiform Encephalopathy (BSE), a disease that leads to deterioration of the brain tissue in beef and dairy cattle. It is the hypothesized link between MCD and Creutzfeldt Jakob Disease (CJD), the similar condition in humans, which led to the imposition of the export ban on British beef. This ban effectively crippled the UK beef export market in addition to sending a shock throughout Europe concerning beef safety. News of this ban spread throughout the world portraying images of infected cattle stumbling and falling unable to maintain their balance. While the scientific community was, at the time, unsure of the nature of the link between MCD and CJD, if any, the effects of the outbreak may have been much further reaching than Europe. It is arguable that North America was also impacted to some extent, whether long term or short, by the MCD scares.
... BSE, and hoof and mouth diseases could contaminate the beef in the United States. I do ... of problems with beef in other countries, people assume that diseases such as mad cow diseases, also known as ... of how disease outbreaks are few and far between, and that the nutritional value of beef is ... the following reasons to justify my claims: because beef is an excellent, more balanced, source of protein ...
In the days that followed the ban on UK beef, beef futures in Chicago fluctuated as traders were torn between two economic forces. The first was the expected increase in demand. Markets formerly serviced by UK beef must now be filled with beef from other countries. The second impact deals with the adverse effects the outbreak could have on consumer demand. North American consumers were also exposed to a great deal of media propaganda as most, if not all, North American newspapers and television news broadcasts, national and local, carried stories of the outbreak and the horrific consequences of CJD. If concerns of food safety in the United States and Canada were never of concern there would have been no need for the United States Department of Agriculture and Agriculture and Agri-Food Canada to issue statements testifying to the safety of each of the countries respective beef.
This paper looks at the short-term impact of the MCD media propaganda on consumer demand for beef in the United States. Due to the fact that the supply of beef in the United States is completely risk free, or at very most extreme minimal risk, in terms of MCD and the risk to the beef supply, the scientific side of the story need not be considered. Changes in US consumers’ demand beef, ceteris paribus, can therefore be credited to changes in preference resulting from media propaganda. The hypothesis studied in this paper is presence and prevalence of media propaganda dealing with the threat of MCD resulted in a negative impact, at least in the short term, on demand for beef in the US. There are two primary purposes for setting out to study the fore mentioned problem. The first is to satisfy the authors own curiosity, in an empirical manner, as to how “fickle” consumers preferences actually are.
... opposite way. Here are some impacts of media on our society. Positive Impacts: Media and society: The electronic media direct social change and values. The electronic ... Impacts: Objectivity and Media: The objectivity of media is becoming increasingly questionable, the coverage of issues of national interest demand care and restraint. The media ...
The second deals with the availability of data and the realization the USDA has readily available, extensive data for livestock consumption. The organization of this paper is as follows: 1) a brief review of the literature relating to the impact of media and propaganda on the public; 2) an overview of the methodology used to evaluate the problem and the data used; 3) the type of estimation used; and 4) results are presented with a critique of the results. Summary and conclusions follow. LITERATURE REVIEW It can be said that the impact of media, correct or otherwise, also has considerable impacts on consumers’ perceptions, and preferences. The literature is quite extensive in this area and all seem to report a similar result.
The public is indeed impacted by the media. A few examples are listed here with a more comprehensive review of the literature contained in Roberts and Bachen. In the article by Page, Shapiro, and Dempsey, the authors find the content of television news accounts for a proportion of aggregate changes in US citizens political preferences. Cook et al. uses an experimental design to look at the impact of a single media event on the general public and policy makers and find media does indeed influence these groups. When evaluating concepts and ideas like freedom of the press, Graber suggest that the costs of such concepts be included with the benefits.
The preceding literature review is quite abridged as much is omitted. Advertising could also be reviewed here and would return a similar result, consumer behavior impacted by what is seen and heard. METHODOLOGY To evaluate the impact of structural change a suitable demand system is needed. For the purposes of this paper the Almost Ideal Demand System (AI model) derived by Deaton and Muelbauer is implemented. This model has several favorable features that lend favorably to analysis of structural change, the foremost being its linearity. The AI model has been used in a number of studies to evaluate structural change in meat demand (see Moschini and Meilke, and Eales and Unnevehr).
