~ ~ - _ - 9" ”4.3 , - HIV-'1. .‘g“"‘ I‘ _.;.I" " ,1: \QI‘IJU‘ 9'9 9‘47.
'..., 9 9' ' .{r 9 9|r."' wmrw - .9 .'.‘.‘,l", |“‘IH.~"V’”"|:9U" ‘
9‘, 9 "9‘ 9 -| " .'L- 9‘ '9 . 9‘. "9 y; .‘9 ’ 9m A". “”5... 39 ~.{ 9"9,‘.":,9..‘.

' 9" “I - _ 1"" H“ ‘ 1' ’ 1 .‘h‘ .3“; ‘ :l: . {:9 ..t",|‘"'ut"[n’"- 9-. :Id' {
,i',‘u.‘. , " 3- "99 .'.‘ u .-
‘ 1“ u" via I -‘ 'l‘ 9 “Ctr"; I” "I‘I'li
.,' mg] “'99., .. .2 ', .‘ I,

  
   
   
   
     
  
 
   
    
   

, 9 >"vlqw‘ 9, '- |'_|r ”n _
- _ Q .." vu ’ ll] u‘
. rt . .A . ’
.. .‘J . o" 1.“; ”tad.“ . ‘ "I
. . 1“

1-”)?‘19‘21-{3‘ 3v
.‘ ' :9

  

9 .
.' ' II.‘
'9,

   
   

u
I ll
.I9. .

 

 

-‘+ "91'3“: u
«‘5 "'W Mffi‘ ‘
Ifly‘a“.
L7-
0““

k

I

A 5‘

. - 'xv
jg“. J'A

   

"L 9 ‘5:
”9:9 99%

xx
Q "

93%;“ t

'31"-

L— ‘
.1 ("At-T?“ M'
Q“ ' .p:
b I.

93,

  
 

. . l

”'I‘Jfl‘ ‘ ‘ .
'\ I
9" . _

EM Vi" -'

A
D

I v“. "T? 1 “2. | -. '1 9.
. * 9:? ..;_r3?\-1..qr:&ap,"LL-2
' (I, 'I ' . 3‘6 g') 0’

  
 
  
 
 
  
 
 

 
  
      

    
 
   

  
   
    
    

H’Y , . . s. .
“ firfl ‘9‘" (.fi.‘ I. "“3? I}... 1"." N f‘ HT“ ‘
_j .| . u. 1'9: {3}...- ' _‘. .;,' ZM’a-f u
”(:‘I‘.'EL'.H‘:':|'.&‘\I>:;I if. $3? I ,'I 9
' '1: ,, 9- ..r.. 9 ~‘.|.,9,\r 9 1m.—
9-99.99 9 9 .
. --' 0 "-9 , 9..",..' |' \‘9‘. ‘ .::‘ .4 I...“ _-
9 ”"3: I?” .' ‘ |I I‘l . ”‘wa H ' ‘
u 9 “‘III" ' a! I. I' will. 9‘“ H "hulli : I. {I ‘ .
a- 9' I} "I ' “1.”! 'H' «1: "° " "‘I" ‘3 '
9. ' ‘ . " ,, '. " ""I'H I! I‘ll . .

 

9

Illlllllllllllllllllllllllllllllllllllll 7

mass 31293107720918 ' ӎ- v - .
iiflfiiififfia: £914,192
answer-say ;

m

This is to certify that the

thesis entitled

Consumer Response to Food Contamination:
A Case Study of Heptachlor Contamination of Oahu Milk

presented by

Mark Eugene Smith

 

has been accepted towards fulfillment
of the requirements for

 

Master of Science dcgmmin Agricultural Economics

 

% , ./fl17w @W
Major professor a

Eileen 0. Van Ravenswaay

 

Date JUlY 9, 1984

 

0-7639 MSUis an Affirmative Action/Equal Opportunity Institution

 

 

 

 

 

MSUJ— RETURNING MATERIALS:
Place in book drop to
LIBRARJES remove this checkout from

your record. FINE§ will
be charged if book is

 

 

9 r», x: ‘ 4 g" .« -. A. I" k, ‘ 1 .1
9 . g f M C l.' .H
l J.

,4 hl’

 

. 82

returned after the date
stamped below.

'——-—.———_.——._...—-' 'c _.

 

 

 

 

 

 

 

CONSUMER RESPONSE TO FOOD CONTAMINATION:
A CASE STUDY OF HEPTACHLOR CONTAMINATION

OF OAHU MILK

By

Mark Eugene Smith

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Agricultural Economics

1984

Copyright by
MARK ELGENE SMITH
1984

ABSTRACT

CONSUMER RESPONSE TO FOOD CONTAMINATION:
A CASE STUDY OF HEPTACHLOR CONTAMINATION
OF OAHU MILK
By

Mark Eugene Smith

The 1982 pesticide contamination of Oahu. Hawaii. milk with
heptachlor provided a rare opportunity to study consumer response to
food contamination. A major objective was to develop methodology to
estimate producer cost from lost sales due to consumer awareness of
contamination. Other objectives were to estimate sales loss to Oahu
dairymen. examine such incidents' characteristics. and draw pol icy
implications.

An econometric model considering consumer awareness of
contamination. contamination's extent. and possible consumption habit
changes was less restricted and provided better loss estimates than
previous models.

Policy implications include protection of long-term industry
interests through protection of short-term consumer interests by
recalling all contaminated product. Following contamination. even
favorable news may not increase sales; attempts to counter bad

publicity are not recommended. Significant producer incentive exists

Mark Eugene Smith

to avoid lost sales. but unless benefits or costs of individuals'
actions are fully captured. market forces alone will not guarantee food

safety.

To my parents. Eugene J. and Marie 8.. and my brothers. Paul A.
and Tony 8.. without whose help and support this would not have been

possible. Also to A.M.D.G.

ACKNOV LEDGMENTS

I thank many for their help with this thesis.

Special thanks go to my major professor. Eileen 0. Van Ravens-
waay. who was a source not only of ideas to explore. but of leads to
pursue. I am also grateful to committee members Stanley R. Thompson
and Daniel B. Suits for their invaluable econometric assistance.
Robert Gustafson and Lester V. Manderscheid provided further economet-
ric and statistical advice. For demystifying the intricacies of dairy
marketing. I am indebted to Glynn C. McBride. I appreciate Frederick
Fico's introduction to relevant controversies in the field of
journalism.

To thank all those who provided information and data would
require a chapter in itself. I am most.thankful to Hawaii Milk
Commissioner Robert Yara. Milk Analyst Denis Shimamoto. and former Milk
Commissioner Roy Matsuura for their patience answering numerous ques-
tions about the Oahu milk market and generously providing data. I am
also grateful to Foremost Dairies. Inc;. for providing information
under trying circumstances.

Also. I thank Merlinda Ingco for her advice. encouragement. and
support.

I appreciate the wherewithal provided by the Economic Research
Service of the U.S. Department of Agriculture and Michigan State Uni-

versity. which made this study possible.

TABLE OF CONTENTS

LIST OF TABLES C O O O O O O I O O O O O O O O O O O O O O O O O 0

LIST OF

Chapter
1.

2.

FIGURES O O O O O O O O 0

INTRODUCTION . . . . . . .

1.1 BaCKground O O O O O O O O O O O I O I I O 0 O O O 0
Organization of

l.2 Research Approach and
THE OAHU MILK MARKET . . .

2 1 Production . . . . .
2 2 Processing . . . . .
2 3 Government Regulation
2.4 Coordination . . . .
2 5 Marketing . . . . . .
2 6 Consumption . . . . .
27 Smmmy. ... ...

THE OAHU MILK CONTAMINATION INCIDENT . .

3.1 Chronology . . . . .

3. l. l The Contamination

2 The Recalls and Their Aftermath

3 Renewed Controversy Over
and Appropriate Action

4 Subsequent Contamination . .

5

6

.7

3 .3 Advertising Response
THEORY AND METHODOLOGY . .

4.1 Adjustments to Standard Theory

Assigning Responsibility
The Safeway Controversy
Beyond the Study Period
3.2 Legalities After the Incident .

Thesis

Duration of
Level

Exposure

4.2 Awareness of Contamination and Mediating Factors . .

Page
vi

vii

u—l

on

Milk Substitute

4.3 Specific Response to Negative Product Information
and Recalls . . . . . . . . . . . . . . . . . . .
4.4 Mass Media and Interpersonal Communication Effects
on Awareness . . . . . . . . . . . . . . . . . .
4.5 Consumer Response to Health Warnings and
Contamination Incidents . . . . . . . . . . . . .
4.5.1 Studies Without a Quantified Media Variable
4.5.2 Studies Using a Quantified Media Variable
4.6 Theoretical Statement of the Problem . . .
4.7 Methodology . . . . . . . . . . . . . . . .
5. DATA AND FINDINGS . . . . . . . . . . . . . . .
5.1 Data . . . . . . . . . . . . . . . . . . .
5.1.1 Consumption . . . . . . . . . . . .
5.1.2 Population . ....... . . . . .
5£L3 Prices .. . .. . .. . . . .
5.1.4 Income . . . . . . . . . . .
5.1.5 Advertising . . . . . . . .
5.1.6 Media Coverage . . . . . . .
5.2 Econometric Findings . . . . . . .
5.2.1 Determination of Appropriate
5.2.2 Model Estimation . . . . . . . . . . . . .
5.3 Estimates of Lost Sales . . . . . . . . . . .
5.3.1 Quantity Estimates of Lost Sales . . .
5.3.2 Estimation of the Value of Lost Sales
5.4 Media Findings . . . . . . . . . . . . . . .
5.4.1 Effect of All Media on Sales . . . .
5.4.2 Trends in Media Coverage . . . . . .
5.5 Demographic Findings . . . . . . . . . . .
6. SUMMARY AND IMPLICATIONS . . . . . . . . . . .
6.1 Methodology . . . . . . . . . . . . . . . .
6.2 Estimate of Losses . . . . . . . . . . . .
6.3 Characteristics of Massive Food Contaminati
Incidents . . . . . . . . . . . . . . . . . . .
6.4 Policy Implications . . . . . . . . . . . .
6.5 Further Research . . . . . . . . . . . . .
APPENDICES.......................
1. THE DATA . . . . . . . . . . . . . . . . . . . .
2. CONTENT ANALYSIS OF OAHU NEWSPAPER ARTICLES . .

REFERENCES .

Page

124
125
146
155

Table

2.1
5.1
5.2
5.3
5.4
5.5
A.1
A.2

A03

A.4

A.5

A.6

B.1

LIST OF TABLES

Forms of Oahu Milk Utilization . . . . . . . . . . . .
Estimated Elasticities of Retail Fluid Mi1k Demand . .
Comparison of Known and Estimated Sales Loss . . . . .
Monthly Sales Losses . . . . . . . . . . . . . . . . .
Effect of Publicity on Milk Purchasing . . . . . . . .
Whether Whole Milk Purchases Would Return to Normal .
Adjusted Consumption (Pounds). March 1982-June 1983 .

Consumption. Calendar Composition Adjustment Factors.
and Mi1k Dumped O O O O O O I I O O O O I O O I O 0

Milk Price. Fruit Nectar Price. and Consumer Price Index

Hawaii State Personal Income. Resident Population. and
Per Capita Personal Income . . . . . . . . . . . . .

Visitors to Oahu. Average Length of Stay. and Average
Daily Visitor Census . . . . . . . . ... . . . . . .

Oahu Resident Population and De Facto Population . . .

Components of Media Coverage . . . . . . . . . . . . .

vi

Page
17

85

96
104
106
130

131
134

137

140
143
154

Figure

2.1
4.1
4.2
5.1
5.2
5.3
5.4

LIST OF FIGURES

Oahu Milk Production and Utilization . . . . .
Areas of Loss . . . . . . . . . . . . . . . . .
Expected Response to Media Coverage . . . . . .
Per Capita Consumption: Actual and Estimated .
Per Capita Consumption: Projected and Estimated
Negative and Positive Media Coverage . . . . .

Components of Negative Media Coverage . . . . .

vii

Page
16
7O
74
89
91

101
103

CHAPTER ONE

INTRODUCTION

In March 1982. most milk on the island of Oahu. Hawaii. was
discovered to be contaminated with the pesticide heptachlor. a
suspected carcinogen. Since the mi1k market there is physically. and
by state law. legally. separated from any other major market. the
situation provides a natural laboratory for the economist to study the
response of consumers who were aware of contaminants in their milk
supply. One of the purposes of this study is to develop a methodology
to determine the loss to producers from lost sales due to consumer
awareness of food contamination. The methodology developed here is
applied in a case study of the Oahu incident. Characteristics of the
incident and government and industry response are examined to draw
public policy recommendations.

This research merits attention by several groups. Knowing the
magnitude of this cost would assist government regulators who weigh the
balance between increased monitoring or imposed fines to enforce food
safety regulations. .As government regulation comes under increasing
scrutiny. the question of government product inspection versus private
inspection by producers. especially of livestock and livestock prod—

ucts. is raised. Is the threat of lost sales because of consumer

awareness of contamination enough incentive to prevent the producer
from violating food safety regulations? Knowing the characteristics of
massive food contamination incidents would also help government offi-
cials devise strategies to handle them. to protect public health and
the industry involved.

Producers who are victims of such an incident will benefit
from this study. They often often cite a decline in sales. which is
attributed to consumer fears of the contamination. However. in law-
suits and testimony before government bodies. this additional cost has
not been quantified and so has not been recoverable from those respon-
sible for the contamination. Losses from product and feed dumped.
laboratory. veterinarian. and legal fees. and losses from death of
livestock are more easily quantified.

The effects of a food contamination incident are not limited to
producers of the contaminated product. Producers of the same. though
uncontaminated. product may suffer sales loss. Producers of substi-
tutes may benefit as consumers alter consumption temporarily or
longer-term consumption habits. Producers of vegetarian foods may
benefit from concern over additives and contamination of meat products.
Conversely. producers of complements of a contaminated product may
suffer reduced sales after a contamination announcement. Knowing the
economic consequences of contamination will help such producers.

Further. consumers will find this study of interest by under-
standing the characteristics of such incidents. 'They may then know

what to expect in the aftermath of such occurrences. Understanding the

environment in which government officials responsible for public health
operate. consumers may exert influence to achieve the desired gov-

ernment response.

1.1 BiQkQEQUDd

Consumer protection from adulterated food has been a federal
concern since the beginning of this century when the Pure Food Act
(1906) and Federal Meat Inspection Act (1907) were passed. As food-
production processes became more complex. so did federal regulation
(see Kessler. 1984). Maintenance of food safety involves protection
from mislabelling. poor packaging. additives (e.g.. preservatives.
dyes). residues (e.g.. animal antibiotics). and environmental contami-
nants. such as heavy metals. aflatoxin. pesticides. and other organic
compounds that are unintentionally present in the environment of people
or animals.

Most of the more prominent contamination occurrences involve
this latter group of substances. The earliest in recent history was
the "Great Cranberry Scare of 1959." when shortly before Thanksgiving
the Department of Health. Education. and Welfare announced that cran-
berries may have contained residues of amino triazole. a herbicide and
potential carcinogen. As a result. producers claimed sales were
depressed at a time of normally peak demand. In 1968. about 20 percent
of Montana's milk supply was contaminated with the pesticide heptachlor
after chlordane. a related chemical. was sprayed on alfalfa fed to
dairy cows. One dairyman noted:

The great furor of publicity that this [contamination] aroused

caused the public to be afraid to drink milk and caused a very
noticeable drop in our sales. It was months before the milk out-
lets returned to normal. and in many cases. it still is not.
(Boylan. 1969)

The largest LLS. food contamination incident occurred in 1973.
when polybrominated biphenyl (PBB). a fire retardant. was accidentally
mixed with feed in Michigan. Initially believed to be an isolated
occurrence. the statewide scope of the problem was slowly realized by
government officials. It is believed that about 90 percent of Michigan
residents at the time were exposed to P88 in their beef. pork. lamb.
eggs. and milk (Chen. 1979). The state spent about 5250.000 in a
campaign to restore Consumer confidence in Michigan agricultural prod-
ucts after Canada temporarily closed its border to Michigan meat prod-
ucts and consumers sought out-of-state foods. The Congressional Office
of Technology Assessment (1979) estimated the cost of condemned food
at $215 million. Eleven years after the occurrence. lawsuits are still
being resolved. and at least one dairyman believes milk sales are still
slightly depressed.

Since 1973. a succession of contamination incidents has cap-
tured public attention. In 1979. a damaged electrical transformer in a
Montana meat-packing plant leaked polychlorinated biphenyl (PCB). which
contaminated animal feed and food for human consumption. The incident
affected 18 states and two foreign countries at a cost of $11 million
to producers and state and federal agencies (USDA. FSQS. l980bL
Again. concern was voiced about a slump in sales of livestock products
(see Rede. 1979). In 1982. heptachlor contamination of Oahu milk was

discovered. leading to numerous recalls of milk and dairy products and

consumer doubts of island milk quality. {After reports of dioxin con-
tamination of some Great Lakes fish. sales to New York City. a major
market. dropped by about 80 percent. even of those species that were
uncontaminated (Peterson. 1983M

Most recently. concern over the fumigant ethylene dibromide
(EDB. not technically an environmental contaminant) caused confusion
in supermarkets. State and federal agencies scrambled to define safe
levels of the compound amid rumors that large amounts of grain would
have to be dumped. Sales of cake mixes. after their EDB concentrations
were publicized. were particularly affected. "Cake-mix sales have
dropped to zilch--peop1e aren't even going near that aisle" (Beck &
Hager.1984L

The environmental contamination problem of livestock is one of
relatively few large-scale incidents and many other small-scale inci-
dents for which there is little aggregated data. Van Ravenswaay and
Smith (forthcoming) estimated that nationally from 1974 through 1978.
about 1.000 cattle and 200 hog operations were annually affected by
environmental contamination. .Aggregate direct costs borne by such
producers were very small compared to the worth of the industry as a
whole. but individuals may have borne significant costs. and it is
unknown how minor incidents affected sales of meat products. Some fear
a shift away from meat consumption and a rise in more vegetarian
habits. Such incidents are often handled quietly to minimize consumer

concern over food safety.

The Office of Technology Assessment (1979) surveyed 50 states
to determine the cost of environmental contamination in food but did
not estimate reduced sales stemming from the incidents. Of 32
responses. only 18 reported at least one incident between 1968 and
1978. Only six estimated the cost of food condemned for a total of $67
million. excluding Michigan's PBB incident. Since many states were
unable to provide estimates. the authors concluded this figure may be
grossly underestimated.

Another question associated especially with the larger
incidents is that once they occur. government action may protract
consumer awareness of the problem and may worsen consumer attitudes
toward a contaminated product. Investigations (Hawaii Senate. 1983;
USDA. FSQS. 1980a) may reveal poor government performance in protecting
public health. and accusation of a cover-up raises questions of gov-

ernment's intent and desire to safeguard the consumer.

1.2 W

To estimate producer sales losses due to consumer awareness of
contamination. an estimate of what sales would have been in the absence
of the incident is compared with sales in light of the incident. Hence
an econometric model of Oahu fluid milk demand is estimated. An analy-
sis of government and industry response to the incident. based on a
review of more than 500 newspaper articles. government reports. and
numerous conversations with related industry and government officials
in Hawaii. California. and Washington. 0.0.. is included to supplement

the study of consumer response.