... to access the company systems; and Transmitting personal data via WLAN connections can be compromised by someone ... companies IT infrastructure) to gain access to personal data using multiple resources4. The areas affected by workstations, ... violations, where some users continue to leave private data in the open where others can see it; ... IMPACT OF A DATA CLASSIFICATION STANDARD Being a mid-level ...
To test for structural change in this study an intercept dummy is introduced in a manner similar to that employed by Eales and Unnevehr. However, this dummy is introduced subjectively in the sense that it appears only for certain dates when a change in demand is expected to occur. In other words, this dummy attempts to capture the short term effects rather than looking for a permanent structural change (more explanation of this will be given shortly).
In addition to the structural change dummy a trend variable and eleven monthly seasonal dummies are introduced.
The stochastic version of the model is thus written 1) where: goods are indexed i = 1… 5; t is a time trend variable; r is a structural change dummy with value 1 when a shock is hypothesized and zero otherwise; d is a monthly dummy; x is the monthly disposable income; and. 2) Due to the non-linear relationship in P, when estimating the model this parameter is approximated using the Stone Index. To maintain the theoretical properties of homogeneity in prices and income, and Slutsky symmetry the following equation and cross equation restrictions are imposed: ; 3); 4); and 5).
6) The first condition is homogeneity condition, the second two are adding up, and the fourth restriction is Slutsky symmetry.
However, in my specification there are thirteen additional variables that must also satisfy adding up. The model specification is thus complete by imposing the following thirteen constraints; 7); and 8) for = 1… 11. 9) As mentioned previously, the structural change dummy is used subjectively in the sense that a value of one is entered for only those months hypothesized to be negatively affected by the media. For the study at hand this corresponds to media related to MCD, particularly the type of news coverage that would be expected to lead to a negative change in consumption behavior for beef. The following lists chronology of events and the specific months when the author feels media reporting on MCD would be such to impact consumers choices.
The chronology is as follows: 1) March 1996 – McDonalds suspends sale of British beef products, – European ban on British beef imposed, 2) April 1996 – Britain offers a host of solutions to deal with the MCD epidemic including the slaughter of thousands of animals, 3) May 1996 – British prime minister John Major states that Britain will no longer cooperate with EU in business until the ban is eased, 4) June 1996 – European Court of Justice rejects Britains bid for immediate removal of the ban, 5) July 1996 – EU scientists say BSE can infect sheep; 6) August 1996 – British coroner finds rules that a 20 year old vegetarian who died of CJD contracted the disease from eating meat as a child, 7) October 1996 – A team of scientists led by Professor Collings at the Imperial College School of Medicine at St Mary’s, London find evidence of link between a new variant form of CJD and BSE, 8) November 1996 – EU steps up research funding for BSE, 9) January 1997 – The US Food and Drug Administration (FDA) proposes ban slaughtered animal parts in livestock feed because of links to MCD, 10) March 1997 – World Health organization announces that people at risk from CJD should be banned from giving blood, 11) April 1997 – Beef, corn and soybean futures fell on the Board of Trade with a newspaper report that an Indiana man died of an ailment linked to mad cow disease, 12) June 1997 – The British government is to extend ‘mad cow’ controls to sheep because of fears that they may also have become infected with the “fatal brain disease”, and will order the compulsory slaughter of all sheep suspected of infection with scrapie, 13) July 1997 – Burger King pulls out of France after 16 years, half of customer loss due to BSE fears, 14) September 1997- Scientists claim to have “proved” that MCD disease has caused CJD, 15) October 1997 – three unconfirmed cases of CJD in Florida, 16) December 1997 – US restricts the import of cattle, sheep and some livestock products from 21 European nations until there is proof of no ‘mad cow’ risk from them, 17) January 1998 – US beef producers go to court with Oprah Winfree over comments made about beef and MCD. (sources: NY times, CBS news, Wall Street journal, Reuters World Report, London Times) The discussion of MCD in the US seems to taper off greatly after the Winfree trial in Texas and is thus no major news reports were beyond January 1998. In total seventeen months reporting news that were deemed by the author to have impacts adversely affecting consumption for beef in the US. DATA The MCD epidemic in Europe started in early 1996 which makes the use of annual and even quarterly data unacceptable in terms of estimating a demand system targeted at capturing the impact of media on consumer behavior.