The next two chapters familiarize the reader with the Oahu milk
market and the impact on it of heptachlor contamination. Chapter Two
presents a study of the Oahu market. the first of its kind in almost 25
years. Chapter Three presents a chronology of the incident which
details consumer confusion and resistance to purchasing milk and milk
products after the contamination announcements. It also describes
actions by State officials and the Hawaiian and Californian milk
industries in response to the incident. Chapter Four develops the
theory and methodology of the study. A review of the literature of
consumer response to negative product information and to recalls and
health scares in particular is presented. The appropriate econometric
model is then developed. Chapter Five presents the findings of the
model. a study of media coverage of the incident. and demographic
analysis of consumer response to the incident. The last chapter
summarizes the study and discusses the economic consequences of food
contamination. including characteristics of massive food contamination
such as the Oahu heptachlor and Michigan PBB incidents. Public policy
implications are drawn with regard to minimization of public health
threats and industry losses after a food product has been contaminated
and to the roles of government and industry in enforcing food safety

regulations.

CHAPTER TWO

THE OAHU MILK MARKET

To build an econometric model of demand. an understanding of
the market under study is fundamental. This chapter builds upon
studies conducted in Hawaii in the 19505 through the early 19705. more
recent reports including those by the U.S. Justice Department. and

personal conversations with industry and government officials on Oahu.

2.1 Winn

Milk is the only major food in which Hawaii is self-sufficient.
Of the 25 herds in the State with ten or more cows. 16 operate on Oahu
to produce about 80 percent of the State's supply. A similar
percentage of the State's population resides on the island. In 1980.
Hawaii had the nation's largest average herd size of 520 cows per dairy
with mostly Holsteins in drylot operations (Koshi. 1980). Since at
least 1968. all production has been of Grade A quality. Average 1981
production per Hawaiian cow was 11.811 pounds. below the national
average of 12.147 pounds. but slightly higher than that of Florida and
most other southern and south-eastern states. In 1981. Oahu held
9.600 of the State's 12.700 cows. Monthly Oahu production

averaged about 9.893.140 pounds over the 1977-1981 period. Average

1982 output fell 8 percent to about 9.116.060 pounds and was the lowest
in six years (Hawaii ARS. 1983). Production varies monthly. usually
peaking in late spring and bottoming out in the fall. Since at least
the early 19605. production has been kept 10 percent above "normal
demand" to handle fluctuations (Mollett. 1961).

The cost of milk production in Hawaii is the highest in the
country (Cohen & Eisenstat. 1983). Chief problems as seen by the
industry relate to high feed costs and low reproductive rates
(Morrison. Kefford. 8 Harada. 1981). Without pastures on Oahu. all
feed must be purchased and much must be imported. Reliable feed
supplies are sometimes precarious since only one firm transports feed
from California and feed storage facilities are lacking. Imports of
prepared animal feed. mostly from the mainland. averaged 108.000 tons
per year from 1977 through 1981. Because of the heptachlor contamina-
tion incident. green chop (ground pineapple plants used for forage) is
no longer used requiring more imports.

Further adding to high costs is unionized dairy farm labor
earning an average of $7.30 per hour (Hawaii DPED. 1983). Wasteful
labor use was found on Oahu dairies in Mollett's 1961 study. but it is
unknown to what extent this has continued. Dairy replacement costs
also boost production costs. Replacement calves are shipped to the
Outer Islands to be raised. and then returned to Oahu.

Despite the problems they face. Oahu dairymen earn a relatively
favorable return on investment. The 1979 average return for 12 dairy

operations was 10:77 percent (Donoho. 1980). The 1979 prime interest

10

rate was 10.91 percent. Conscious of possible mainland competition.
the industry has pressed the State and University of Hawaii to
investigate means to reduce costs. At least one dairy cooperative was

trying to use a local grass hay by 1980 (Koshi. 1980L

2.2 messing

At the processor level. a duopoly exists on Oahu and apparently
has since at least 1958. 'The larger of the two. Meadow Gold Dairies-
Hawaii. is a subsidiary of Beatrice Foods and has operations on other
islands. It received 5959percent of milk produced on Oahu prior to
the contamination announcement.(Hawaii Senate. 1983L Foremost
Dairies. Hawaii. a subsidiary of Foremost-McKesson. handled 40.5 per-
cent of the pre-recall production (Hawaii Senate. 1983). In addition
to island milk supplies. Meadow Gold imports dry milk fat from the
mainland while Foremost imports it from New Zealand (Harpham. 4/20/83).
Processors supply: homogenized. 2 percent. 1 percent. skim and
flavored mi1k. buttermilk. half-and-half. ice cream. ice milk. ice milk
mixes. yogurt. cottage cheese. and sour cream. Both are heavily
involved in fruit juice processing also. Local production accounts for
all fresh fluid. and 75 percent of ice cream consumption (Morrison.
Kefford. & Harada. 1981L

A5 are costs of milk production. processing costs are higher
than those on the mainland. (Information for this discussion is mostly
from Harpham. 5/12/83J Labor. utilities. transportation. and
packaging costs are greater. Likewise. since the price processors pay

for raw milk is higher. the L5 percent shrinkage (the same experienced

11

on the mainland) costs more. In a pocket economy. processors cannot
achieve the economies of scale possible on the mainland. One Honolulu
plant with 60 employees processes 17.000-20.000 gallons daily. while a
mainland plant could employ 70 and process seven to nine times as much.
The same Honolulu plant could increase production four times to reduce
per gallon costs. but demand does not warrant increased production.

In early 1983. Oahu dairies and processors employed 574 people
(360 in processing/distribution and 214 on farms); total state dairy
industry employment was 720 workers. Directly and indirectly the
industry supported about 1.900 jobs or a little more than 044 percent

of total Hawaiian employment in 1981 (Hawaii DPED. 1983).

2.3 W

Beyond government involvement to ensure a safe milk supply. the
State regulates production. transportation. processing. storage.
distribution. and delivery of milk under the Milk Control Act. The
Division of Milk Control establishes quotas and minimum producer
prices. Similar to mainland markets. producers are paid a blend price
based on utilization. The 1967 Act was passed in an atmosphere of
violence and milk dumping. Producers felt they were not receiving a
fair price from the two processors who had not increased producer price
for 15 years (Lynch. 9/29/83). The two milk sheds of Honolulu and
Hawaii were created. (Operations are fully integrated on Maui and
Kauai; five producers sell to two processors on HawaiiJ Only with a

license granted by the Board of Agriculture may a firm import fluid

12

milk. Military commissaries are not subject to the Act though and may

import whatever they wish.

2.4 {Gannfiinitinn

The Division of Milk Control and two Oahu dairy cooperatives
help achieve horizontal coordination among producers. Raw milk supply
is regulated by quotas the Division establishes for each producer.
Five producers who account for about 40 percent of the island's quota
form the Oahu Dairy Cooperative. Eight others. producing 27 percent.
are members of the 50th State Dairy Farmers' Cooperative.

Mechanisms to improve vertical coordination are known to exist
at two levels. At the farm input level. most Oahu dairymen‘joined the
Green Feed Cooperative to harvest and distribute green chop. Green
chop is available from the two major pineapple companies which would
normally burn and clear the fields in preparation for the following
year's crop.

At the producer-processor level. the State market order helps
ensure dairymen a market. even though neither producer cooperative has
any processing capacity. Both cooperatives and the University of
Hawaii's herd supply both processors. Both processors themselves are
fully integrated. Based upon quota allocations. Meadow Gold's farm is
the second largest on Oahu. and accounts for 17 percent of island

production. Foremost Farms ranks third with 15 percent.

13

2.5 Marketing

The sizes of milk marketing channels are unknown. even to State
officials (Cohen. 1983). 1A beef marketing report (Garrod & Ching.
1982) is the most recent study showing marketing channels for any
Hawaiian agricultural commodity. In 1980. almost half the market sup-
plies were distributed through retail grocers. a third through restau-
rants and hotels. a tenth through the military. and the rest through
institutions and direct marketing. While beef marketing channels dif-
fer from those of milk. this gives some idea of the relative magnitude
of sales through each type of distribution. It is known that military-
related sales have accounted for 12 to 15 percent of fluid milk consum-
ption. and tourist-related sales have been about 10 to 12 percent over

the last five years.1

During the school year. about 20 percent of
fluid consumption is purchased by the Department of Education (calcu-
lated from Harpham. 9/25/82L

With a duopoly among processors. competitive behavior would not
be expected. Though perhaps tacit. collusive pricing would be expected
in recognition of their mutual interests. Likewise. advertising and
product strategy would be undertaken with consideration of possible
countermeasures by the other. From Wm: weekly
"Retail Food Price Guide" from January 1977 through December 1982.

there is no evidence of a fluid milk price war between the two.

Monthly average whole milk prices in the first half of 1983 were

 

lResults of a survey conducted by the author of individuals
knowledgeable of the Oahu dairy industry. 1983.

1h

generally lower than 1982 levels though. No generic advertising is
undertaken; milk is differentiated by brand advertising. The milk
companies are not considered maj or advertisers by some in the adver-
tising trade. Historically. advertising by one increased with the
introduction of a new product (dai ry-related or not). Before the
spring of 1983. the last such event occurred a little more than five
years ago. Based upon television advertising expenditure data from two
of Oahu's four commercial stations. Meadow Gold spent almost twice as
much as Foremost in 1981 and 1982. and more than seven times as much in
the first half of 19$. Also. following the heptachlor incident.
Meadow Gold changed its package design. While this change was part of
a national marketing strategy it did help Meadow Gold break with a bad
product image. In May 1983. Foremost packaged "Dai ryland." a new
private label milk for Foodland Super Markets. the largest grocery
chain in the state. That month. "Dairyland" sold for $1.59 per hal f-
gallon. or 5 percent less than the Foremost-Meadow Gold price of $1.67
(Harpham. 5/12/83). It thus appears some competitive behavior does
exist between processors.

An interesti ng aspect of milk marketing on Oahu is that the
processor truck drivers play a key role balancing quantities supplied
with quantities demanded. They remove expired milk or milk nearing its
expiration date and leave enough fresh milk to meet what they believe
the store will need. No grocery store executive has easily available

data on the amount of milk returned from his/her store.

15

Shipment of milk between islands is infrequent since each
island is basically self-sufficient. The last milk shipment from Oahu
to Hawaii was in 1980. Milk was shipped from other islands to Oahu
after the contamination announcement in March. April. May. June.
October. and December 1982. Approval for such shipments is granted by

the State's Milk Commission.

2.6 Consumption

Per capita Oahu milk consumption has been declining. In
1962. per capita annual fluid consumption averaged about 214 pounds.
but fell to 160 pounds by 1971 (Hogg. 1974L By 1981. consumption was
about 153 pounds per year per person or about 80 percent of that on the
mainland (Morrison. Kefford. & Harada. 1981). Ethnic origin influences
consumption; the Caucasian population consumes more than other racial
groups (Consumer Nutrition Center. 1981). In 1962. Scott (1967) found
that Caucasian adults and children drank milk more because they liked
it than for its nutritional value. while other groups (mainly Japanese.
Chinese. Hawaiian. and Filipino) consumed it more for its nutritive
value than taste. Caucasians account for 33 percent of the Oahu popu-
lation. followed by Japanese (25 percent). Filipino (13 percent). and
Hawaiian (11 percent) (U.S. Census. 1982).

Raw milk is utilized in several ways (Table 2.1). It is
assumed that Class I utilization approximates the quantity demanded of
fluid milk at least over the pre—contamination period. Seasonality is
evident in its demand (Figure 2.1). Fluid consumption falls with the

summer recess of school and rises with September school openings. In

-mnm_.>cm::mw .co_umN__wu:.ocm comuozooca x__E osmous._.m oL:m_u

mam—

Nmm—

.mm_

.mmm_ 6:33

0mm. mum. mum.

Km—

111

23. uwo 23.. uma 23—. uma 2:5 umo 23.. one za. own 2:... 25.
. L b b L _ p _ h P _ _

l6

 

... 2. (N... m a... (a. ,. .72.). .\/ N. ... .3. a
X . . . . . .~ ’ k . s . . > m x
. or . IS a \ a . ~ . e N ,\ s s
o a s S s I ’ o \
K .. R C ’k :~ . . w ...
m .. ... 1. z.\ a . . ... N
‘ .R u .. a N

 

   

‘
_‘-O-----".‘

”$212--.._.

 

 

 

[-

mzowmuuomm 2. mu~¢w>3uo emu—SSE ...... T v

22.33.22. zuzho 1-1
cum—So x4; owhczutchzou ...---
zozcfiazq. a mmcqullu

to—

 

 

Nu
mac—.3;

0.932003

17

the past. troop movements in and out of the state significantly

affected consumption. but it is unknown how important a factor this is
now; In general. Class I-A and Class II utilization rises in the summer.
only to decline (often to zero) when schools open. However. summer
Class I-A and Class II utilization does not always rise enough to
counter the fall in fluid consumption. "Traditionally. the Honolulu
Milk Shed experiences its greatest loss through dumpage during the

summer months" (cited in Cohen 8. Eisenstat. 1983).

Table 25L--Forms of Oahu milk utilization.

 

Utilization Description

 

Class I Used for fresh fluid consumption (e.g..
whole. 2%. and other fluid milk).

Class I-A Skim milk available for use in filled
(imitation) milk.

Class II Soft manufactured dairy products (e.g..
ice cream. yogurt).

Export Fresh fluid milk exported to Neighbor
Islands or for use on airlines and ships.

Salvaged Raw milk from which the cream is skimmed
and used in some Class II products. The
skim milk is then discarded.

Dumped Raw milk disposed of due to excessive
heptachlor residues.

 

There are several fresh milk substitutes available on Oahu.

The milk processors produce filled. or imitation milk. and recombined

18

milk. The same year the Milk Control Act was passed. processors intro-
duced filled. or imitation milk. This is made "by adding vegetable oil
in place of butterfat to either fresh skim milk or reconstituted skim
milk" (Hawaii Senate. 1983). Its market share had fallen from 20
percent in 1967 to 7 percent in 1981 (Renaud. 1971; Morison. Kefford. 8.
Harada. 1981). Filled milk prices are less than that of fresh milk.

In July 1983 (latest month for which data are available). the filled
mi1k price was $1.39 per half gallon compared to $1.68 for whole milk.
Recombined milk "is a product which results from the combination of
nonfat dry milk. dry cream. [and] milkfat with potable water" (cited in
Hawaii Senate. 1983). Ultra-high temperature (UHT) milk was allowed to
enter the Hawaiian milk market with licenses granted in fall 1983 to
Dairymen. Incorporated. and Real Fresh. Incorporated. a California
firm. Other milk substitutes include powdered and nonfat dry milk.
neither of which is produced locally. Dried milk and cream imports
from 1977 through 1981 averaged 647 tons per year. ranging from 611
tons in 1981 to 750 tons in 1977 (U.S. Army Corps of Engineers. 1977-

1981).

2.7 Summary
The Oahu milk market is high-cost. imperfectly competitive. and

confronts the seasonal demand of consumers. many of whom do not drink
milk for taste alone. Government regulation limits the entry of new
dairymen as does the duopoly at the processing level. Further. fresh

fluid milk supplies from sources beyond the State were. until very

19

recently. limited by logistical problems of shipping a perishable
product over great distances (2.500 miles from California) and by the
State's Milk Control Act. However. such protection may have been
necessary to preserve the islands' milk sel f-sufficiency. and even
before the contamination incident. Oahu dairymen were concerned about
possible competition from outside sources. Per capita milk consumption
is below the US. average. reflecting the racial differences in tastes
for milk. The insular nature of the Oahu milk market makes it a prime

market to study consumer response to food contamination.

CHAPTER THREE

THE OAHU MILK CONTAMINATION INCIDENT

The following details most events surrounding the heptachlor
contamination of Oahu milk and its consequences. It focuses on the
origin of the contamination. and consumer. government. and industry
reaction from the initial discovery in January 1982 through June 1983.

Specific information on the culpability of individuals involved is

available in the Wanna:
W (1983). This chronology is com-

piled from that report. over 500 newspaper articles and letters to the
editor in the two major Oahu newspapers. the Honolulu Advertiser and
‘flgnglu1u_§tar:fln11et1n. numerous other articles. and conversations with

many individuals in Hawaii and elsewhere.

3.1 Qnmmlogx
3.1.1.1he_92ntamination
The story of milk contamination began in Oahu's pineapple
fields. Pineapples suffer from mealybug wilt. named after the vectors
which feed on the crop. Ants protect the insects from predators
because ants consume the mealybugs'"honeydew" secretion. It was

found that by eliminating ants. mealybug wilt could be controlled. To

20

21

do this. DDT was first used. followed by mirex. followed by heptachlor.
all organochlorines. and hence extremely persistent in the environment.

In the 19605 the U.S. Food and Drug Administration (FDA) set a
zero tolerance for heptachlor in food. An action level (the level of
residue which will prompt FDA action on a product lot) of 0.3 parts
per million (ppm) on a fat basis was established since no smaller
quantities could then be detected. In laboratory animal tests. hepta-
chlor has caused liver and kidney damage and is a suspected human
carcinogen. It is estimated that about 90 percent of Americans carry
heptachlor residues in their bodies. The EPA banned heptachlor use in
1978. but granted an exemption to the Hawaiian pineapple industry. A
stipulation was that growers had to wait one year after the last appli-
cation before a field could be harvested for animal feed.

Pineapple plants had been harvested as "green chop." a cheap.
local substitute for imported dairy forage since 1958. With help from
the Department of Agricultural Engineering at the University of Hawaii
(UH). harvesting became more efficient in the late 19705 and included
the base of the plant. where heptachlor apparently accumulated. It was
later found that green chop samples from April 1981 contained high
heptachlor levels. Being fed to dairy cows. the pesticide was excreted
in the milk. It is now believed the Oahu milk supply contained hepta-
chlor as early as October 1980. Violative levels were present in milk
bottled in April and May 1981. and high but nonviolative levels were
found in ice cream made that May; However. milk bottled in July and

August was found to be below the action level. 'The State Department of

22

Health (DOH) semi-annually checked for pesticide contamination in
milk. and no Violative levels of heptachlor were detected in the July
1981 tests.

On January 22. 1982. DOH Food and Drug Branch Chief Karl
Tomomitsu received test results from a sample of homogenized milk and
samples of raw milk from three dairy farms. all violating the action
level. These samples were taken January 6. Other samples taken
January 13 showed no detectable heptachlor residues. Upon recommenda-
tion from the lab. Tomomitsu sent the older samples to the FDA lab in
California. Three days later. the lab sent the samples with no note of
urgency attached. 'Tomomitsu did not develop an action plan in case the
original results were confirmed. and did not inform his superiors. By
January 22. all milk produced January 6 and used for fluid consumption
had either been sold or had expired. but Class II products could have
still been for sale.

Thirty-five days later on March 1. confirmation by the FDA was
received. Tomomitsu initiated an investigation to detect the source
and extent of the contamination. assisted by Milk Commissioner Roy
Matsuura. Neither informed their superiors. Eight farms. including
the three previously tested. and both processors were sampled March 9.
Two days later. Tomomitsu informed his superiors of the contamination.
including DOH Deputy Director for Environmental Health Melvin Koizumi.
Dean Noel Kefford of the UH College of Tropical Agriculture was
consulted after the UH lab confirmed the contamination. Kefford wrote

Commissioner Matsuura:

23

. . . Continued consumption of milk with the reported heptachlor
epoxide residues would not appear to constitute an unreasonable
hazard to the general public. even those judged to be most
sensitive. More serious would be an announcement of this technical
violation of a tolerance . . . and subsequent prohibition of the
sale of milk from dairies and processors. No amount of explanation
of the technical nature of the violation would expiate the damage
done to the reputation of milk as a wholesome food and coinci-
dentally to the dairy industry . . . [erecipitate action which
results in a perception by the public that the milk supply contains
hazardous materials should be avoided. (Hawaii Senate. 1983)

On March 15. test results indicated both processors and seven
out of the eight farms sampled were in violation of the action level.
The next day. Deputy Director Koizumi informed DOH Director George
Yuen. an engineer. who ordered more tests to determine whether
contamination levels were falling. A day later. the same eight farms
were retested. By Thursday. March 18. the media learned of the
contamination and a report appeared about possible State action in the
afternoon paper. At 4:30 mew before the latest sampling results were

known. the DOH ordered the first of eight mandatory recalls.

3.1.2 W

Oahu residents read across their next morning's newspaper.
"Toss out that milk in fridge--it contains pesticide poison." Pulled
from shelves were homogenized. 2 percent. 1 percent acidophilus (for
those allergic to milk). and half-and-half milk. Since heptachlor
concentrates in the butterfat. skim milk was not pulled and neither
were those products processors said were made with imported butterfat
(ice cream and other Class II productsk Imitation milk was declared
safe. Foremost did not announce that it also pulled buttermilk and

sour cream it believed might have been contaminated. Health Director

24

Yuen believed this was an isolated incident and said the public was not
endangered even though heptachlor levels three to six times the action
level were found. Supplies on grocer shelves that day were deemed
safe. When asked about the delay in pulling milk. Yuen answered. "It
took time to determine the exact course of action to take" (in Hastings
& Harpham. 3/19/82L The Department of Education switched from serving
fresh to imitation milk in schools.

Both processors offered consumer refunds as about 180.000
gallons of milk were dumped. Meadow Gold President Robert Milne
encouraged the public to call his company with any questions.

Dr.1James Koshi. former dairy specialist with the UH and general man-
ager of the 50th State Dairy Farmers' Cooperative. assured the public.
"We're out to produce the best product" (in Hastings & Harpham.
3/19/82).

Within 36 hours of the recall announcement. acceptable milk
supply was only more than half of normal supply. Both Meadow Gold and
Foremost cut half-and-half production. with Meadow Gold supplying nor-
mal levels of whole and 2 percent milk. Meadow Gold's estimated loss
from dumped milk and milk refunds was 5250.000 (Harphanu 3/20/82).
Foremost. whose supply was more drastically affected. increased produc-
tion of imitation milk. Governor George Ariyoshi and health officials
assured that the public was not endangered. and most. but not all.
medical experts who were quoted agreed. Lack of consensus of hepta-

chlor's health effects was constant throughout the incident. Criticism

25

of the DOH's delay began to be voiced. as well as dairymen's anger that
they had not been informed of the problem.

Supermarkets reported heavy returns and initially brisk sales
of powdered and canned milk before consumers learned that imitation
and skim milk. in plentiful supply. were safe. Neighbor Island milk
sales were down. even though contamination was limited to Oahu. Calls
to hospital emergency rooms increased. Meadow Gold was flooded by
telephone inquiries but continued to invite such calls. saying "We feel
people have the right to know" (in Harpham. 3/20/82).

Meadow Gold dumped so much milk down the storm drain and not
into the sewer system as directed by the DOH that the channel into
which it emptied became de-oxygenated. Tens of thousands of fish died
and attracted newspaper photographers. Dairy cow slaughter was halted
until pesticide tests were performed. which later showed heptachlor
levels above the action level.

Headlines on Tuesday. March 23. reported that two of the three
acceptable dairies supplying Oahu were actually in violation of the
action level. Hence. it was possible unacceptable milk had been
consumed over the weekend. It was later learned that Meadow Gold knew
some of its supplies were probably contaminated. Foremost's only fresh
milk supply was imported from Hawaii and Maui. No milk was recalled.
but one supermarket voluntarily pulled milk from its shelves. Some
stores had posted signs saying the milk was safe. and several store
managers were now confused about milk safety. 'The DOH allowed the

processors to use contaminated milk to make skim milk. and most milk

26

available was skim. imitation. and recombined. Consumer refunds
continued. Senate Health Committee Chairman Benjamin Cayetano
heightened criticism of Health Director Yuen. just stopping short of
calling for his resignation.

Less than a week after the first recall. the DOH ordered Meadow
Gold 2 percent milk pulled as a precaution for Violative heptachlor
levels. Meadow Gold offered a refund. Some stores reported milk
stock-outs by closing time. but many others reported no milk. even skim
and imitation. was selling. Sales of condensed milk were up. and runs
on powdered milk were reported. Comments by store managers reflected
the situation: "Consumers are confused--they just don't know what's
safe now" (in Harpham 8 Hastings. 3/24/82). "People . . . have kind of
lost confidence and think no one really knows what's happening. So
they're staying away until things are sorted out. . . . Recent events
have set back the dairy industry 10 years" (in Kakesako. Games. 8
Morita. 3/25/82). "The state really blew it" (in Watanabe 8 Morita.
3/24/82).

'Two small dairies were cleared. but 95 percent of Oahu cows
were contaminated. Chairman of the UH Department of Animal Science
Richard Stanley stated that one bad batch of green chop fed for two to
three weeks in December was responsible for the incident. It was a
"one-shot incident” (in Harpham 8 Hastings. 3/24/82). Meadow Gold
joined Foremost by importing milk from other islands; total March
imports were 8.250 gallons. ‘The DOH announced that test results were

needed before a recall could be ordered. a policy not followed for the

27

first recall though. The health director said the DOH lacked adequate
lab facilities. but that it was trying to increase the frequency of
testing and improve internal communications.

A week after the first recall. Gov. Ariyoshi. Health Director
Yuen. and UH officials again tried to reassure the public that milk was
safe. the recalls were precautionary. and that there was "no evil force
at work" (in Kakesako. Comes. 8 Morita. 3/25/82). Dean Kefford
explained that the action level is a regulatory level and does not
necessarily indicate a health threat if violated. But Sen. Cayetano
began publicly questioning the use of raw contaminated milk to make
skim milk and asked the Attorney General to investigate its legality.

Ten days after the first recall. Foremost milk distributed in
Waikiki was pulled for heptachlor levels above the action level. 'The
milk came from a dairy previously cleared by the DOH. and Foremost
President Paul Heckenlively declared. "If they [the DOH] dontt get their
act together. the milk industry will be destroyed" (in Ong. 3/28/82).
Dairymen also blamed the confusing DOH test results for causing
confusion in the marketplace. The State agency announced it would try
a new sampling procedure "because some of what [the DOH has] gotten up
to now has not been representative" (in Harpham. 3/30/82).

March went out like a lion. The fourth recall in less than two
weeks was announced. together with Health Director Yuen's immediate
effective resignation and the appointment of his successor. The DOH
pulled Meadow Gold cultured buttermilk. and the Attorney General began

investigating whether the company knowingly used milk of an unknown.

28

and so possibly contaminated. origin. Again. Gov. Ariyoshi declared
the safety of milk. but processors had more milk than could be sold.
Supplies were low. but demand was apparently less. Ample supplies of
imitation and recombined milk were available. Yuen cited personal
reasons. and not the milk crisis. for leaving. He partly blamed the
dairy industry for the crisis and accused the media of accentuating the
negative. Sen. Cayetano. unsure if milk on the shelves was safe.
claimed Yuen's resignation was the first step toward restoring public
confidence. That confidence had been shattered. and former Superin-
tendent of Education Charles Clark. with no public health background
but a reputation as an efficient administrator. was chosen by the
Governor to succeed Yuen.

April began with much media attention given to the poor State
handling of the situation. Sen. Cayetano criticized the Governor for
failing to take action and blasted the DOH for slow action. which he
claimed eroded public confidence in government. He accused the DOH and
dairy industry of putting economic considerations before protection of
public health. The Senator was also appointed Chairman of the Senate
Special Committee to investigate the incident. Charles Clark was
endorsed by the H9n91n1u_Ad1eLtiser. which editorialized. "Many people
will be understandably skeptical about the milk situation for a while
. . . despite assurances from some in government. including Yuen and
medical people that milk was safe to drink all along" 04/1/82). Also
in early April. health department tests cleared yogurt. cottage cheese.

and Meadow Gold ice cream. even though the ice cream was unexpectedly

29

found to have heptachlor at the action level. The DOH had previously
believed only mainland butterfat was used in its production. so no
heptachlor should have been present. No recall was ordered since the
action level had not been violated.

Milk sales remained sluggish. Chairman Jack Suwa of the Board
of Agriculture said. "[Milk] is on the shelf and it is up to the milk
industry to convince [people] it is safe. . . . This whole thing (the
milk crisis) wasn't handled according to the consumer angle and that is
why we have this fuss" (in Harpham 8 Burris. 4/2/82). Safeway offered
to import milk. but did not press the matter given ample milk supplies.

On April 5. recalls extended beyond fluid products to other
dairy products. Meadow Gold voluntarily recalled some of its ice cream
for suspected contamination. Meanwhile. the UH Pesticide Hazard
Assessment Project tested mothers'lnilk from six women and found all
six with heptachlor levels four to ten times those found two years
earlier. One sample was above the action level.

The next day. Acting Health Director Clark introduced an
improved sampling procedure by saying. "We think we have made our last
recall" (in Harpham. 4/6/82). With the cooperation of the agriculture
department. both processors. and the UH. milk would be tested three
times: upon delivery to processors. in their holding tanks. and in
finished products. FDA experts were requested to help improve DOH
testing. Meadow Gold and Foremost would hold products until cleared.
Clark praised both processors for their cooperation. and the Attorney

General cleared Meadow Gold of intentionally using possibly

30

contaminated milk. Human error on an employee's part was blamed.
Skinuning contaminated raw milk was still of questionable legality. but
it seemed the situation was finally getting under control.

Clarkfls blitz to restore public confidence was undercut the
next day. when Meadow Gold voluntarily pulled yogurt. (Yogurt had
previously been cleared) Worse. the yogurt sample in which excess
heptachlor was detected was not an "official" sample but part of a DOH
employee's lunch who asked that it be tested "for the hell of it" (in
Harpham 8 Hastings. 4/8/82L (Mark admitted that the DOH had to trust
processors to ensure all contaminated products were pulled.

Low sales figures. ample supplies of state-cleared milk at the
processing plants and in the stores and rising sales of competing
products such as milk powders or imitation milk already make it
abundantly clear that Oahu consumers don’t believe state assurances
that all whole milk released to market is wholesome and safe and
cleared of heptachlor. (Lynch. 4/8/82)

Sen. Cayetano criticized the DOH for considering the economic
future of the dairy industry instead of public health. Meadow Gold was
blamed for the last recalls because its officials told DOH that only
mainland cream was used in the recalled products when actually some
Oahu cream was used. The Senate investigating committee later learned
that Meadow Gold quietly pulled ice cream other than that previously
recalled.

A telephone poll of consumers conducted for Foremost Dairies
showed wholermilk consumption was down 73 percent. 2 percent down 75

percent. and imitation milk sales up 169 percent. About 16 percent

said if the crisis ended they would buy a lot less milk. 8 percent said

31

a little less. and 70 percent said they would resume normal consump-
tion.

Mid-April brought the fifth DOH recall. this time of some
Meadow Gold low-fat cottage cheese with heptachlor residues more than
twice the action level. The company's production practices and ethics
were questioned. and the company voluntarily pulled all its cottage
cheese. Cl ark again blamed the processor for misleading the State into
believing no raw Oahu milk was used in some Class II products. Both
Clark and Cayetano called for a second investigation of Meadow Gold.
which the company president welcomed. Cayetano claimed the DOH had
"put misplaced trust in Meadow Gold" (in Harpham 8 Hastings. 4/15/82).
He again questioned the legality of making skim milk from contaminated
raw milk. To avoid further recalls. Clark ordered the processors to
reject all Violative milk and not use it for skim. Meadow Gold com-
plained about the order. and in his response. Clark questioned why the
company had not conducted independent lab tests as had Foremost. Where
the DOH had previously concentrated its testing on fluid milk. it
expanded its tests to Class II products. including some from the main-
land. Again Clark assured that no violative product would reach the
shelves but had to qualify that by adding. "that is. assuming my orders
are followed" (in Kakesako. 4/14/82).

Meadow Gold officials claimed the recalled cottage cheese was
made from the same mix used in a batch previously cleared of contamina-
tion. Clark expressed confusion over how a product made from accept-

able milk could contain unacceptable heptachlor levels. The ”skim milk

32

syndrome" received attention as a possible explanation. Research found
that by skimming contaminated milk to remove heptachlor in the butter-
fat. the fat remaining had a higher heptachlor concentration than the
fat removed. The FDA suggested the EPA re-evaluate the heptachlor
action level.

On April 19. the FDA banned Meadow Gold milk from interstate
travel because samples taken three and five days earlier showed
unacceptable heptachlor levels. Clark requested an expanded Attorney
General probe because supposedly the milk had been tested before
delivery. The next day. despite test results showing acceptable
heptachlor levels. Clark recalled all Meadow Gold products made from
Oahu milk and banned its further use of Oahu milk. The armed forces
likewise halted purchases of Meadow Gold's fresh and imitation milk.
Foremost. whose profile during the whole incident had been low.
accepted milk usually sold to Meadow Gold but was careful to avoid a
recall. It held the milk until testing was completed. resulting in a
fresh milk shortage for a few days. Meadow Gold responded by revealing
its offer to allow DOH personnel to monitor the plant 24 hours a day.
but that the DOH declined. The company indicated it would sue.the
State and seek a temporary restraining order to block State action.

When asked if Foremost could meet the increased demand for its
milk. its new president. Donald Bender. said his production could
increase. but Foremost did not have a large enough market.

We're at a point where consumers are so confused and concerned that
they're avoiding anything in the dairy case except for certain

products. . . . We've enjoyed a tremendous increase in imitation
milk and juice sales have just skyrocketed.(in Watanabe. 4/22/82)

33

Two days after its ban against Meadow Gold. the DOH allowed the
company to process Oahu milk. but there were implications Meadow Gold's
threatened lawsuit may have influenced the decision. The processor
agreed to improve its tracing and testing procedures to lessen the
chances of viol ative products reaching store shelves. A week after the
ban. Oahu-supplied Meadow Gold products were again available. In
attempts to restore sales. the company announced the expanded testing
measures. a consumer-information program. and plans to import about 500
cows from the mainland and other islands.

Because of public concern about Oahu-produced milk. Foremost
shipped in milk from Hawaii. Total April imports by both companies
were 12.733 gallons. Skim and 2 percent milk were not being produced
on Oahu.

The pace of events slowed in May. The Senate investigation
continued. and for one week. the Navy flew in five milk shipments for
its commissaries. Meadow Gold flew in cows from California with much
publicity and dropped its milk prices for about two weeks. Losses to
dairymen from dumped milk exceeded 3600.000 ("Dairy Farmers' Loss."

5/ 14/82). and the Governor declared an economic emergency so they could
receive aid.

By mid-month. processors claimed public concern was still
depressing sales. Before the Senate investigating committee. Health
Director Cl ark claimed Oahu milk was safe to drink and explained pre-
cautions being taken to ensure milk safety. He admitted that the DOH.

Department of Agriculture. the UH. and the FDA were unprepared to deal

34

with the crisis. "There was a breakdown." he said (in Harpham.
5/21/82). Deputy Health Director Koizumi said. "What we were not
prepared for was the need to get information out to the public. . . .
We were hesitant. we didn't know how to handle it. Had we been more
aggressive in getting information out there would have been fewer
problems" (in Harpham. 5/21/82). From the hearings. it seems Foremost
took more precautions than Meadow Gold. The situation for Meadow Gold
improved when the FDA ended the ban on interstate shipments of its
products. and the military resumed purchases. Dairymen expressed con-
cern that consumption habits may have changed due to the crisis.

By month's end. seven of the island's l6 dairies were cleared.
skim milk returned to store shelves. public confidence was slowly
returning. but imitation milk sales were still up. Supply and demand
were about balancing out. but the periods of peak demand and supply did
not always coincide. Some spot shortages were reported. The discovery
of contaminated green chop as early as April 1981 initially attracted
little attention but was to later develop into a new phase of the
controversy--determination of the duration of public exposure to hepta-
chlor.

Early June brought another recall. this time a federal recall
of L5 million cans of tuna packed in Honolulu for improper canning.

By mid-June. nine out of 16 dairies were cleared. and while sales were
lower than usual. the milk commissioner felt consumer confidence was

returning. Attention began shifting to the Department of Agriculture

35

and the supply situation rather than the health department and milk
safety.

Another farm was cleared in July. and supply exceeded consump-
tion. The milk commissioner anticipated financial problems for Oahu
dairymen in August or September if sales did not improve.

As the Senate investigating committee held hearings. more about
the crisis was revealed. Cayetano accused Meadow Gold of deliberate
use of contaminated cream and other illegalities. The company harshly
criticized the tactics used in the committee's investigation. Foremost
was praised for voluntarily pulling products and halting the skimming
of unacceptable milk. Both processors claimed the DOH advised them

such skimming was permissible.

3.1.3 Won
W
W

The controversy entered a new phase in late July. Guberna-
torial candidates tried to use the incident to their political
advantage as evidence of the incumbentfls poor leadership. Lt. Governor
Jean King disclosed a UH report indicating unacceptable heptachlor
levels in milk as early as April 1981. which she claimed Health
Director Clark tried to cover up. and which the Attorney General kept
from Sen. Cayetano's investigation. Clark denied such an attempt. but
length of exposure became an issue since the DOH originally believed
milk was contaminated only four months before the first recall.

In the first half of August. milk sales began to rise. and

newspapers reported heavily on the length of exposure and possible

36

lowering of the action level. The political aspects of the issue
became less salient as questions of milk safety were again raised. ‘The
EPA and an independent lab confirmed the UH finding that Oahu milk
contained unacceptable heptachlor levels in April 1981. Originally.
officials believed the exposure was only short-term. and ill effects
were downplayed. Now. the serious effects and cumulative build-up of
heptachlor in the body were reported. With the finding that Oahu
consumers. especially children. were exposed for about a year. sub-
chronic liver damage was considered a real possibility. and the EPA
considered lowering the action level.

The findings of longer contamination were an embarrassment to
the health department. which detected no violations in its July 1981
tests. Clark attacked the findings and defended the DOH lab. generat-
ing more publicity about the contamination. Five months earlier. a
consumer had informed Sen. Cayetano that she had found some frozen milk
bottled in July and August of 1981 in the bottom of her freezer. The
Senator referred her to the health department. but only in late July
did the DOH test the milk and find no Violative heptachlor residues.
Confusion resulted since milk was apparently contaminated in April 1981
and January 1982. but not in July 1981. To add to this. the FDA
confirmed that acceptable heptachlor levels were present in the July
milk. though three times higher than the DOH found. It was later
discovered that the DOH knew in April 1982 that Meadow Gold ice cream

made in May 1981 had heptachlor residues very close to the action

37

level. Cayetano criticized the department for not disclosing such
findings to help determine the length of exposure.

As EPA officials visited Oahu in mid-August. the question of
the appropriate action level was further highlighted. .After meeting
with Clark. EPA Director of Pesticide Programs Edwin Johnson tried to
calm public fears about heptachlor. though he still expressed concern
for infants. .Johnson acknowledged that determination of the action
level would have to consider the economic repercussion on the dairy
industry. The DOH said it would follow the action level set at the
federal level. To resolve the question of contamination in milk
bottled in April 1981. the EPA submitted its sample to another inde-
pendent lab. which later confirmed the contamination.