... produced by a slowly changing magnetic force." This was Michael Faraday's equation that sparked new interest ... best of the equation. "A lot of electricity was produced by a rapidly changing magnetic force, whereas ... had discovered how airplanes flew. Using Bernoulli's equation, one can see that since the upper jet ... square of the distance between them.This equation told scientists in NASA how to escape ...
... used to calculate price elasticity: •Test markets •Analysis of historical sales data •Conjoint analysis Price elasticity is always negative, although ... , this negative value will lead to a negative price elasticity of demand value. Determinants •Substitutes: The more substitutes, ... the higher the elasticity, as people can easily switch from one good to ...
Ideally data available at the weekly would be ideal, however such a data set would be difficult to compile. Instead, monthly data are used. The demand system estimated makes use of beef, pork, lamb and mutton, and chicken disappearance and price data compiled from publications of the United States Department of Agriculture (USDA).
Per-capita monthly consumption of the four above meat groups is all computed in the same fashion: production plus imports less exports all divided by the US population. Beginning and ending stocks are not contained in this data set, as such data were not readily available. It is thus assumed that these carry-overs and carry-ins remain relatively constant or are negligible.
The disappearance data for beef, pork, and lamb is relatively straightforward in one all encompassing category, that is total pounds of production. To represent disappearance figures for chicken, broiler data are used. Figure 1 depicts per-capita consumption for each of the four meat groups. Figure 1: Monthly per-capita Consumption by Meat Type (source: USDA) From the figure 1 we see that beef consumption appears to hold relatively constant and is much unchanged throughout the twenty-year period. The above graph is recent enough that the apparent structural change between the 1970’s and 80’s that yielded a fair amount of study (see Eales and Unnevehr 1993, and Moschini and Meilke 1989) does not appear. It is tough to see from the above graph whether or not there has been a significant change in beef consumption for the time near and after 1996.
Such a shift cannot be discerned from a visual glance due to the variability of the data. Pork also appears to remain relatively unchanged as does lamb which is just barely visible at the bottom of the figure. The increase in chicken consumption is fairly uniform except for a spike that occurs in and around mid to late 1997. Closer inspection of the raw data revealed that this was due to a large jump in production during the months of June, July, August, and September of 1997.
... ——Teaches——Class A conceptual data model identifies the highest level relationships between the different ... entities, whereas physical data models represent how the model will be built in the database. ... A physical database model shows all table structure including column name, column data type, column constraints, primary ...
No adjustments were made to account for these seemingly odd observations. Price data, as with the consumption data, are also compiled from USDA publications and are reported in nominal dollars per pound. These price trends are reported in figure 2. Figure 2 – Monthly Commodity Prices (USDA) The model to be proposed in following section aims to capture impacts within the red meats group in addition to movements away from red meats all together.
To capture this impact a composite good is used to develop a more complete demand system. To compute this monthly disposable income is required. Yearly data from the Statistical Abstract of the United Stakes publications for per-capita consumption expenditures are used to derive monthly income using linear approximation. This data is shown in figure 3. Figure 3 – Personal Income Spent on Consumption Expenditures (Source: USDA) Income growth appears to growth at roughly the same rate throughout the study period. ESTIMATION PROCEDURE The paper by Eales and Unnevehr, which employs a similar method of dummy intercepts to capture structural change as in this paper, estimate an AI model in first difference.
For the purposes of this paper however the standard AI model given above is used 10) Where P is an approximation of P using the Stone index (i. e. ).
The error term, u, satisfies the following stochastic assumptions: for all i; 11) for all i; 12) for all i, j and for all t; and 13) for all i, j and all t s. 14) The above assumptions are incorporated in the estimations, which makes use of the Seemingly Unrelated Regression (SUR) procedure. From the estimated share equations, Marshall ian elasticities may also be calculated directly.