In the meantime. 12 dairies met the action level of 0.3 ppm.
but sales had not reached pre-recall levels. Dr. James Koshi: "The
public has yet to be convinced that the milk is now safe to drink"

(in Watanabe. 8/2/82). ‘To help restore consumer confidence. Meadow
Gold. now under Raymond Jarman. established a new milk hotline. a
speakers bureau. and an informational brochure. but refused to disclose
to the Senate committee how its ice cream was made.

On August 17. after releasing findings that Oahu infant livers
were unaffected by heptachlor. Clark officially declared the end of the
heptachlor crisis. He again assured the public that milk was safe.
that monthly monitoring of milk for pesticides would continue. and that
95 percent of milk products contained less than half the permissible

level of heptachlor. Cayetano disagreed with Clarkds decision to close

38

the issue. Some health experts recommended that consumers still refrain
from milk consumption.

As August closed. attention began to shift to possible federal
aid for the dairymen and the political implications of the incident.
The W rel eased results of its poll taken mid-month
when the EPA-DOH dispute was publicized and the appropriate action level
was debated. Fifty-four percent of those sampled felt "pretty sure"
that milk was safe. but 40 percent were not sure. Thirty-five percent
said they were drinking less milk; mi1k sales were down by 20 percent.

Schools opened in September serving fresh 2 percent milk. as
usual. Papers reported the anticipated EPA recommendation to lower the
action level and its effect on island milk supplies. No shortages were
expected since sales were still below normal and acceptable milk was
being dumped. All but one dairy were cleared. and both processors were
careful to accept milk sufficiently below the action level to minimize
the possibility of a recall. As a result of the Department of Agricul-
ture's pesticide use/misuse investigation. the State cited Del Monte
and issued a warning to Dole for premature harvest of fields sprayed
with the pesticide.

September 10 brought EPA's recommendation to the FDA to lower
the action level to 0.1 ppm. Less than 1 percent of mainland milk
would have been unacceptable under the new standard. which provided
adequate protection for persons older than four months. An action
level of 0.05 ppm would provide adequate protection for all persons.

but about 2 percent of mainland milk would exceed the level. Health

39

Director Clark said he would act after the FDA accepted the recommenda-
tion but that no recall would be issued. It was unclear whether com-
mingling of acceptable and unacceptable milk would be allowed; it was
not in April 1982. One editorial noted that commingling could increase
doubts about milk safety. 'The EPA also revealed that low-level con-
tamination of Oahu milk was likely from October 1980. or 18 months
before the first recall. The possibility of subchronic liver damage
again received attention.

Around mid-month. another failure by the DOH to protect the
public was disclosed. A few months earlier. Violative levels of the
pesticide endosulfan were found in Oahu watercress after the crop had
been sold. The DOH gave the public no notice of its findings.

On September 21. the FDA lowered the action level to 0.1 ppm.
claiming the length not the amount of heptachlor exposure was the
problem on Oahu. Both processors acted immediately. and the DOH lowered
its action level the next day upon receipt of the FDA's official
notice. Milk sales were still below supply; milk from two dairies was
not being processed because of inadequate demand. The DOH stepped up
testing.

No commingling was allowed. and this apparently surprised the
processors who had earlier predicted no shortages. By September 24.
Foremost was unable to meet demand. and by September 27. with some of
its milk diverted to Foremost. Meadow Gold too had trouble supplying
adequate amounts of milk. Schools were supplied with imitation mi1k.

Clark quietly acknowledged that milk was contaminated in April 1981.

40

after the EPA tests showing such contamination were confirmed. By
September's end. more watercress was seized for endosulfan contamina-
tion. and a well was closed for contamination by dibromochloropropane
(DBCP).

Principal October news events included a recall of Tylenol in
Hawaii. and the Senate investigation hearings. 'The State claimed the
pineapple companies allowed premature harvest of their fields. It was
announced later that even four years after application. heptachlor
residues could be detected in green chop. Most damaging to Foremostus
reputation was the revelation that on March 25. it did not recall
possibly contaminated yogurt. but quietly sent salesmen to all outlets
to buy it back. Sen. Cayetano criticized the company for not publicly
recalling the product. Citing the attitude taken to the discovery of
cyanide in Tylenol. Cayetano said. "What we hope to get is that kind of
attitude to any contamination" (in Harpham. 10/6/82). The press again
reported the Senator's criticism of the DOH for its lack of promptness
and good management in handling the situation and that economic consid-
erations were of greater concern to it than the public interest.

No related items were reported in the November newspapers.
After importing about 1.800 gallons of milk in October. none was
imported in November. The milk crisis was not a sufficient embar-
rassment to Gov. Ariyoshi to prevent his re-election. Shortly before
Thanksgiving. Hurricane Iwa struck and Oahu suffered substantial

damage.

41

3.1.4anm

On December 8. antibiotics were detected in unacceptable
amounts in Meadow Gold milk. About 6.000 gallons of skim and
recombined milk were "voluntarily" recalled before the DOH ordered it.
No further shortage resulted as safe milk supplies were quickly
restored. The next day. Meadow Gold whole and 2 percent milk were
found to contain excess antibiotics but had already been pulled from
the shelves. The DOH investigated the incident. especially trying to
determine how recombined milk. supposedly not made with any Oahu
ingredients. could contain antibiotics. Meadow Gold President Jarman
claimed the contamination occurred in the plant's milk lines. Foremost.
with a better testing program than Meadow Gold. dumped contaminated milk
before it entered the plant. Despite DOH clearance of Foremost. the
company still received many calls about the contamination and antici-
pated lower sales. Milk sales had yet to recover to pre—heptachlor-
contamination levels.

By December 11. antibiotics were cleared from the milk supply.
but Meadow Gold continued to face close State scrutiny. DOH officials
considered further penalties against the processor but thought the
recall probably hurt the company more than any fine. The incident was
kept alive in the press since the contamination source could not be
identified. To Meadow Gold's chagrin. it was finally determined to be
its own farm.

Near the holidays. it was discovered that nonviolative levels

of heptachlor had been detected in Meadow Gold imitation milk in

42

mid-November. Skim milk could be used in imitation milk. so the DOH
began consideration of label changes for the product. Health Director
Clark also recommended that fines for violators of the State's mi1k
regulations be stiffened from a maximum of $500 to a minimum of $2.500
per violation and that each violation be considered a criminal offense.
not a misdemeanon. The health director suggested the processors be
required to test milk before it is received into the plant. Milk
Commissioner Roy Matsuura. later accused by the Senate committee of
compromising his duty by his dealings with Meadow Gold. retired.

The new year brought increased milk supplies. Schools resumed
serving fresh 2 percent milk. but consumer confidence had yet to be
restored. The proceedings of the Senate hearings were reported
throughout January. .Sharp criticism was directed at State officials
and Meadow Gold for not making public health their highest priority.
Milk on the island of Hawaii was recalled once for antibiotic
contamination. On January 29. a small article announced that Safeway.

Inc.. had applied for a milk distributor's license to import milk.

3.1.5 Assisnimflasmnsibflm

In February. the Senate investigating committee rel eased its
findings. which were critical of both federal and State agencies and
the milk processors. among others. The FDA and EPA allegedly failed to
ensure public health by allowing heptachlor use in Hawaii without
specific tests to determine the possible health effects from its use on
the pineapple crop. 'The pineapple companies were found negligent for

not monitoring heptachlor application and harvesting of fields. The

43

DOH was accused of being more interested in avoiding recalls and public
criticism than in protecting the public. Allegedly. it failed to
quickly develop a plan of action. was too slow in its recalls. and
sometimes waited for test results before pulling products known to be
made from contaminated milk. 'The DOH was criticized for allowing
processors to use contaminated milk to make skim milk. Clark did not
plan to discipline DOH officials criticized in the report because they
helped the State through the crisis. The committee accused Meadow Gold
of engaging in illegal acts. including the intentional use of contami-
nated milk for products beyond just skim milk. and the destruction of
pertinent production records. 'The company denied some charges and
claimed others were based on "unverifiable assumptions" (in Watanabe 8
Mayer. 2/3/83). ‘The Attorney General was charged with poorly conduct-
ing its Meadow Gold investigation and was accused of justifying the
health department's mistaken advice allowing contaminated milk to be
skimmed rather than correctly interpreting the law. In general.
"Although the heptachlor contamination crisis may have been caused in
large measure by the activities of State officials. much of the blame
for its intensity and duration must be placed on the private sector.
particularly Meadow Gold" (Hawaii Senate. 1983).

Among other recommendations. the committee suggested that the
State Legislature:

--stiffen penalties to deter pesticide misuse.

--require better pesticide-use records from applicators.

44

--transfer responsibility for pesticide monitoring from the
Department of Agriculture to the DOH. and

--provide the DOH the authority to acquire pertinent informa-
tion concerning possible food contamination.

The Governor was exhorted to direct State agencies to:

--review business contracts with Meadow Gold and offer State
contracts to more reliable suppliers.

--appoint a special prosecutor to prosecute Meadow Gold for
use of contaminated milk and Foremost for using contaminated
milk to make skim milk. and

--obtain legal advice from the Attorney General and not offer
uninformed legal opinions.

It was recommended that the DOH:

--impress on its staff its primary responsibility to protect
public health. not industry.

--review and improve its testing program.

--routinely test animal feeds treated with pesticides. and

--announce all recalls ordered by the DOH.

Also in February. after sampling breast milk from 166 mothers.
the findings of the UH Pesticide Hazard Assessment Project were
released. Almost 90 percent had levels too high for infants under four
months. and about 60 percent had levels over theeO.1 ppm action level.
The State Legislature tried to prevent future occurrences of feed
contamination by giving the agriculture department authority to test

all animal feed. The DOH requested the power to inspect or seize

45

records of a company suspected of producing an adulterated or
mislabeled product.

With the completion of the Senate investigation and Safeway's
bid to enter the market in the background. the public health aspects of
the issue began to fade. First. though. Oahu consumer and environmen-
tal groups sued both processors for $250 million each for breach of
contract by selling contaminated milk. Among other demands. the groups
sought a register of infants born from October 1980 to October 1982 and
their health monitoring until the age of 18. a medical expense fund for
children's illnesses due to heptachlor exposure. punitive damages of
$1.000 per milk purchaser. and weekly publication of heptachlor levels
in the processors' products.

In mid-March. Real Fresh. Inc.. a California producer of steri-
lized milk. won the right to a federal court hearing to challenge
Hawaii's authority to restrict interstate milk trade. The legislature
killed a bill mandating an eight-day shelf life for fluid milk. which
was introduced to help restore consumer confidence. The bill would
have effectively killed any attempt to ship fresh fluid milk from
California since the transit is five to six days. Oahu processors
voluntarily set shelf life for their fluid products at 10 and 12 days.
Milk sales. growing slowly. were still down. and acceptable milk was
being dumped when Meadow Gold unveiled its fluid promotion campaign.

As part of a national marketing program introducing a new carton. the

company placed full-page advertisements in the papers for four weeks.

46

Toward the end of March. Safeway's application received more
attention. .As the two processors were threatened with a union strike.
Safeway officials admitted they applied for the license in part because
of the contamination incident and said they would offer Lucerne brand
milk for 10 to 15 percent less than local milk. They foresaw supplying

about 10 percent of Oahu's needs.

3.1.61ha_5af.aaay_c.ontmansx

The latest phase of the milk issue. protection of local
dairies. spilled over when the Board of Agriculture infuriated many
consumers by announcing its intent to deny Safeway's application
because it was "not in the public interest" (in Harpham. 3/31/83). The
Board cited destructive competition and claimed the market was ade-
quately served since acceptable milk was still being dumped daily. The
number of milk-related letters to the editor of both papers jumped from
zero from November through March to 17 in April. overwhelmingly in
favor of Safeway imports.

In response to the Board's intent to deny the license. Safeway
requested a public hearing. Despite reports of traces of DDT in Oahu
milk. and heptachlor in island wildlife. Gov. Ariyoshi asked Safeway to
use Oahu milk in its Lucerne brand. With milk sales still low.
Ariyoshi asked the processors to operate more efficiently. Foremost
considered impOrting milk from California and introduced gallon
containers for whole. 2 percent. and imitation milk. and orange juice.
Fresh milk specials were offered. cutting prices from $1.68 per hal f-

gallon to $1.54. By the end of April. specials continued. but so did

47

dumping of acceptable milk. When an organization named Friends of Oahu
Children asked the'Governor to allow milk importation for health
reasons. Ariyoshi tried to "dispel the notion that Mainland milk is
clean and local milk is not" (in Harpham 8 Oshiro. 4/20/83). He also
offered to make DOH test results public on a regular basis. The legis-
lature voted to fine those responsible for food-safety violations a
maximum of $00.000 and gave the DOH power to seize records pertinent to
possible food contamination or mislabeling.

May brought news that milk from all dairies was below the
action level. but dumping continued. The Hawaii Consumers' League was
formed and began petitioning to pressure the State to approve Safewayhs
application for reasons of health. price. and freedom of choice.
Seventy-five people attended its first meeting. Of the 32 relevant
letters to both papers. two to one favored Safeway. The League's
president wrote:

Few of that segment of the public (about 27 percent) which has
lost so much confidence in the local milk industry as well as in
the government's ability or willingness to protect the consumer
interest will ever buy the local product again--even in a re-
designed package. However. many will tell you they are tired of
powdered milk and want an alternative fresh source. (McMurdo.
5/6/83)

Foremost introduced "Dairyland.".a private label for Foodland stores.
with a price 5 percent less than Meadow Gold or Foremost mi1k. Milk
specials continued and merited mention in an editorial.

The Fresh Milk Industry of Hawaii. consisting of dairymen and

processors. took the offensive against Safeway by placing an advertise-

ment. not in Hawaiian papers. but in the newspapers of Oakland.

48

California. where Safeway's headquarters are located. The ads ques-
tioned Safeway's conscience for trying to destroy the Hawaiian dairy
industry. Dr. Koshi. now executive director of the dairy organization.
claimed Safeway's imports could lead to the industry's demise. leaving
the islands solely dependent on an uncertain supply from the mainland.
Other stores also applied for licenses to import milk if Safeway were
allowed. Koshi claimed the market was adequately supplied because
dumping of acceptable milk continued.

By mid-May. an initial $6.4 million in federal aid from the
Dairy Indemnity Program was made available to Oahu dai rymen. who lost an
estimated $8.5 million. Later in the month. Safeway ran a full-page ad
in the Oahu papers asking. "Who's milking the Hawaiian consumer?" and
implying a tradeoff between 500 jobs and 5.000 babies. A poll con-
ducted for Safeway revealed that 62 percent favored Safeway's applica-
tion although more than half believed that local milk was safe. Of the
32 relevant letters to the editor printed in both papers. more than two
to one were in favor of milk imports.

June was appropriately designated "Dairy Month" by the Governor
since the milk controversy was discussed in the papers almost every
day. The Fresh Milk Industry of Hawaii began a series of weekly ads to
answer consumer questions about milk safety. milk supplies. and the
need for local industry protection. Several dairymen sued Safeway for
$4.5 million for false and malicious claims about the Hawaiian
industry. Despite rumors of heptachlor. and mention of salmonella in

Californian milk. the Hawaii Consumers' League gathered 12.000

49

signatures in five weeks. Before the first week of June ended. though.
the unions employed at Safeway went on strike through June and into the
summer. The local dairymen's claim of uncertain supplies seemed con-
firmed. A study by the Hawaii Department of Planning and Economic
Development was rel eased. estimating costs to the islands of $33 mil-
lion if importation were allowed. By month's end. the Hawaii Con-
sumers' League. the Libertarian Party. the 50th State Dairy Farmers'
Cooperative. the Oahu Dairy Cooperative. and the U.S. Justice Depart-
ment requested permission from the Board of Agriculture to participate

in the Safeway hearings.

3.1.7Baxond_th.e_fitudy_l2aci.od

Since June. events have evolved more slowly. In late August.
the Board of Agriculture granted Dai rymen. Inc.. a license to import
UHT milk. In early September. before its court case challenging
Hawaii's restriction of interstate mi1k trade. Real Fresh. Inc.. was
granted a similar license. The constitutionality of Hawaii's Milk
Control Act was still unresolved.

From September 28 to October 17. the Board of Agriculture heard
testimony regarding Safeway's application. As the local industry
claimed imports would lead to dependence on uncertain external mi1k
supplies. unions at both processors went on strike. Dumping and spot
shortages occurred as management tried to operate the pl ants. Accord-

ing to one industry official. it was only in December 1983. 22 months

50

after the first recall. that milk sales returned to pre-contamination
levels.

Early in 1984. Chairman Jack Suwa of the Board of Agriculture
surprisingly announced he favored granting Safeway's request.
Apparently the controversy was re—ignited. and the hearings were to be
re-opened. Before they were. though. a federal judge in Hawaii ordered
the State to grant Safeway a distributor's license. and in May. fresh
California milk was available to Oahu consumers. The Governor was
considering signing legislation mandating a 10-day shelf life. The

Oahu milk controversy is far from settled.

3.2 W

The legalities surrounding the incident reflect the confusion
stemming from it. One group of consumers sued the State and processors
for ill effects supposedly suffered from heptachlor exposure. Another
sued the State. Gov. Ariyoshi. and Health Director Clark to enforce the
zero heptachlor tolerance level. while another group sued the two
processors for half a billion dollars for selling contaminated milk.
The dairymen sued the pineapple companies for selling contaminated
feed. the State for causing confusion in the market. and the heptachlor
manufacturer (Velsicol Chemical Corp) and its distributor (Brewer
Chemical Coer. The pineapple companies sued the State and Velsicol.
and counter-sued the dai rymen's Green Feed Coop. The company that
insured the Green Feed Coop against such lawsuits itself sued the Coop
to escape liabilities. Meanwhile. Real Fresh. Inc.. had taken the

State to court challenging the constitutionality of the Milk Control

51

Act. and Safeway was sued for defaming Oahu dairymen. As of June
1983. just about every party involved faced legalities. Total awards
could be over $620 million. Value of sales from Oahu dairies were

about $25 million the year before the crisis.

3.3 Advantisincjasoonsa

The newspaper advertising strategies pursued by the two
processors and their rivals differed markedly. Meadow Gold advertised
much more than Foremost. perhaps because the former had a larger milk
supply and more unfavorable media coverage. Rivals stressed the
wholesomeness of their products that were not made in Hawaii.

Meadow Gold responded three days after the first recall with a
ful l-page ad announcing a new milk supply and stating that all Meadow
Gold milk on grocer shelves had been approved by the DOH. Two days
later. Foremost's full-page ad claimed its milk was "good and
wholesome." but if unavailable. "Ditto." Foremost's imitation milk. was
just as good. A week after the first recall. Meadow Gold blitzed the
week's papers with ads describing its imitation and recombined prod-
ucts. even after its buttermilk was recalled.

In early April 1982. Real Fresh. Inc.. began a three-week
campaign of weekly ads stressing that Real Fresh milk. available in
cartons in the commissaries and cans to the public. was "sterilized for
[consumers'] safety" and free of pesticides. "KLIM." Borden's dry.
whole milk. also stressed its wholesomeness and added. 'Drink to your

health. Hawaii."

52

Early in May. Meadow Gold again promoted its imitation
products. Later in the month and through June. new ads with coupons
introduced a promotion where. by buying a half-gallon of milk. the
coupons could be used for savings on other products.

August brought Carnation's annual "Healthy Baby Contest." which
ran through October.

With milk shortages in late September 1982. Foremost
re-introduced its ad assuring consumers of the wholesomeness of its
fresh and imitation milk. and offering coupons for the latter. No ads
were printed from the end of October through February.

Almost on the anniversary of the first recall. and about six
weeks after the Senate investigating committee findings had been
released. Meadow Gold introduced its new carton with weekly full-page
ads saying. "Freshness Never Tasted Better." For the purchase of two
half-gallons of fresh. whole milk. the consumer could get one free
half-gallon of fruit or other drink. The campaign lasted about a
month.