The equations for calculating own price, cross price and income elasticity are given here as; 15); and 16) respectively. 17) If the elasticities are to be calculated at the mean, then by normalizing prices and income, without loss of generality, to a value of 1 (i. e. p = x = 1 for all p) the first two equations can be re-written: ; and 15′).
16′) The share equation, at the mean, can then be written as. 18) One further simplification can be made to ease the computation of elasticities. If the time trend and dummy variables, both structural and seasonal, are normalized to zero, once again without loss of generality, the share becomes equal to the intercept… The own price, cross price, and income elasticities are then written as follows: ; 15”); and 16”) respectively. 17′) Due to the adding up restrictions, only n-1 equations need be estimated. For this paper, the share equation for the composite good is omitted for estimation.
RESULTS The 4 equation AI demand system with the homogeneity and symmetry restrictions imposed is estimated using TSP statistical software version 1. 15. The results from these estimations are provided in table 1. The additional parametric restrictions imposed by adding up are used to derive the parameter estimates for the composite good.
The fit of the data as indicated by the R-square indicates a good fit for the beef and pork share equations, fair for the lamb share equation, and poor for the chicken share equation. The lower bound for the Durbin-Watson test for serial correlation at the 1% level of significance with 150 observations and 19 parameter estimates (20 less the intercept) is 1. 355 and the upper bound is 1. 913. The DW statistic for beef, pork, and lamb all fall into this region suggesting that autocorrelation may be a problem. The DW statistic for chicken is found to be well below the lower limit suggesting that autocorrelation is a factor.
Table 11, 2 Seemingly Unrelated Regression Parameter Estimates for the Basic AI Model Equation Parameter beef Pork Lamb Chicken CG 3 Intercept 17. 35 8. 813 0. 147 6. 004 -31.
314 (279. 56) (232. 149) (21. 92) (101. 82) (na) Trend -0. 43 -0.
051 -0. 0125 0. 1677 0. 3258 (-9. 77) (-1. 94) (-2.
006) (4. 396) (na) Structural -0. 399 -0. 5157 -0. 0125 0. 684 0.
2432 Change (-2. 394) (-5. 028) (-0. 67) (4. 54) (na) beef -6. 99 -4.
205 -0. 028 -1. 821 13. 044 -5.
16) (-6. 741) (-0. 181) (-2. 368) (na) Pork -4. 205 2. 855 0.
026 -0. 651 1. 975 -6. 741) (5. 562) (0.
3004) (-1. 354) (na) Price Lamb -0. 028 0. 026 0.
11 -0. 314 0. 206 -0. 181) (0. 3004) (2. 297) (-3.
019) (na) Chicken -1. 821 -0. 651 -0. 314 6. 651 -3. 865 (-2.
368) (-1. 354) (-3. 019) (7. 503) (na) CG 13.
044 1. 975 0. 206 -3. 865 -11.
36 (na) (na) (na) (na) (na) Expenditure -0. 642 -247. 335 -0. 338 -23. 47 272. 785 (-21.
56) (-14. 82) (-0. 092) (-1. 156) (na) Jan.
0. 221 -0. 061 0. 0071 0. 271 0.
5619 (1. 073) (-0. 486) (0. 34) (1. 434) (na) Feb.
1. 539 -0. 387 0. 00853 0. 682 -0. 84253 (7.
45) (-3. 072) (0. 404) (3. 606) (na) Mar.
-0. 479 -1. 309 0. 00462 0. 111 2. 67238 (-2.
317) (-10. 37) (0. 217) (0. 591) (na) Apr.
0. 725 -0. 14 0. 0216 0. 7311 -0. 3377 (3.
505) (-1. 113) (1. 0129) (3. 86) (na) May 0. 476 -0. 575 0.
0112 0. 648 0. 4398 Seasonal (2. 298) (-4.
537) (0. 5233) (3. 413) (na) Dummies June 1. 408 -0.
93 -0. 0006 0. 865 -0. 3424 (6. 781) (-7. 339) (-0.
0268) (4. 559) (na) July 1. 891 -1. 156 -0. 00409 0. 907 -0.
63791 (9. 218) (-9. 238) (-0. 188) (4.
842) (na) Aug. 1. 532 -1. 412 0. 0025 0. 84 0.
0375 (7. 392) (-11. 16) (0. 115) (4. 427) (na) Sept. 2.
093 -0. 894 -0. 00588 1. 113 -1. 30612 (10. 102) (-7.
067) (-0. 273) (5. 861) (na) Oct. 1.
27 -0. 571 0. 0081 0. 787 -0. 4941 (6. 126) (-4.
506) (0. 379) (4. 144) (na) Nov. 1. 813 0. 322 0.
0081 0. 82 -1. 9631 (8. 755 (2. 55) (0. 384) (4.
332) (na) R-square 0. 9845 0. 960 0. 839 0. 437 (na) DW 1. 65 1.
69 1. 62 0. 95 (na) 1 All parameter estimates have been increased by a factor of 1000. 2 Values in parentheses represent t statistics.
Stars denote probability of type 1 error 10%, 5%, 1%3 Composite Good. Turning to the parameter estimates and first considering the trend variable, the share of beef as a portion of income has been decreasing over time as has been the popular belief. Pork and Lamb also seem to be showing a decrease over time. Chicken share however is on the rise as is consistent with the popular belief that chicken consumption is on the rise. Beef consumption tends appears to be quite sensitive to seasonality as do pork and chicken. The seasonal results for these three variables returned quite impressive results.
Lamb in contrast appears to effect little by seasonality. When the dummy variable for structural change is introduced for the 17 months presented in the methodology section there does appear to have been a shift in demand for beef. This period of time is also accompanied by a significant shift in pork and chicken demand. Pork experienced an inward shift in demand while chicken demand shifted outwards.
Lamb demand did not respond at all to the structural change parameter. The implications of these findings are discussed further in the next section. The last section of results presents the elasticity matrix. While this information is not used explicitly to determine the extent or presence of structural change, it is useful in determining how well the model performs in terms of theoretical prediction.
The elasticity matrix is given in table 2. Table 2 Elasticities beef pork lamb chicken CG beef -0. 76 0. 08 0. 00 0.
12 36. 60 pork -36. 05 0. 01 -0.
43 0. 01 0. 09 lamb 27. 38 -27. 06 -0.
15 0. 20 -0. 25 chicken -2. 12 3.
61 -1. 29 -0. 24 -0. 17 CG -0. 05 0.
13 3. 14 -2. 91 0. 00 Income -0. 01 0. 00 -0.
01 -1. 93 1. 94 The commodity groups in the horizontal row across the top of the table represent the dependant (quantity) and the vertical row to the left represents the independent (price) effect. The own price elasticities are contained in the shaded cells. Income elasticities appear in the lower most row. DISCUSSION AND CRITIQUE OF RESULTS Upon first glance of the results in table 1 the model seems to perform in a satisfactory manner if one ignores the potential autocorrelation problems.
There are a large number of significant variables and the structural change parameter, act in the desired direction for the most part. As mentioned previously the fit for beef, pork and lamb is quite good while poor for chicken. The introduction of the eleven seasonal variables for each of the share equations, while requiring an additional 44 degrees of freedom did indeed prove beneficial as many of the seasonal variables are found to be significant except in the case of lamb. Demand for beef has the expected inward shift representing a decrease in demand, supposedly caused by the incidence of media related to developments in the MCD saga. The chicken shift can also be supported in this same context suggesting that chicken and beef are substitutes and that MCD has no biological effect on chickens, or there has yet to be any evidence presented. Lamb consumption shows no response to this variable.
If one were to predict a direction for this variable it would be in the negative direction since scrapie, the form of MCD in sheep, entered the media in June of 1997. Pork in contrast is a puzzle. A-priori expectations of the author would predict a move similar to that of chicken if any move at all. That is, one would most likely expect the demand for pork to shift outwards as the demand for beef shifts inwards. Never the less the model does not reciprocate this predicted move. The elasticity matrix, which offers insight into the theoretical soundness of the model estimated, seems to indicate problems with the model.