The end of June 1983 brought ads and coupons for Foremost's
gallon container. which had been on the market already for about two
months.

Interviews with advertising directors of the four commercial
Oahu television stations indicate that after the initial recalls.
processors promoted imitation milk. fruit drinks. and some processed
dairy products. At some times. though. when unfavorable news was

reported. both companies completely cut television advertising.

53

creating confusion for the stations. Some directors felt no effective

countermeasures were taken by the companies.

CHAPTER FOUR

THEORY AND METHODOLOGY

With this background of the Oahu milk market and the contami-
nation incident. the theory and methodology of this study may be

developed.

4.1 MW
Standard economic theory holds that quantity demanded is a

function of the product price. price of substitutes. income. tastes.
and preferences. Implicitly assumed is perfect knowledge and maximiz-
ing behavior on the part of the consumer. In a study of consumer
reaction to food contamination. these assumptions are especially doubt-
ful. Bartlett (1973) stated that imperfect knowledge creates uncer-
tainty and the opportunity to influence decision makers by subsidizing
information. Cox (1967) found that consumers with high risk percep-
tions. as could be expected following a contamination incident. seek
information to reduce that risk. It was assumed that the information
they find will affect demand.

Heiner (1983) stated that people cope with uncertainty by
developing behavioral rules and patterns to use in recurrent
situations. As the individual follows habits. Opportunities to

maximize utility are overlooked. The relevance for this study is that

5h

55

consumption habits are used to reduce risk and uncertainty; brand
loyalty is an example of such behavior; Heiner claimed that such rule-
governed behavior will change when the reliability of a new action to
improve performance exceeds a minimum level of reliability. Applied to
food contamination incidents this implies that habits may change when
the risks of consuming a product under question exceed a certain level.
Folkes (1984) implied that negative and unexpected events may lead to
re-evaluation of purchasing behavior. Thus. one would expect a
threshold of negative information before the consumer changes

consumption.

4.2 WWW:

A key assumption of this study is that awareness of contamina-
tion. not contamination itself. affects consumer response. However.
awareness alone may not effect a change in behavior. After a contami-
nation incident some persist in consuming food of questionable safety.
Hence. awareness is apparently mediated by several factors. Source
credibility is one. Sternthal. Phillips. and Dholakia (1978) defined
credibility in terms of expertise and trustworthiness. They cited
evidence that "highly trustworthy and/or expert sources produce a more
positive attitude toward the position advocated than sources that are
less trustworthy and/or expert . . . [and] highly credible sources
often induce more behavioral compliance than do sources having less
credibility" (p. 287). Weinberger and Dillon (1980. Table 3) presented

evidence that implies purchase intentions are more strongly affected by

56

unfavorable product information from an independent testing agency than
favorable information from a trade or professional association.

Aspects of the individual as receiver of information affect
his/her reaction to it. Sternthal. Phillips. and Dholakia (1978)
concluded that audience predisposition affects the influence of source
credibility. "When an audience is favorably predisposed toward an
advocacy. a low credibility source will induce greater persuasion.
whereas when an audience is negatively predisposed. a highly credible
source will be more influential" (p. 298). Following a contamination
incident when consumers may not trust the product in question. a highly
credible source of information is needed to restore confidence.
Locander and Hermann (1979) cited evidence that as the complexity of a
situation rises. self-confidence affects the tendency to accept influ-
ence. They found that as risk rises. consumers most confident in their
ability to make a specific purchase decision rely most on their per-
sonal experience and independent sources of information. In the after-
math of a contamination incident as risk perceptions may rise. some
individuals will mostly trust their own experience and independent
sources of information. Those who are less confident may be more
easily influenced and if most of the information to which they are
exposed concerns the health hazards of the contaminated product. one

would expect sales to fall.

57

”WW
anLBecalls

Studies of the attitudinal and behavioral effects of negative
information are relatively recent. Cusumano and Richey (1970) deter-
mined that negative information given to subjects about the character
of a stranger has a greater effect on evaluations of the stranger than
positive information. Further. the order in which information is given
(first negative. then positive and vice versa) has little effect on the
final evaluations. They concluded that the effect of negative infor-
mation overwhelms that of positive information. Richey et al. (1975)
again presented favorable and unfavorable information about a stranger
to subjects who then rated his characten. Results were that a single
negative behavior counteracts five positive behaviors; the stranger was
rated the same when five positive and only one negative statement were
made of him as he was when five positive and five negative statements
were made. Thus "one slip—up" is all it takes to influence an indi-
vidual's reputation among those who are not familiar with him/ her.
Weinberger and Dillon (1980) applied such work to consumer purchase
intentions in response to favorable and unfavorable product informa-
tion. They found that unfavorable ratings have a disproportionately
greater impact than favorable ratings.

With respect to product recalls. Mowen (1979. cited in Mowen.
l980)9c1aimed to have conducted the first experimental investigation
into consumer perceptions of product recalls. He found that a company
that makes a series of recalls is held more responsible for the defect

than a company that makes fewer recalls. Similarly. the longer a

58

company delays in making a recall. the lower are perceptions of its
concern for consumer welfare. Further. the extent of injury resulting
from the defect also influences perceptions of the company and
intentions to purchase a replacement product. In his similar 1980
study. Mowen surveyed a sample of secretaries. homemakers. and members
of the League of Women Voters after presenting them with information
about recalls by Corning Ware and a fictitious company. The more
familiar. reputable company was held less responsible for its product
defect than the fictitious company. Surprisingly. when a company
voluntarily recalled its product (while the Consumer Product Safety
Commission (CPSC) was still studying the matter) it was held more
responsible for the defect than when the CPSC recalled the product.
Mowen hypothesized that when the CPSC recalls the product. it (somehow)
shares responsibility for the defect.

Mowen. Jol ly. and Nickell (1981) improved the external validity
of Mowen's earlier studies by surveying a random sample of people for
their perceptions of four companies. Ford Motor Company. Firestone.
Inc.. Corning Glass Works. and Conair. Inc.. which recalled products.
Different factors related significantly to perceptions of different
companies. suggesting mediating factors influence perceptions of indi-
vidual companies. Ford and Firestone's social responsibility (or lack
thereof) was related to perceptions of those companies. Other factors
found to be related to Firestone were the consumer's knowledge of the
product recall. the defect danger. and whether any other tire companies

made recalls. Perceptions of Conair were related to the danger of the

59

defect and corporate responsibility for it. No significant effects
were found to explain perceptions of Corning Glass. which Mowen
explained was due to either the small number of respondents aware of
its recall or its good reputation. Mowen claimed length of time to
recall. not found significant here. was indirectly related to consumer
perceptions through impression of corporate social responsibility. He
did not reconcile the finding that the number of previous recalls was
not significant here. though it was in his 1979 work. Mowen and Ellis
(1980. cited in Mowen 8 Pollman. 1982) again presented a fictitious
magazine article about a company recall. The severity of injury.
number of previous recalls. and length of time to recall influenced
perceptions of the company and interest in purchasing a replace-

ment product from it.

Mowen and Pollman (1982) continued the research exposing under-
graduate business students to press releases about a recall. The order
of release of information was studied. ‘The company that first
announced the most severe outcome and later issued less severe warnings
was perceived more favorably. and as more honest. credible. and con-
cerned for consumer welfare than companies whose press releases grew
more alarming with ti me. "Messages which were in-role. such that they
follow the obvious interests of the communicator. tended to be per-
ceived as less credible and as made for the benefit of the sender and
not the receiver" (p. 219).

The above research may be extended to consumer response to

food contamination incidents. One would expect that response is

60

influenced by the severity of the contamination's effect (e.g.. botu-
lism in canned food). the company's responsibility for the contamina-
tion. the number of previous recalls. and the company's social
responsibility. related to the length of time to recall. Announcing
the worst possible outcome first enhances the company image and may
limit sales 1055 in the long run. If government is responsible for
overseeing food safety. questions of its responsibility for the con-
tamination and concern for consumer welfare may influence perceptions
of the food product and hence sales. These factors may be mediated by
the credibility of the source of the information and consumers' predis-

position. risk perceptions. self-confidence. and other factors.

4.4 Wu
W

Awareness of product quality is gained through the mass media
and interpersonal communication. In the field of journalism. it is
believed. though not without controversy. that the effect of mass media
on the public is to set the agenda of issues which people consider.
The agenda—setting hypothesis states that the "media may not be espe-
cially powerful in telling people what to think. but they can be quite
successful in determining what people will think about" (Cohen. cited
in Chaffee. 1980). "Newspapers clearly state the value they place on
the salience of an item through headline size and placement within the
newspaper. . . . Agenda-setting asserts that audiences learn these
saliences from the news media. incorporating a similar set of weights

into their personal agendas" (McCombs 8 Shaw. 1977). People take a cue

61

from the priority given an issue in the media in determining the
issue's priority for themselves. Thus when much attention is devoted
to a contamination incident. people will consider the contamination.
Some may discount it. but they will consider it.

A weakness of the agenda-setting hypothesis with respect to
issues of product quality is that much of the evidence supporting it
stems from the effect of media coverage of political issues on persons'
priority attached to those same issues. However. one study by Bloj
(cited in Sutherland 8 Galloway. 1981) suggests that the agenda-setting
hypothesis also applies to product quality perceptions. He found a
correlation between the prominence of air crash coverage. a fall in
ticket sales. and an increase in sales of flight insurance.

A second avenue in which consumers are exposed to news about
product quality is through interpersonal communication. Eugene Shaw
(1977) found evidence implying that issues given prominence in the
media are also those issues discussed among people. Again. though. his
study concerned political coverage and discussions leading up to the
1972 Presidential campaign. Arndt (1967) found that word-of—mouth
advertising among homemakers does affect purchases and that those with
the highest risk perceptions are most affected by word-of-mouth expo-
sure. Therefore. following a contamination occurrence. consumers may
gain awareness of the incident through the mass media. and this may be
reinforced by interpersonal communication. Hence. studying media cov-
erage may closely reflect total consumer awareness of food contami-

nation.

62

4.5 Wm
ContaminatioLInCJdents

Given that food contamination incidents have gained attention
principally after the 1973 FEB incident. it is not surprising that
there are few studies of consumer response to food contamination.

These studies are scattered among the literature of the marketing.
consumer research. and agricultural and environmental economics profes-
sions. They either employ a dummy variable or a variable quantifying
media coverage of the health threat to examine the effect of a health

warning on product sales.

45.1mm.
QuantifiaLMediLllaniabla

The first known work (Brown. 1969) dealt only with a possible
change in cranberry price elasticity following the 1959 Department of
Health. Education. and Welfare announcement that cranberries could
contain residues of the herbicide and suspected carcinogen amino tria-
zole. 'The pre—Thanksgiving announcement was particularly untimely for
producers. From 1957 through 1962. 300 Atlanta. Georgia. families
recorded weekly food purchases. Brown hypothesized that since adver-
tising decreases price elasticity. product contamination or adverse
advertising would increase it. Per capita purchases regressed on
average price. age of homemaker (a one-four dummy variable). and per
capita income in periods before. during. and after the scare revealed
no significant change in elasticity. Per capita purchases did fall 26

percent in 1959 but regained the 1957-58 level during 1960-62. The

63

specific extent and duration of reduced purchases are unknown since it
was not a study objective. Brown suggested that industry advertising
may offset changes in demand elasticity.

Hamilton (1972) examined the relative effects of advertising
and health warnings on cigarette demand from 1925 through 1970. Annual
per capita cigarette consumption by persons over 13 years of age was
regressed on measures of price. per capita income. and advertising.
Three zero-one dummy variables were also used. representing. respec-
tively. the first major cigarette health report (1953). the 1964 Sur-
geon General's report. and the beginning of the significant amount of
broadcast anti-smoking messages in 1968 mandated by the Federal Commu-
nication Commission's Fairness Doctrine. Hamilton found that the
health scare decreased consumption more than advertising increased it.
An interesting point is that the 1968 dummy (representing anti-smoking
broadcasts) had a greater effect than the single 1964 Surgeon General's
report. which in turn had a greater effect that the initial report. If
true. then this would illustrate the relative impacts of a one-time
announcement compared to repeated exposure to a health warning. This
conclusion may not be drawn. though. since the greater impact of the
1968 dummy variable may reflect carryover effects from the 1964 warn-
ing and other factors.

Witt and Pass (1981) conducted a very similar study on
cigarette consumption in Great Britain. Three major health warnings
were represented by three dummy variables. The authors found that the

health scares reduced consumption so much (3 to 7 percent) that

64

advertising expenditures would have had to double to compensate for the
1055. Using annual data to determine the duration of the health
scares. their demand equation was estimated three ti mes with the dummy
variables taking the value of one only in the year of the respective
health warnings. also in the year following each health report. and in
the year of the warning and two subsequent years. The authors con-
cluded the health warnings had a two-year lag. ‘This is an improvement
over Hamilton's work. whose dummy variables retain a value of one for
every year following the respective health warning. However. Clarke
(1976) stated that the duration of advertising's effectiveness is
overstated in models using annual data and that advertising's effect
does not last more than a year. Thus. a two-year effect on sales is
questionable if a health scare is perceived similarly as advertising.
If the negative information from a health scare is weighted more
heavily than advertising information. as may be the case. then a two»
year lag is possible.

Hoffer and Wynne (1975) studied consumer recalls of subcompact
cars by examining changes in subcompact market shares of seven domestic
and foreign manufacturers. Monthly independent variables included own
price and advertising. the differences between own price and that of
competitors. the differences between own advertising and that of
others. and the number of recalls. Each safety-related recall was
assumed to be of equal weight; attempts to quantify media coverage
"proved to be highly subjective and unsatisfactory" (p. 214). An

instantaneous. noncumul ative relationship between market share and the

65

number of recalls yielded best results. Findings indicated that
recalls had an insignificant impact on market shares for all makes
except Vega. The authors noted that Vega had slightly fewer recalls
(4) than the average make (4.3) but that they were in consecutive
months. ‘They contrast this to two other makes with six and eight
recalls distributed over the observation period. which suffered no loss
of market share due to the recalls.

Vega's loss may be due to heightened consumer awareness from
the consecutiveness of the recalls. Strong (1977) recommended that
advertisers group advertisements and schedule these groupings close
together to maximize effectiveness. Given a flurry of press coverage
surrounding the monthly recalls. it is suspected this same effect
worked to Vega's disadvantage. Alternatively. the reduction in market
share could have been due to consumer perceptions of a greater safety

hazard with Vega than with other makes.

4.5.2 Wang
Ouanfliiedfledialaniahle

Schuker et a1. (1983) examined the effect of the 1977 saccharin
health controversy on sales of diet soft drinks. The FDA announced
that it would withdraw approval of saccharin use after studies indi-
cated its possible carcinogenicity; Congress placed a moratorium on
the ban but required a warning label on saccharin-containing products.

Diet soft drink sales were regressed on diet soft drink price
and advertising. regular soft drink price. trend and seasonality

variables. and two dummy variables. The warning-label dummy variable

66

took the value of one after the warning labels appeared. The other
took a value of one only in those periods when (print) media coverage
exceeded the average of all periods. This major news variable was
found superior to other measures of coverage. such as the number of
editorials. number of all news items. column inches devoted to the
issue. the tone (positive. neutral. or negative) of advocacy about the
saccharin ban. and weighted tone (number of items in a period
multiplied by the average tone of the items). Fdndings were that the
major news dumnw'had an insignificant impact on diet soft drink sales
while the warning label did explain a reduction in the rate of growth of
diet sales. The authors explained media coverage as insignificant due
to measurement problems. and that intense coverage of a short duration
was expected to have less effect than a continuous warning. Further.
in their demographic analysis of the warning's impact. college-educated
consumers were first found to reduce consumption. followed by those
with young children. The elderly and consumers without a high school
diploma did not alter consumption. in spite of the warnings and media
attention.

Schulstad and Stoevener (1978) incorporated an information
variable in their demand analysis to determine the loss of consumer
surplus as consumers altered consumption to avoid a contaminated
product. ‘They estimated the effect of information of mercury
contamination in pheasants on the average number of pheasant-hunting
days per hunter per season. and its impact on the number of pheasant

hunters in Oregon from 1950 to 1971. As an information variable. the

67

cumulative number of articles two years prior to the hunting season was
found superior to column inches or number of newspaper articles one
year before the season and cumulative column inches of articles two
years before the season. All four were highly correlated though.

The information variable was not significantly related to the
average number of hunting days per hunter. though it was highly
significant in explaining the number of hunters per season. Hence. it
appears people will completely exit the market of a contaminated good
rather than simply reduce consumption. This must be qualified since
there are many substitutes for pheasant hunting. If a product with few
substitutes is adulterated. consumers may have no exit option and may
limit consumption.

It was estimated that about 17.600 hunters stopped hunting in
1971 because of the contamination information. The percentage of all
Oregonian hunters this was is unknown. Aggregate costs to persons who
eschewed hunting to avoid contamination were estimated to be $1.35
million.

Swartz and Strand (1981) similarly incorporated an information
variable in their model of oyster demand in the Baltimore. Maryland.
market. Supposedly. only uncontaminated oysters were sold in
Baltimore. but 200 miles to the south. kepone contamination had closed
the James River oyster beds and had attracted much media attention.
Newspaper articles were rated according to their negative impact on
oyster consumption. As a proxy for their likelihood of being read.

each story was weighted by the cost of advertising in the space

68

occupied by that story. Using biweekly observations. the authors found .
the media variable was significantly related to the fall in oyster
sales with a two-month lag. even though the oysters were uncontami-
nated. Estimated losses to consumers and producers from this imperfect
information were about 5 percent of total value of marketing during the
news-release period. Losses from a failure of oyster price to rise
after the James River was closed. and losses to other seafoods. were
not calculated.

The authors also examined the timing of the media coverage and
oyster sales and concluded that losses may be affected by when the
government announces the contamination. It was estimated that had
coverage occurred earlier in the oyster season. losses would have been
25 percent greater. whereas if it occurred near the end of the season.
losses would have been similarly reduced. .As a policy implication.
they stated that if government officials know with certainty that
consumption of one product is safe even though a related product is
contaminated. losses as consumers attempt to avoid poisoning may be
minimized by properly informing the public. (The implicit assumption
is that consumers believe government is acting on their behalf. and not
to protect the industryJ However. the authors admitted that certainty
in the event of a contamination incident may be unlikely. Even with
all available information. consumers may still be uncertain about food
safety and will change consumption to avoid contamination.

This study is an improvement over other studies in that it

allows one to observe the relationship between different levels of

69

current and lagged media coverage and sales response. It lends
understanding to the distributional impacts of a food contamination
incident in that even conscientious producers of an undifferentiated
product suffer when less ethical or competent producers create a health
scare. or when an unforeseen incident strikes some producers. However.
their weighting scheme is questionable since the value of advertising
on the front page is unknown. Further. the authors only consider those
news articles that negatively influenced oyster consumption. .Attempts.
if any. by government officials to restore confidence in the industry
were apparently overlooked. ‘The question remains of the effectiveness
of government in restoring confidence once the contamination has been

announced.

4.6 .InaQna11Qal_StatamanI_QI_IhQ_ELthflm

Following a food contamination incident. consumer tastes. pref-
erences. and perhaps habits change. shifting the demand curve back. If
supply is unaffected. as assumed in the studies above. and assuming it
is fixed in the short run. the revenue loss to producers is the area
POP1ba in Figure 4.1a. If however. product is recalled. shifting
supply back as well. several results are possible. If a shortage
develops. price is expected to rise. offsetting some of the sales 1055
due to product dumping. (Area PzPodc compared to loss due to dumping.
area QoadQ1 in Figure 4CHL) If demand falls as much as supply is cut.
producer market 1055 is limited to the amount dumped. (Area QoaeQz in
Figure 4.1c). However. if demand falls more than supply is reduced

(i.e.. consumers refrain from purchasing the product). prices will

 

 

 

 

Sales Loss Only

0
k .
t

Qo

(a)

 

 

 

 

 

52 S'0
\ {
8
Q2 0.0
(a)

Dumping Loss Only

 

 

 

 

 

 

Q1

Q2
(1:)

w
U1
0

 

 

 

 

 

Dumping Loss Offset by Price Rise

A

 

Q3 Q0
(d)

Dumping and Sales Loss

Figure “.l.--Areas of loss.

71

 

 

 

 

 

 

 

Oahu Dumping and Sales Loss

Figure 4.i.--Cont'd.

72

fall. Now. in addition to dumping losses. producers suffer the addi-
tional loss due to this consumer resistance. (Area P0P3fg in addition
to area Qoagos in Figure 4:“1) Competitors may attempt to capitalize
on the contamination by entering the market and may permanently change
purchasing behavior. In the case of a differentiated product. market
shares may shift as seen in the case of auto recalls. In cases of
industry-wide contamination. consumers may change habits. such as
switching from meat consumption to a more vegetarian diet.

Estimating costs to Oahu dairymen requires modification of this
theoretical approach. Through Hawaii's Milk Control Act (1967). the
producer price received for milk used for fresh fluid use (Class I) is
set by the milk commissioner. Over the course of the contamination
incident. this price remained constant at $21.09 per hundredweight.
Hence. producer costs are two-fold: the value of milk dumped due to
contamination (QoahQ4 in Figure 4.1e). and the value of acceptable milk
that could have been used for fresh fluid use. but because of reduced
demand was used otherwise (Q4hiQD). The amount of contaminated milk
dumped is known; the problem at hand is to measure the losses due to
reduced demand.

Only awareness of the contamination will create such a
response. From the Witt and Pass (1981) and Swartz and Strand (1981)
studies. negative response to a health warning or contamination incident
is temporary. One expects consumption to drop as awareness of contami—

nation rises. As the contaminant is cleared from the production system

73

and reported so by the media. consumption should approach if not return

to normal levels. Expected response is shown in Figure 4.2.

4.7 Methodology

Because many factors influence fresh fluid milk consumption on
Oahu. an econometric approach is used to determine how sales were
affected as a result of the incident. Assuming supply is fixed in each
period. a single-equation model was estimated. As did Brown (1969).
publicity of the contamination was considered adverse advertising for
the industry. In this respect. estimates of advertising's effect on
milk demand by Thompson and Eiler (1975b. 1977). Thompson. Eiler. and
Forker (1976). and Kinnucan (1983) were particularly useful.

Models of quantity demanded frequently include explanatory
variables such as the product's price. prices of substitutes. income.
population. demographic and other changes over time. seasonality. and
advertising. However. additional factors influence consumption after a
contamination announcement and must be included in the model. One of
these factors is consumer awareness of food safety. which has three
components: food quality (the danger from the contaminant in the
product). adequacy of government protection of food supplies from
contamination. and hence protection of public health. and the
producersfl/processors' integrity (their social responsibility and
concern for consumer welfarek To reflect awareness of food safety.
measures of negative and positive media coverage of food safety were

included in the model.

 

7h

 

rConsumption

..... Negative media coverage

—--—-Positive media coverage

 

 

 

TIME

Figure h.2.--Expected response to media coverage.

>—orn::

MQ>ZM<OO

75

The scope of contamination may also affect consumer response.
and consumption habits may change. To account for the contamination's
scope. the amount of product dumped was included in the model.
Initially. one would expect this to be related to supply and have no
effect on demand. However. if most product is dumped due to contamina-
tion. consumers may suspect even that product allowed on store shelves.
As the extent of contamination diminishes. consumers may become less
concerned about it. Indeed. as the problem fades. new concerns about
the dangers from the contamination may have less effect on consumers
because most of the problem has been corrected. This was particularly
appropriate in study of the Oahu incident because 95 percent of island
cows were contaminated and consumers apparently suspected milk on the
shelves too. (Indeed. heptachlor levels were elevated. even though
below the allowable or "action" levelJ Given that consumers were
repeatedly asked to return or dispose of their milk and dairy products.
it is possible that the slight downward trend in milk consumption may
have steepened as milk consumption habits changed. Hence. factors
reflecting this are also incorporated into the model.

The complete model of monthly milk consumption is given below:

m
Q = f(DPM, DPS, DPCPI, Seasonality, TIME, 2 ADt-i’
i=0
n p )
2 NM _., 2 PM _ , DUMP, DV, DVTRND
J80 t J k=0 t k

where:

0.

fresh

76

fluid milk consumption per capita per day

adjusted for the calendar composition of each month.

DPM

DPS

DPCPI

TIME

retail whole milk price deflated by the Honolulu CPI.
retail deflated price of substitutes.
deflated per capita personal income.

a time trend variable (1. 2. 3. ...) to account for demo-

graphic and other changes over time.

Seasonality =

Eleven monthly dummy variables reflecting the

pattern of consumption during the year using December
as the base.

m
2 AD-
i=0 t

[11

ZNM.=
jO to

I =

current and lagged deflated advertising expenditures by
the industry.

monthly negative or unfavorable media coverage of milk
safety and lagged values.

monthly positive or favorable media coverage of milk.

DUMP = pounds of contaminated milk dumped monthly.

DV =
tion.
DVTRND =
by an
trend.

a dummy variable. zero before the March 1982 contamina-

one after. and

a factor to account for a change in habits reflected

(expected) shift downward in the slope of the
It takes values 1. 2. 3. ... after the con-

tamination incident. DV was included to allow a shift
in the intercept of the trend line.

The model was judged by several criteria. First was its

consistency with economic logic including reasonable elasticities.

Second was its ability to explain the variation of the dependent

variable since it was used to project sales in the absence of the

incident. Third was the statistical significance of the coefficients

and absence of serial correlation.

77

From the model. projected sales were compared with estimated
actual sales. The difference reflected the amount of sales lost due to
consumer awareness of the contamination. However. it is known that
much milk was dumped because of excessive pesticide residues. Hence.
to determine the amount of sales lost over and above the loss due to
product dumping. the amount of milk dumped was subtracted from the
difference between projected and estimated actual sales. To determine
if any mention of milk at all after the contamination announcement
influenced current awareness of milk safety. the model was estimated a

second time with all articles considered negative information.

CHAPTER FIVE

DATA AND FINDINGS

5.1 .Qata

Compared to areas under federal milk market orders. data
regarding the Oahu milk shed are much more difficult to obtain. Due in
part to the duopoly at the processor level. the State Division of Milk
Control reports only aggregate data. Moreover. numerous lawsuits and
investigations following the incident made processors and others
reluctant to disclose information.

Nevertheless. adequate data were obtained to estimate fluid
milk demand. A period of five years before the incident and 15 months
afterward (through June 1983) was selected. Appendix One details the
description of the data below. Estimation beyond 15 months was limited
by the lack of access to Honolulu newspapers and the unavailability of
accurate population estimates (principally the number of tourists and

their average length of stay).

5.1.1 Consumption

Fluid consumption data were obtained chiefly from the~Hawaii
Division of Milk Control. It was assumed that before the contamination
announcement. Class I utilization. the amount of milk that processors

bottled. approximated quantity demanded. Because this assumption was

78

79

questionable after the contamination announcement. Class I utilization
was corrected for imports and estimated route returns. Further. from
September through December 1982. schools were supplied with imitation
milk to avert retail shortages. To accurately reflect quantity
demanded. the amount of milk that would have been sent to schools was
added to the Class I utilization data for that period. Failure to
correct for this underestimates the dependent variable in October and
November by about 15 percent. and to a lesser extent in September and
December. It was assumed that Class I utilization corrected by these
factors approximated quantity demanded.

Because milk sales vary according to the day of the week. per
capita consumption for each month may vary between years simply because
one month may have more "high sales" days in one year and more "low
sales" days in another. Hence consumption per capita per day for each
month was corrected by a monthly adjustment factor (see Schlenker 8

Christ. 1971).

5.1.2 Emulation

To calculate per capita consumption. estimates of the de facto
population (actual number of people on the island) were required. It
is known that on a given day. tourists comprise about 10 percent of
total population though this varies monthly. Hence. estimates of the
Inonthly number of tourists on Oahu (derived from data from the Hawaii
Visitors Bureau) were added to estimates of monthly resident population

interpolated from annual data.

80

5.1.3 Euros

Retail food prices in eight Oahu supermarkets were surveyed by
the Market News Branch of the Hawaii Department of Agriculture (1977-
1980) and the Honolulu Adlertiser (1981-present). Average retail sales
were used for a one-half gallon paper container of whole milk. and a
variety of possible milk substitutes including one-half gallon con-
tainers of imitation milk and fruit nectar. a l4-quart package of non-
fat dry milk. 46 ounces of fruit drink. and 12 ounces of canned soda.
In 1977. each was deflated by the Honolulu Consumer Price Index for
urban wage earners and clerical workers. From 1978 onward. each was
deflated by the Index for all urban consumers. The Bureau of Labor
Statistics linked the two series so that both have the same value in

December 1977.

5.1.4 Income

Per capita personal income was calculated in two steps.
Quarterly state personal income was interpolated to derive monthly
estimates. This was divided by monthly estimates of state resident
population (similarly calculated) to estimate per capita personal
income. It was assumed that Oahu and state per capita income were

similar. Income was deflated by the same method as were prices.

5.1.5 thirertising

The work by Thompson and others previously cited shows generic
milk promotions significantly affected milk sales in some New York mar-

kets. Hence it was suspected that milk advertising may have a similar

81

effect in the Oahu milk shed. No generic advertising is undertaken
there. but the two processors advertise their brands. Due to the
legalities ensuing after the contamination. both processors refused to
disclose mi1k advertising expenditures. The two major newspapers also
declined to reveal the amount of advertisement the companies had pur-
chased. Two of four television stations provided gross advertising
expenditures for each processor. but this included advertising for
other dairy and non-dairy products. Hence. the advertising variable
was not included in the estimations. Conversations with those in the
advertising industry in Oahu indicated that fluid milk advertising was
not heavily undertaken by either company. so this was not a major over-
sight. An exception might have been March and April 1983 when Meadow

Gold unveiled its new milk container.

5.1.6 Maoiaiorerage

Given that consumer awareness of the contamination may affect
sales. a proxy for that awareness was necessary. As discussed pre-
viously. the mass media will affect public awareness; hence a measure
of media coverage was derived. It was assumed newspaper coverage
reflected total media coverage of the incident. ‘This was supported by
the national sales representative of one Oahu television station who
said television coverage of the contamination was similar to that of
the newspapers.

All articles related to milk and the incident were obtained
from the two major Honolulu newspapers. the.flono1u1u_Adxertiser and the
.flgnglu1u_§1ar:fiulletin. Each was coded with respect to whether they

82

presented positive. negative. or neutral information about milk safety.
As each article was coded. confidence in the code was graded (4.0 =
very sure; 04) = not sure at all). This gave a measure of the relia-
bility of the coding. ‘To determine how media coverage of the incident
evolved. each article was coded not only with respect to overall milk
safety. but also with respect to the components of milk safety. that
is. milk quality (presence and/or danger of the contaminant in milk).
the level of government protection. and the processors' concern for
consumer welfare (their integrity).

Many measures of the amount of newspaper coverage exist:(see
Budd. Thorp. 8 Donohew. 1967. also Berelson. 1952). However. to best
reflect the prominence of an article. the "Attention Score" developed
by Budd (1964) was used which weights articles based on their placement
in the paper. position on the page. size of headline. and length of
article. Weights range from zero for a small article on the bottom
half of page six. to five for a lengthy front-page article with a
banner headline. The negative media variable was the sum of weights of
those articles coded negatively in each month; similarly for the posi-
tive media variable. The content analysis. including coding criteria.

and Budd's "Attention Score" are detailed in Appendix Two.

5.2 EconometriLEindings

5.2.1 Determination:
Aoorooriatoifllkjuostitute

First to be resolved was the choice of an appropriate substi-

tute to include in the model. From Thompson and Eiler (1975a).

33

appropriate choices may be beer. coffee. or canned soda. Knowing the
Hawaiian mi1k market. imitation milk may be more appropriate. However.
a consumer survey was obtained from one of the processors showing that
after the contamination announcement. more people switched to consump-
tion of fruit juice or drinks than any other substitute. Hence. pre-
contamination consumption was regressed on the deflated prices of whole
milk (DPM) and fruit nectar (DPFN) and then on deflated prices of milk
and fruit drink. The superior substitute in terms of sign and signifi-
cance of coefficients. estimated elasticities. and ability to explain
variation in consumption and absence of serial correlation was fruit

necta r.1

This compared favorably with Hogg's (1974) estimation using
annual data. (As with Hogg's study. income was of unexpected sign and

insignificant when includedJ

 

1The regression was:

0 = 6.81 - 4.49 DPM + 2.26 DPFN + 0.341 JA + 0.322 FE + 0.221 MA
(1.33) (1.87)* (0.830)* (0.145)* (0.145)* (0.152)

+ 0.610 AP + 0.516 MY - 0.244 JE - 0.250 JY - 0.169 AU
(0.151)* (0.151)* (0.151) (0.152) (0.152)

+ 0.610 SE + 0.373 00 + 0.219 N0
(0.155)* (0.151)* (0.152)

R2 = 0.617 o.w. = 1.84 price elasticity (np) = -o.so3
Figures in parentheses are standard errors.
*Significant at the a S 0.05 level.

Other substitutes were also tested but none found any better.

84

5.2-2 W

Estimation of the model yielded the following results:

(1a) 0 = 10.2 - 4.69 DPM + 0.926 DPFN - 0.000500 DPCPI
(4.09)** (2.16)** (0.836) (0.000709)

- 0.00663 TIME + 0.233 JA + 0.222 FE + 0.0955 MA + 0.451 AP
(0.00199)*** (0.131)* (0.130)* (0.131) (0.141)***

+ 0.413 MY - 0.379 JE - 0.367 JY - 0.234 AU + 0.515 SE
(0.140)*** (0.136)*** (0.137)*** (0.137)* (0.139)***

+ 0.368 00 + 0.161 NO - 0.149 0v - 0.0240 DVTRND
(0.133)*** (0.131) (0.365) (0.0276)

- 3.15 x 10'7 DUMP - 0.00301 NM - 0.0271 PM
(7.34 x 10'3)*** (0.00292) (0.00892)***

-0.572

R? = 0.938
= 2.1 -0.417

n
D.W. p

4 ny

Standard error of regression = 0.224 oz./person/day
Mean of dependent variable = 5.14 oz./person/day

Figures in parentheses are standard errors.
*Significant at the 0.5 0.10 level.
**Significant at the a.$ 0.05 level.

***Significant at the 0.5 0.01 level.

The estimation contained several unexpected results. A
negative though insignificant relationship between deflated per capita
personal income (DPCPI) and consumption was indicated. ‘This may have
resulted from income being correlated with another factor which
negatively influenced consumption. The negative coefficient of the
positive media variable (PM) may be due to its high correlation with

negative media (NM). (The coefficient of correlation was 0.767.) A

85

positive coefficient of the dumnw'variable was expected if the downward

trend in consumption steepened.

that the trend not only steepened but still shifted downward.

Its negative coefficient indicates

Confi-

dence in the coefficients of both DV and DVTRND is low though. Income

elasticity is unusual. but insignificant. Price elasticity is high.

but not unreasonable compared to elasticity estimates shown in Table

5.1. Because retail milk prices in Honolulu are the nation's highest.

one would not be surprised if the price elasticity were higher.

Table 5.l.--Estimated elasticities of retail fluid milk demand.

 

 

Source Period Location Own-Price Income
George and King 1946-67 U.S. -0.35 0.20
(in Tomek. 1981)

Renaud (1971) 1950-67 Hawaii ... 0.27

Prato 1 Florida -0.58 ...

(in Cook et al..

1978)

Boehm (1976) 1966-75 Average -0.12 to 0.07 to
across -0.30 0.14
U.S. cities

Thompson and Eiler 1971-74 New York -0.19 0.29

(1975b) City

Kinnucan (1983) 1978-81 Buffalo. -0.66 to 0.23 to
New York -0.73 0.39

 

86

The positive media variable was dropped due to its unexpected
sign and correlation with negative media. ‘The new estimation is shown

below:

(1b) 0 = 10.3 - 4.39 DPM + 0.826 DPFN - 0.000539 DPCPI

(4.37)** (2.31)* (0.893) (0.000758)

- 0.00688 TIME + 0.249 JA + 0.218 FE + 0.0670 MA + 0.422 AP
(0.00213)*** (0.140)* (0.140) (0.140) (0.151)***

+ 0.350 MY - 0.353 JE - 0.367 JY - 0.183 AU + 0.430 SE
(0.148)** (0.145)** (0.147)** (0.145) (0.146)***

+ 0.359 CC + 0.191 NO + 0.142 DV - 0.0449 DVTRND
(0.142)*** (0.139) (0.377) (0.0285)

- 4.60 x 10'7 DUMP - 0.00888 NM

(5.93 x 10'8)***

R2 = 0.929
0. =

W. 2.19

Standard error of regression =

Mean of dependent variable

(0.00233)***

[b

”y

-0.535
-0 0449

0.239 oz./person/day
= 5.14 oz./person/day

*Significant at the a f 0.10 level.

**Significant at the a.§ 0.05 level.

***Significant at the a g 0.01 level.

The sign of income's coefficient was again unexpected but still
insignificant. Because income was expected to be positively related
to consumption. a conditional regression was estimated fixing income

elasticity at that found by Renaud (1971).2 (See Table 5.1.) The

 

2Other options were explored including the omission of DV and
DVTRND. However. DVTRND is "statistically suggestive" and when both
were dropped. the sign of income's coefficient became positive as did
that of DPM (though both were insignificantL Simply dropping income
from the regression yielded coefficients that were insignificantly
different from those generated by the conditional regression.

87

quantity on = 0 - [0.267 x ((T/ffizP—I) x DPCPI] was used as the dependent
variable.

Further. the lag on the negative media variable was also
explored. .An ad-hoc method was used. successively lagging NM until its
coefficient became insignificant or until the sign changed (Gujarati.
1978L A usual problem with this approach is that multicollinearity
may distort the t-statistics. In the case of NM. multicollinearity was
doubtful since values changed from a series of zeros to very high
values (122) as the recalls were announced. These values then declined
but with much fluctuation. It is recognized that geometric and
polynomial distributed Tags are commonly used as well (e.g.. Yon 8
Mount. 1975; Butler 8 Thompson. 1979).

The model selected involved a one-month lag. (The two-month

lag had a positive coefficient.) Results follow.

88

(1c) 0 = 5.68 - 3.07 DPM + 0.663 DPFN - 0.00582 TIME + 0.309 JA
(1.25)*** (1.78)* (0.845) (0.0196)*** (0.133)**

+ 0.275 FE + 0.117 MA + 0.514 AP + 0.457 MY - 0.300 JE
(0.132)** (0.135) (0.137)*** (0.137)*** ( 0.136)**

- 0.319 JY - 0.125 AU + 0.528 SE + 0.400 CC + 0.207 NO
(0.139)*** (0.141) (0.150)*** (0.138)*** (0.137)

+ 0.0305 0v - 0.0303 DVTRND - 3.70 x 10'7 DUMP
(0.384) (0.0298) (8.29 x 10'8)***

- 0.0103 NM - 0.00463 NM(-1) + 0.000311 DPCPI
(0.00229)*** (0.00282)

R2 = 0.929 up = -0.511
Standard error of regression = 0.236
Mean of the dependent variable = 3.77

*Significant at the a g 0.10 level.
**Significant at the o g 0.05 level.