Firstly, the own price effects for all goods should be negative, at least for any goods of concern in this paper. This is indeed the case for beef, chicken and lamb. However the opposite is found pork and the composite good. While the standard errors have not been approximated to test for the significance of these elasticity estimates, the negative own price effect for pork does pose a problem.
This is a possible indication of a problem with the estimation of the model. This helps explain why the structural shift result reported earlier in the pork market needs to be treated cautiously. The second problem deals with income elasticities. The AI model is designed in such a fashion to let the data speak for itself rather than imposing unrealistic assumptions implicitly in the model as in the case of a linear demand system.
The income elasticities of all the meat groups is indeed less than one as is consistent with the belief that food products have an income elasticity of less than 1. It is however difficult to believe that a product like beef is an inferior good. Noting that the beef category is for all beef consumption in this model, few people would include a T-bone steak into the same group as macaroni and cheese. Once again, caution must be taken when interpreting these elasticity estimates as the standard errors have not been estimated. These elasticity estimates are thus able to give an indication of how the data seems to be speaking through the model and should not be taken too seriously. CONCLUSIONS Now that the results generated by the model have been thoroughly criticized, a few saving comments are made in support.
The primary purpose of this paper was to ascertain the impact of adverse media on consumer behavior. Application to the MCD epidemic in Europe and the impact on beef consumption in the US is used to explore this issue. Keeping in mind the criticisms of the results generated by the model presented, the author feels that a humble statement addressing the adversity of MCD news propaganda on beef consumption in the US can be made. That is, the media attention given to the MCD outbreak in Britain and the possible links to CJD in humans appears to impact beef consumption in the US.
This finding is in light of the fact MCD poses no threat, in the past or immediate future, to the US beef supply. The findings presented in this paper, when considered with caution, do seem to support the deeper notion that consumers do not think as much for themselves as given credit for. Is the consumer always right, or is the consumer just a pawn reacting to whatever is thrown into their open, ignorant hands? This question, while harsh in nature, may lead to an interesting answer when one considers the amount of resource that enters into advertising and media reporting. Why would a consumer ever pay twice as much for a pair of running shoes just because they bear a tiny insignia? Why do voter polls switch back and forth during a campaign daily based on a teary eye or a kiss in public? These questions are far beyond the scope of this study but would provide intriguing insight into how consumer behavior is impacted by irrelevant propaganda. If consumers are knowledgeable and are able to discern in the news media what impacts them and what does not, then something like MCD should have no impact. Otherwise the consumers preferences and consumption behavior may be swayed on a whim…
and Brutus is an honorable man? REFERENCES Cook, F. L. , Tyler, T. R. , Goetz, E. G.
, Gordon, M. T. , Process, D. , Leff, D. R. , and Moloch, H.
L. “Media and Agenda Setting: Effects on the Public, Interest Group Leaders, Policy Makers, and Policy.” Public Opinion Quarterly, 47 (spring, 1983), 16-35. Deaton, A. , and Meullbauer, J. “An Almost Ideal Demand System.” American Economic Review. 70 (June 1980): 312-26.
Eales, J. S. , and Unnevehr, L. J. “Simultaneity and Structural Change in U. S.
Meat Demand.” American Journal of Agricultural Economics. 75 (May 1993): 259-268. Green, R. , and Alston, J. M. “Elasticities in AIDS Models.” American Journal of Agricultural Economics.
72 (May 1990): 442-45. Moschini, G. , and Meilke, K. D. “Modeling the Pattern of Structural Change in U. S.
Meat Demand.” American Journal of Agricultural Economics. 71 (May 1989): 253-61. Page, B. I. , Shapiro, R. Y.
, and Dempsey, G. R. “What Moves Public Opinion.” The American Political Science Review. 81 (Mar. 1987), 23-44.
Roberts, D. F. , and Bachen, C. M. “Mass Communication Effects.” Annual Review of Psychology. 32 (1981): 307-56.
Stigler, G. J. , and Becker, G. S. “De Gustibus Non Est Disputandum.” The American Economic Review, Vol.
67, No. 2. (Mar. , 1977), pp. 76-90.