***Significant at the o g 0.01 level.

Signs were all as expected and the Durbin-Watson statistic indicated a
low probability of serial correlation. ‘The coefficient of lagged NM
and DVTRND were insignificant but "statistically suggestiveJ' The
model explained more than 90 percent of the variation of the dependent
variable; actual and estimated per capita consumption are shown in
Figure 5.1.

The direct and carryover effect of negative media coverage is

the sum of the coefficients of current and lagged negative media. or

39

.mmm_ ocsa

-mmm_ >Lmacmw .oOume_umo ocm _o:uu< ":o_qu:mcou mu_omu cud--._.m 0c:m_d

3m.

_

Nam.

5m—

P

0mm—

; _

mum. mmm_ sum—
22. uuo 22. uma 2:... own 22. uwo 2:... own 2:... own 2:... 23.

_

_ _ b

 

 

 

 

zouhmzzmzou owhctzmu ....-.
23.—$5980 ..cahoz 1.1

\'

 

 

ILJDHD ODZUUM LNG DC)-

90

-4L0149. Evaluated at the mean of the variables. a negative media
elasticity of -0.0219 was calculated.3 This indicates that a 1 percent
increase in negative media coverage reduces consumption about 0.02
percent.

It is interesting to note that Thompson. Eiler. and Forker
(1976) estimated the direct and carryover elasticity of generic milk
advertising in New York City to be 0.0212. Kinnucan (1983) estimated
long-run milk advertising elasticity to be about 0.12 in the Buffalo.
New York. market. (This is an average elasticity from five functional

forms Kinnucan used to estimate fluid milk demand.)

5.3 .EaiimaIBS_QI_LQ§1_Sale§

5.3.1.0uanI1Ix_E511moie&_oi_LQ§I_Sales

With Equation lc. monthly sales in the absence of the incident
were projected and compared to estimated actual sales. This reflects
sales lost due to consumer awareness of the contamination considering
changes in deflated milk and fruit nectar prices. income..and season-
ality. Projected and estimated actual sales are shown in Figure 5.2.
In the aggregate. from March 1982 through June 1983. the difference
between projected sales in the absence of the contamination and esti-
mated actual sales was about 39 million pounds of milk. To determine
the quantity of sales lost beyond the amount dumped. monthly estimates

of total sales loss were corrected for the amount dumped each month.

 

3Calculations were:

993 x W or -0.0149 x 5.55 = -0.0219

dNM fi .7

w
\l

91

.mmm_ aces
ummm. >cmscnoo .o0um5_umo ocm oOHOOHOLa "cowuasamcoo ou_aoo c0¢nn.~.m 0L:m_u

 

 

 

mmm. mmm_
z a >oz amm one ><z ¢<z
>4: m«: A _ _ . _ _

o

- a
s
c
a

- N
....
u
22:5sz 3.835;-.- - m a
22:5sz 8:228 I m
u
u
.. e z
s
a
a ‘00:. \s\ oo sin: I m D
,x ... .....--.--.... ........ ... .x .... .. a
I'lllllsss loans oII|I0\ D
._
- m a.

 

 

 

92

For example. the difference between projected and estimated March 1982
consumption was 3.617.860 pounds of milk. However. 3.155.170 pounds
were dumped. The difference was a loss of 462.690 pounds of milk that
was available but not used for Class I purposes. By inspection of
the utilization data (Figure 2.1). any uncontaminated milk not used as
Class I from March 1982 through February 1983 could have been used as
Class I if there were normal demand for it. Hence. the true sales 1055
can be accurately determined over this period.3 This is compared with
estimated losses in Table 5.2. (Zeros indicate months when milk dumped
was greater than estimates lossesJ

The model underestimated known sales. but did generally follow
the trend in sales loss. It correctly identified periods of no sales
1055 in May. October. and November because all available milk was used
for Class I utilization. Losses were underestimated in late spring-
early summer 1982. however. It was estimated that about 6.5 million
pounds of milk that would have been used as fresh fluid consumption
were not. due to awareness of the contamination. Using actual losses
from March 1982 through February 1983 instead of estimates. this figure

was about 8.4 million pounds.

 

3Beyond February. some of the milk not used as Class I may have
been used in other uses even had the incident never occurred. This
milk cannot be considered a cost from the contamination. Beginning in
March 1983. it appears that demand approached pre-contamination levels.
To determine how closely it did requires a more discriminating tech-
nique than just inspection. Hence the econometric model was devel-
oped.

Table 5.2.--Comparison of known and estimated sales 1055 (pounds).

93

 

 

Known Estimated
L055 L055
1982 MAR 1.396.888 462.690
APR 298.429 0
MAY 0 0
JUN 270.791 0
JUL 598.988 0
AUG 597.589 992.735
SEP 26.907 535.495
OCT 0 0
NOV 0 0
DEC 505.146 231.486
1983 JAN 264.726 325.027
FEB 877.757 346.342
February Subtotal 4.837.221 2.893.775
MAR ? 593.279
APR 1 851.822
MAY 7 1.073.750
JUN 1 1.064.680
Total 6.477.306

 

5.3.2E511mat1on_of_tne
1a].u.e_of_Loet_5alee

Assigning value to uncontaminated milk involves a complex
system of classified pricing. Assigning value to the milk sales lost
is likewise complex.4 If the milk had no other use. the loss because

consumers did not purchase it would be the full $21.09 per hundredweight

 

4Estimation of the effect of the incident on the average. or
blend price producers receive for their milk was not possible. Actual
monthly blend prices are available. but computation of projected prices
in the absence of the incident requires access to confidential informa-
tion the Division of Milk Control would not release.

94

producers could have received for Class I milk. However. at a
minimum. producers received $2.92 per hundredweight for milk that had
been salvaged. Alternatively. milk not used as Class I could have been
used as Class II (producer price: about $13.50 per hundredweight) or
for Class I-A or export (worth $18577 per hundredweight). Hence. the
loss to Oahu producers through June 1983 could have ranged from about
$195.000 to $0.530.000 assuming. respectively. all the milk was used as
Class I-A or salvaged.

More precise estimates were calculated. Actual sales 1055 from
March 1982 through February 1983 was determined by multiplying the
amount of milk used in each non-Class I use by the difference between
the Class I and each alternative use price. Beyond February. the
estimated loss was calculated to determine how much milk not used as
Class I would have been in the absence of the contamination. The
amount of milk salvaged up to this limit of the sales 1055 was multi-
plied by the monthly Class I-salvage price differentiaL. If less milk
were salvaged than sales lost. then the amount of Class II milk was
added up to the limit of the sales 1055 and multiplied by the Class I-
Class II differential.

For example. given the June sales 1055 was 1.064.680 pounds.
then all 1.004.506 pounds of milk salvaged was valued at the Class I-
salvage price differential for a value of $180.510. An additional
60.174 pounds (1.064.680 - 1.004.506) from Class II use would also have
gone to fluid consumption if the incident never occurred. Thus. an

additional $4.591 was lost (60.174 pounds x Class I-Class II price

95

differential [$0.07631). Monthly sales losses are shown in Table
5.3. Total sales 1055 due to consumer awareness of the incident from

March 1982 through June 1983 was about $626.000.

5.4 .Meo1a_Eindln9§
5.4.1.EIieQI_Q£_All_Medlo_Qn_Soles
To determine if any mention of milk adversely affected
sales (e.g.. by reminding the public of the current or past problem).
total media coverage (TM) of milk. the contamination incident. and
Safeway's bid to enter the market replaced negative media coverage in

the conditional regression. Equation 1c. The results were:

(1d) Q,= 6.41 - 3.77 DPM + 0.155 DPFN - 0.00618 TIME + 0.308 JA
(1.20)*** (1.70)** (0.787) (0.00184)*** (0.125)**

+ 0.283 FE + 0.183 MA + 0.534 AP + 0.539 MY - 0.273 JE
(0.124)** (0.128) (0.129)*** (0.132)*** (0.131)**

- 0.303 JY - 0.154 AU + 0.514 SE + 0.402 CC + 0.203 N0
(0.130)** (0.132) (0.138)*** (0.130)*** (0.129)

- 0.294 DV + 0.0702 DVTRND - 2.95 x 10'7 DUMP
(0.399) (0.0326) (9.31 x 10'8)***

- 0.00665 TM - 0.00268 TM(-1) + 0.000311 DPCPI
(0.00121)*** (0.00171)

? = 0.937 np
0.268 (from Renaud. 1971)

D.W. = 2.22 ny

Standard error of regression = 0.222
Mean of the dependent variable = 3.77

*Significant at the a g 0.10 level.
**Significant at the a 5 0.05 level.

***Significant at the a g 0.01 level.

96

 

 

Nm_.m~ m mmm.~mm was. .u_nm
~_N.m ~m~0.o -m_.o mmo.w_s <-. .mn_e
oss.m_ . .mao.o awn—.0 emm.wa_ __ nan_u
. umomo<
~em.os m mmm.mmm m.8. ne_nm
mmm.m Nmeo.o a~m_.o mo~ sue <-_ nmn_u
mm NMNo.o aum_.o mem.~ utoaxm
m_m.om m_m_.o mmNo.o ~m~.o~_ seeds—am
>_ss
mmo.__ m _m~.ou~ nae. ma_nm
mm5.m ~m~o.o nam_.o m_o.am~ <-_ nan_e
am_.e anm_.o ~m~o.o Naa.mm eases—am
econ
o o o ..o. au_nm
>mz
s~m.m w m~m.mm~ awe. .a_am
sum.e ~m~e.o amm_.o mNe me <-_ .nn_e
__.a<
_oo.mo_m www.mmm._ ..o. nu_nm
a. ~m~0.o a~m_.o use <-_ ..a_u
m~_.m~ ~_m_.o ~mno.o amm.m_~ __ ann_u
oem.sa mmao.o e~e_.o e~o.~m_._ eema>.aw
~wm. seen:
A.a_\mv A.a_\mv
mHW~M3uMW4 _o_uc0c0mm_o ou_ca um: quumme x__z mo Om: o>_umcc0u_<
ou_cm o>_umco0u_<

 

.mOmmo_ mo_om >_;ucozu-.m.m o_nmb

97

 

soa.a~ m ama.aam ..o. nu_am
m_m.s_ ~m~o.o -m_.o Ram e_a <-. nan—u
mem.~ Nmeo.o nnm_.o asa.m~. oceaxu
-s.o_ ooao.o men—.0 mm_.am_ __ man—u
>coacnou
~e_.e w m~a.em~ n.o_ aa_nm
~s_.m «mmo.o naw_.o mNN so~ <-_ man—u
mam. >omacmw
.No.os m mw_.mom awe. nu_nm
mme.~ ~m~0.o n~w_.o omm o~m <.. men—u
mm_.mm wm~_.o ..mo.o omm.sm. auma>_am
LOQEOUUQ
o o o mmO. mo_mm coneo>oz
o o o mmo_ mo_mm concuoo
sue a aom.m~ and. me_am
eNe Nmuo.o amm_.o mom mm 4-. .mn_a
cooEOHQOm
A.a_\mv A.a_\wv
mHn—umwswwmg —m_uc0._0n_::o OUT—m Om: MMWHUMMW x22 $0 0m: 0>mumcgouz<
OUTBQ O>_umcL0u—<

 

.ease_eaou--.m.m n.5ac

 

98

 

 

mmw.m~mm th.o~u.m mu_en one. .o
u:.m> .ouou oeums.umm
_o_.mm_w omm.wmo._ and. an... eaens_u.u
.mm.s mono.o osm_.o em. as __ nme_e
o_m.om_ ~m~_.o ~.mo.o mom.soo._ eaen>_nm
0:33
Nmm.mm w om~.mmo._ nno. no.8. enume.onu
m_o.e ~m~o.o new..o oo~.mm~ <-_ .aa_u
o__._ ~m~o.o mnm_.o m~m.as 0.6axm
meo.me maao.o oem_.o mm_.eem __ ann_a
.om.mm new..o meso.o omm.~o~ eama>_am
>oz
omm.~a . Nam._mm ..o. .8... eaeae.ama
mm... «muo.o -m_.o mm: o. <-_ ..e_e
.eo ~m~o.o -m_.o ~mo.a~ utoaxm
ma~.ms mm.o.o .mm_.o mmm.o_o _. ..a.e
om~.s~ ame_.o ~_so.o ~a_.me_ eaen>_nm
...a<
om:.mm w mnm.mmm mmO. mo.om ocuoE.umm
~mm.m wmuo.o mum—.0 .mw.o.: <1. «mo—u
m_~.. NMNo.o nua..o _-.~m etoaxw
ssm.m mmno.o .mm_.o m~m.mm __ nmm_u
_o~.m mmo_.o ..eo.o .me.om euen>_am
cute:
mo_mm umo. .mmwmwmwww.a OMMMM\MW3 Amo:30av ...: mo om: o>.umcc0u_<
mo o:_m> cu... . o>.uoco0u_< >u.ucmad .

 

.easa_ucoa--.m.m u_ane

99

Price elasticity is high but still acceptable. The coefficient of TM
lagged one month was negative and "statistically suggestivefi' The
coefficient of DV became negative while that of DVTRND became positive
though both were very insignificant. The direct and carryover effect
on consumption of all media coverage related to the incident. whether
negative. positive. or neutral. was -0.00933. Evaluated at the mean.
this yielded a total media elasticity of -0.0351. indicating a l per-
cent rise in any media attention given to the milk issue reduced
consumption about 0.04 percent. Based upon the comparative el astici-
ties between negative media (-0.0219) and total media (-0.0351) it was
suggested that any publicity of milk at all reduced consumption more
than just bad publicity. although a statistical test implied no

significant difference between them.5

 

5Since the variances of the regressions 1c and 1d were
insignificantly different. the elasticities were statistically
compared. A major assumption was that the respective ratios of mean
negative and total media to the mean of the dependent variable were
known with no variance. Adapted from Kmenta (1981. p. 372). the test
statistic used was:
A - B ~ t

sA-B N-K

 

where . ___
A = (z 8”“ x MM/fi)

B = (2 8T" x TM7EC)
and sA_B' [WE/61212 Van-(2 8N") + (ii/1:012 Var(z 8m) - 2 (Til/‘60)
_ A A &
(TM/EC) Cov(z 8N". 2 6mm
It was assumed that the covariance was zero although relaxing this

would increase the t-statistic. The assumption that N-M/Tfi and film
are known with no variance is more problematic.

100

However. because the signs of the DV and DVTRND coefficients
changed with the incorporation of a one-month lag on TM. negative and
total media elasticities were calculated from models in which neither
variable was lagged. Dropping lagged NM did not substantially change
Equation 1c. Dropping lagged TM yielded expected signs in the coeffi-
cients of DV and DVTRND. The total media elasticity O4L0241) was
found to be significantly different from that of negative media

5.4.2 IrendsJDJediLQoxerage

Surprisingly. the pattern of media coverage was not as
expected. Rather than negative media (NM) peaking and slowly falling as
positive media (PM) rose. NM fluctuated over the contamination period
(Figure 5.3). After the initial media attention subsided. new facts
emerged which renewed concern and confusion over milk safety. After
this second peak. a second. lesser contamination incident occurred in
December (antibiotics in milk). ‘This passed as the Safeway controversy
emerged. ‘Throughout the period. media sensitivity to any contamination
was heightened. Discovery of heptachlor in wild game was reported. as
were traces of DDT in the island milk supply (which. health department
officials explained. was not without precedentL Positive media cover-
age did not rise slowly as expected. but instead followed the pattern
of NM. After the heptachlor contamination announcement. government
officials rushed to reassure the public of the safety of milk. After

subsequent negative information about mi1k safety. government officials

lOl

.mmm. ocss
(Nam. >co3cnon .ommco>oo m.oos o>.u.moa ocm o>.ummoz--.m.m ucam.d

mam. «mm.
><z ¢<z z<1 >oz mum 4:5 ><z ¢<z
_

  

 

 

    

 

 

a
cu
m
Dc 0
u
a
cm 2
o
a
u.
on w
b
..
cuomz u>_h_mom.sl. no.
cyan: u>~p¢owz.lll
ou—

 

 

102

would again counteract rt Hence as negative media coverage fluctu-
ated. so did that of positive media. Little attention was given the
May 1983 announcement that all dairy herds were virtually cleared of
heptachlor.

The components of milk safety (milk quality. level of govern-
ment protection from milk contamination. and processor integrity) were
also examined. ‘The measure of processor integrity revealed little
since both processors were rarely criticized. It was conservatively
assumed that negative information about one processor would only cause
consumers to switch milk brands, so only when both processors were
criticized did this measure assume non-zero values. Dairymen were
rarely criticized for producing a contaminated product. Negative cov-
erage of milk quality was generally greater than negative coverage of
the level of government protection (Figure 5.4%. Certainly government
investigations kept the matter alive in the press. but they were not
the only factor. New facts about duration of exposure. a second
contamination incident. and efforts by Safeway to import milk from
California reminded the public of health risks associated with Oahu

milk.

5.5 QemooraohJLELnoinoa

Surveys conducted after the initial contamination announcement
reveal how different segments of the population reacted to it. One
survey was conducted for Foremost three weeks after the first recall.
but before the last two (Table 5.4). More than 99.5 percent of those

randomly sampled were aware of the contamination. A small majority of

103

.mmm. ocsw
(Nam. >cmsonou .ommco>00 o.ooe o>.ummoc so mucocoaeouuu.:.m o.:m.u

mam. «wa—
>mz ¢<z z<o >oz mum 4:5 ><z «<2

 

 

 

 

 

 

 

IOIIIOIIIO\ Q o
1 cu
u at
u on
000’.
u on
zo~humbo¢m hzuzz¢u>oc zoom.2:.
>h~4¢30 :4“: zoom.lll . oo—

 

 

DN—

CI—F—UZI—HOZ ”DOME!

104

.ucuOLOQ _ so u_o;-uco coca anode

.uc. .mo.c.oo unoEULou

”ouc30m

 

 

 

 

 

 

.oSN. .Na. .N... ._o_. .mn_. .mm. ..o.. .mm. .aa. .mm. ..m.. .N... .Nmm. .u.asa. .eaes. "ones
a - - - - - - - - - - - - 36:. u.c6e
- - - _ - - - _ - N - - . >u_u._asa eta»; e.=u>e:
.s Nm an mm mm as N: as an on .5 mm as use.
mm as no cm e: am mm .m No me an as om ano—
n 1 .- n .. ... .. I a I I I .. 0.9..

...:o: x..E >59
» u u u u u a u u u u a u

0.928. n.4: .uauo .mwum -umwwa +om as-mm sm-m~ Lemma +m ~-_ and: .nue»
e_eao.se=

xom

>u.u.ccum

0m<

e. eu.e__;u

 

.mc.mo;uoan x_.E co >u.u..a:n mo uuomuu--.a.m o.omh

105

households without children did not alter their milk purchases. but a
larger share of households with children did. Across age groups. more
of the youngest shoppers reduced consumption than other groups. Unlike
the Schuker et al. (1983) saccharin study. the oldest age group did
alter milk consumption. It is unknown why a lesser percentage of those
in the 25-34-year-old group changed consumption than others. Perhaps
they represent childless individuals who consume little milk. Non-
Caucasians in Hawaii consume less milk than Caucasians. and Scott
(1967) found that non-Caucasians drink milk more for its nutritional
value than for its taste. Hence one would expect more non-Caucasi ans
to reduce consumption than Caucasians. Indeed. Table 5.4 shows the
change in consumption more among Orientals and others (about 70 percent
of the population) than among Caucasians. Finally. the shift was
greater among females than males. This may understate a later shift in
purchases since the telephone survey began the day after the presence
of heptachlor in breast milk was revealed.

The same survey also asked if normal whole milk purchasing
would be resumed after the incident (Table 5.5). Again. a greater per-
centage of males and those without children indicated they would resume
normal purchases than females and those with children. Interestingly.
a higher percentage of those in the oldest age group responded affirma-
tively than among other ages. This suggests the older age group is
more likely to return to pro-contamination habits (i.e.. habits estab-
lished over their lifetime are less easily broken than those of younger

people). Also. a larger percentage of Orientals. who consume milk more

l()6

.uc. .no.c.oa unCEOcou "cocoon

 

Ax..e 0.0:: «mop >35

 

 

 

 

 

.o~_. .es. .... .... .nm. .an. .... .as. .QN. .an. .... .... .oa_. so: as: .eentoa. .u..a
m m o N o a a a a a m a a see. u.=oe
m. o. : 2 ..N o .N a mu ... - o o. «no. .6. a :5 .2
.. - m. ~ 3 m m _. .. o m m a an». o.uu.. - >aa .0:
me me as no as .m ~e a. .8 o. no me .a no»
2:... 0.2:
guns no mc.>:a Cu sumo
om 50> v.30: .ou>.om
0L0: nee—no.3 x..E C.
u u u u u u u u a u u u »
u_.eau ...: .uaae .muuu -uuhuu +em ms-mn sm-m~ .mwws +n N.. «as: .aao
»
xum >s_u_a;em ae< e_oguase:

a. ao.e..gu

 

..msco: Ou ensue. v.50: nonozucan x..e 0.6:: c0:u0£3--.m.m o.aoh

107

for reasons of nutrition. would resume pre-contamination purchases than
Caucasians. A greater proportion of Caucasians than Orientals changing
purchases over the long-term may reflect a greater change in milk
consumption habits among Caucasians than Orientals. Perhaps the former
group perceived the incident as indicative of long-term problems with
contaminated milk supplies and poor protection of public health by the
government. Orientals may have perceived the incident with less con-
cern.

If some believed the government was responsible for poor
protection of public health. this may be reflected in political opin-
ions. Democratic Governor George Ariyoshi was re-elected nine months
after the initial recall. Although voting reflects opinions on a
variety of issues. the state's mishandling of the heptachlor contamina-
tion was apparently insufficient reason to remove the incumbent; Repub-
lican Governor William Milliken of Michigan was also re—elected after
the PBB incident. Indeed. three months before the elections a
political opinion poll was conducted for the.flono1u1u_Adxertiser in
which questions about milk consumption were also asked. Thirty-one
percent of those polled said the incident was "just one of those things
that happen these days." and 30 percent blamed the health department
for the crisis (in Keir. 8/27/82). On average across all political
parties. only 3 percent blamed the Governor. although 6 percent of the
Independents blamed him. Further. the survey showed more Independents
than Democrats or Republicans were still buying less milk than before

the recalls. This is not surprising since the Independent candidate

108

attempted to make heptachlor contamination a key election issue. ‘This
implies that a spokesman who is believed to be more credible than those

in government may influence consumer attitudes.

CHAPTER SIX

SUMMARY AND IMPLICATIONS

This research was undertaken with several objectives in mind.
One was to develop a methodology to estimate producers':sales loss due
to consumer awareness of food contamination. Second. the methodology
was applied to determine costs to Oahu dairymen following heptachlor
contamination of their herds. From this estimation and study of the

characteristics of the incident. policy implications were drawn.

6.1 Methodology

The demand model considering the impact of media coverage of
the incident. scope of the contamination. and a possible change in milk
consumption habits is less restricted than other models which examine
consumer response to food contamination. Following the approach of
Schulstad and Stoevener (1978) and Swartz and Strand (1981). another
model was estimated incorporating a negative media variable but
excluding factors accounting for the scope of the contamination and
possible changes in habits. Compared to known losses from March 1982
through February 1983. this model substantially underestimated monthly
sales 1055 in all but two months in which sales 1055 was greatly

overestimated. Consideration only of negative media coverage restricts

109

110

the model and hampers accurate estimation of sales 1055 following a
massive contamination incident.

The pervasiveness of the contamination is a factor that appar-
ently also affected sales since some consumers may have refrained from
consumption until the problem subsided. It was also suspected that
repeatedly disposing of one's milk and dairy products or returning them
for refunds may have changed consumption habits. ‘The complete model
(Equation 1c) accounted for these factors and yielded better estimates
of sales 1055. Variables accounting for a change in habits were sta-
tistically insignificant but one was "suggestiveJ'

This complete model is most appropriate for study of massive
contamination incidents but is also applicable for smaller occurrences.
In these instances. the dumping variable. the dummy variable. and the
DVTRND variable may drop out of the model if little product is dumped
and if habits do not substantially change. One is then left with a
model similar to those of Schulstad and Stoevener and Swartz and
Strand.

The model developed is not without weaknesses. Although it
followed the trend in monthly lost sales better than the restricted
model. known losses through February 1983 were still underestimated. so
estimated losses calculated through June 1983 may also be an underesti-
mate. A key assumption of this method is that consumer awareness of
the contamination and not the contamination itself leads to sales 1055.
Awareness of such problems (34}. antibiotic residues in beef products)

may depend on the different attention given by different media. and

111

reaction may be influenced by other factors. The Oahu incident was
very intense. and the assumption of similar newspaper and other media
coverage was reasonable. For other incidents. this assumption may be
inappropriate. especially if people gain awareness through one medium
which gives the issue different prominence than others. Further. the
Oahu incident involved a more immediate threat stemming from the milk
in one's refrigerator. Other contamination scares. such as high EDB
residues in some bakery goods. are less direct and hence less attention
may be paid them. The studies cited above share these weaknesses
however.

This methodology is useful to several groups involved in
contamination incidents. At least in this occurrence. government
officials declined to impose fines on Meadow Gold for marketing
contaminated milk in December 1982 because they felt the company had
suffered enough in terms of lost sales. This methodology enables
officials to weigh such choices by determining sales 1055 to violators
of food safety regulations. Producers of the contaminated product will
wish to»quantify their sales losses when they present the economic
impact of the incident to government agencies which may provide finan-
cial aid. Producers whose feed is contaminated by the supplier. or
competent. ethical producers of a quality product who suffer because
their fellow producers market an adulterated product may use this
methodology in lawsuits to recoup the cost of lost sales from those

responsible.

112

6.2 EstimatooLLosees

The magnitude of fresh fluid milk sales 1055 due to consumer
awareness of milk contamination may be approached from several perspec-
tives. It is important to note that these losses are only a portion of
total producer costs which include product and feed dumped. laboratory
testing. veterinarian and legal fees. higher interest and labor costs.
and in cases such as PBB. animal mortality and extra holding costs.
(These losses in turn are only part of the total costs to retailers.
consumers. and government agencies who are affected by a contamination
incident.) It is known that Oahu dairymen received $8.551.515 from the
federal Dairy Indemnification Program for milk dumped (USDA. DIP. 1983
8 1984). Thus the $626.000 sales 1055 due to consumer awareness is an
additional 7 percent of the dumping 1055. Clearly. dumping was the
greater cost. but once the contamination became known. dumping was
inevitable; losses due to consumer awareness could have been minimized
by more effective government and industry response to the incident.
Further. since dairymen were paid for their dumped milk. the sales 1055
became a greater share of their total cost from the incident. By
comparing sales 1055 to actual marketings over the period. it is seen
that marketings could have been 3 percent higher than they actually
were. As noted earlier. Swartz and Strand (1981) estimated that losses
to consumers and producers due to consumer awareness of kepone contami-
nation of oysters were 5 percent of actual marketings. Since the study
period ended before sales had returned to pre-contamination levels.
total sales loss to Oahu dairymen is surely greater than estimated

here.

113

It is worthwhile to compare estimated sales 1055 to fines that
could have been levied for producing a contaminated product. Before
the incident. the maximum fine for violating Hawaiian milk safety
regulations was $500 per violation. In response to the incident. the
maximum fine for violators of State food safety regulations was
increased to $10.000. If all 16 Oahu dairy operations were fined.
penalties would still be only about 25 percent of the sales 1055.
Hence. fear of lost sales in this instance would far outweigh fear of
being fined for marketing contaminated milk.

Although $626.000 is minor compared to the amount of milk
dumped. this was about $39.000 per producer or about $2.400 per
producer per month. Hence the cost in terms of lost sales was still
substantial to each producer. This cost also reflects the amount a
producer would be willing to pay to avoid such a cost. If milk testing
for heptachlor in cows! milk cost the same as testing in mothers'lnilk
($150 per sample). each producer could have spent $2.250 per month to

have milk tested eyery.other.day to avoid a loss from lost sales.

6.3 GrummanoLoLMasoiyLEood
ContaminatJoLInoidents

Government officials. producers of the contaminated product
whether directly affected or not. producers of substitutes and
complements. and consumers all benefit by knowing what to expect from
these incidents. From Michigan's PBB incident and heptachlor contami-
nation of Oahu milk. characteristics of these incidents become evident.

First. costly decisions by industry and government officials are made

114

in an environment of great uncertainty. Without prior studies. scien-
tists cannot give definite counsel to decision-makers who attempt to
balance protection of public health with avoiding costs to the industry
involved. Second. the fear of lost sales to the company or industry
paralyzes decision-makers precisely at the time swift action is needed.
Hesitancy to act was seen in both the P88 and heptachlor cases where
violative samples were not acted upon because they were "unofficialJ‘

Consumers also operate in an environment of uncertainty. In
both incidents. scientists presented contradictory opinions and
evidence about the health effects of the contaminant. Government
officials rushed to reassure the public that the problem was isolated.
that health risks were minimal. and that the situation was under con-
trol. Subsequent events belied these assurances. In both instances
the media shared responsibility for consumer confusion with headlines
such as "Bad Mi1k Called Harmless" (Honolu1n_§1an:8ulletin. 3/20/82)
and "Tainted Food Is Safe. FDA Reports" (in Chen. 1979).

As seen from media coverage of both incidents. publicity does
not quickly diminish. After the initial peak of publicity. new
aspects of the contamination are revealed and media coverage rises.
This may be due to scientific studies about the effect of the contami-
nant or information about the length of exposure. Government investi-
gations keep the matter alive. but heightened media attention about
contamination of any kind. fueled by other contamination problems.

protracts consumer awareness as well.

115

6.4 Was
From a brief comparison with the 1982 Tylenol incident.

policy implications may be drawn for industry and government officials.
Seven people died after using cyanide-laced Extra-Strength Tylenol. a
pain reliever. Johnson and Johnson. the parent company of Tylenol's
manufacturer. voluntarily recalled the product even though the manufac-
turer did not seem to be responsible for the poisonings. Johnson and
Johnson also offered a $100.000 reward for information about the killer
and cooperated with authorities investigating the incident.

Tylenol's market share dropped from 35.4 percent to 7 percent
as the company recalled products at a cost of $100 million ("Tylenol."
10/2/83). Polls indicated that about half of Tylenol's users thought
it unlikely they would use the product again and sales were down 80
percent (Waldholz. 10/29/82). After the heptachlor incident on Oahu.

a survey conducted for Foremost indicated that 24 percent of those
sampled would buy less milk after the contamination crisis passed. and
whole and 2 percent milk purchases were off by about 75 percent. Con-
cerning government's influence on response. one market analyst said:
Federal Drug Administration's [sic] statements about Tylenol and
the companyts steps to recall Extra-Strength Tylenol "could have a
fairly negative psychological impact on the product in the minds of
consumers until the FJLA. tells them that it is all right to use
it again. If the uncertainty goes on for weeks or months. and the
FJLA. does not give the product a clean bill of health. there

could be fairly long-lasting damage to the product." (cited in
Pace. 10/2/82)

In industry-wide contamination incidents. there is no parent
company to act quickly to reassure and protect the public. That role

falls on government officials. The results of this study indicate that

116

government can best protect the long-term interests of the industry by
protecting the short-term interests of the consumer and recalling all
possibly contaminated product. Losses to the industry cannot be
avoided in massive contamination incidents. The question government
and industry officials must ask is not how to avoid but how to minimize
costs.

Amid uncertainty. officials attempt to reassure the public of
little or no health risk associated with the contaminant. However.
when officials retract their statements or are proven wrong. they lose
credibility precisely at the time they most need it to minimize pro-
ducer costs. As risk perceptions rise. the public looks for credible
sources of information. In both P88 and heptachlor incidents. indi-
viduals. whether qualified or not. stated higher health risks asso-
ciated with the contaminant than did government officials. Unless the
government is credible. the information provided by these individuals
will lead consumers to reduce consumption since negative information
has a greater impact than positive information. Hence negative infor-
mation freely supplied to consumers leads them to reduce consumption
more than it would have been reduced. If the government has lost
credibility. it has lost its effectiveness to counter this effect.

Mowen and Pollman (1982) showed that releasing the worst pos-
sible information first followed by less alarming releases resulted in
more favorable perceptions of the company than when information of
possible health risks grew increasingly worse. By maintaining credi-

bility and placing public safety foremost. government officials can

117

minimize the cost of lost sales to producers as well as the threat to
public health. One year after the cyanide poisoning deaths. Tylenol
was once again the nation's leading nonprescription pain killer. Six-
teen months after the initial milk recall. Oahu dairymen were threat-
ened with a complete loss of market by competition from Safeway.

Analysis of media coverage of the Oahu incident supports these
policy implications. Protecting public health by recalling all
contaminated product minimizes subsequent recalls. possible injury to
consumers. continued media coverage. and attention drawn to the matter.
Compared to the Thompson. Eiler. and Forker (1976) study of generic
milk advertising in New York City. negative media coverage appears to
have an effect of similar magnitude but opposite direction as generic
advertising. But rather than present positive information to counter
bad publicity. this study indicates that government and industry can
best minimize sales loss by removing the perceived health threat.
Disputing unfavorable test results only serves to maintain the
prominence of the incident and heighten uncertainty surrounding it.
Following a contamination incident. a statement such as. "This product
does not cause cancer." will draw attention to the possibility of a
product's carcinogenicity.

Emergency plans should be prepared before contamination
strikes a company or industry. Johnson and Johnson's decisive actions
were taken partly because an l8-month review of the company code of

ethics at all levels of management had just been completed. The

118

consensus behind the code helped executives make difficult decisions
(Seibert. 12/25/83). Quick action is expected of government agencies
especially. A delay in disclosure of contamination in the food supply
is a disincentive to later announce the problem because questions about
the delay are raised by the press. The longer an official delays in
announcing a problem. the more embarrassing it is to announce it. On
Oahu. reasons for the delay were part of the investigation of the
incident. A principal constraint frequently faced by government
agencies involved in these occurrences is a lack of laboratory
facilities. Plans for quickly expanding such capacity should be
developed. Specific recommendations for industry handling of product
recall are available in Fiske and Chandran (1975). McGuire (1974). and
Grocery Manufacturers of America (1974).

Market structure will influence the amount of lost sales.
Estimated sales 1055 for Oahu dairymen is low compared to producers of
contaminated products with many substitutes. After a man died of
botulism poisoning from Bon Vivant vichyssoise in 1971. the company
soon filed for bankruptcy (Kleinfiel d. 1982). Oahu consumers who grew
tired of fruit nectar or powdered milk had no alternative but to buy
island milk. If an undifferentiated product is contaminated. the cost
of lost sales will not be limited to those producers directly involved.
Even those whose products were never contaminated will suffer a sales
1055. Dairymen on the island of Hawaii suffered a decline in sales
even though their herds were not contaminated. Even with

differentiated products. other producers will be affected by the

119

contamination. There is conflicting information about the effects of
the 1982 Tylenol poisoning incident on manufacturers of substitute pain
relievers. One report indicated sales of competitors' products
declined ("Tylenol." 10/9/82) while another reported sales increases
(Kneale. 10/13/82).

Barriers to entry will affect sales 1055. Fresh milk from
other sources was not available given the processor duopoly and govern-
ment protection of the island dairy industry. Had an alternative fresh
milk source been available. the sales 1055 probably would have been
greater. (However. few Oahu consumers tried the small amount of milk
imported from the island of Hawaii. It too was apparently suspectJ
In the aftermath of Michigan's PBB incident. food products were simply
shipped in from surrounding states. and in some cases. clearly
advertised as being produced outside Michigan.

Not only will market structure affect consumer response. but
consumer response may also affect market structure. ‘The greatest cost
to Oahu producers was not lost sales. but a change in market structure.
In Hirschman's (1970) terms of exit and voice. Oahu consumers could not
easily exit the milk market. Hence they exercised their voice not to
pressure producers or processors but to pressure the State to allow
competitors into the market. Now that California milk has penetrated
the market. the survival of the local industry is threatened. The
incident precipitated questioning of the market structure and its
protection by consumers. competitors. and federal agencies. ‘This case

shows that consumer awareness of food contamination can bring about

120

more competition in an imperfectly competitive market and provide a
powerful producer incentive to maintain product quality even in a
protected market with limited substitutes.

Finally. this study shows that the market provides financial
incentives not to market an adulterated product. but can reliance be
placed on market incentives alone to enforce food safety regulations?
Effectiveness of the market to do so may be tempered by non—economic
factors. If. as some believe. negative information carries greater
weight than positive information because it is rare (see Weinberger.
Allen. 8 Dillon. 1981). the effectiveness of bad publicity on sales of
a contaminated product may diminish as more contamination is reported
(the "Cancer-of-the-Month" syndrome). As discussed in Chapter Four.
response to information about contamination will be affected by the
credibility of the source and the predisposition. self-confidence. and
risk perceptions of the consumer. which in turn may be influenced by
the perceived degree of health risk (e.g.. death from botulism).
government protection of public health and producer integrity. Charac-
teristics of the populace involved will affect response. Different age
and ethnic groups. those of different household compositions. and men
and women responded differently to the Oahu incident.

A major problem with relying solely on the market to enforce
food safety regulations is that the market punishes indiscriminately
and may not hold those responsible for the contamination fully
accountable. This study shows that each Oahu dairyman lost milk sales

as a result of consumer awareness of the contamination. Yet consumer

121

confidence in one's undifferentiated product is a collective good. The
benefit of acting to ensure consumer confidence cannot accrue solely to
the producer who takes action to maintain such confidence. Indeed.
despite efforts by some producers to maintain product quality. they
may still bear the cost created by another producer who markets a
contaminated product. Likewise. the costs of marketing a contaminated
product do not accrue to the guilty parties. Hence. producers have
incentives to reduce their costs by marketing products of questionable
quality and letting the industry as a whole bear the cost. Unless a
method is found to impose the cost of marketing a contaminated product
on those responsible. market forces alone will not assure consumers a

safe food supply.

6.5 EurtneLBesearoh

Much remains to be learned from heptachlor contamination of
Oahu milk. ‘The period of observation of this study was too short to
determine when milk consumption reached pre-contamination levels. Data
indicated that sales remained depressed in the summer of 1983 but may
have recovered by January 1984.

The existence of thresholds of awareness and risk and their
effects on consumer behavior changes is important. If consumers cannot
find adequate substitutes for a contaminated product. they may eventu-
ally resume normal purchases of it. If a threshold is crossed though.
they may go to great lengths to search for or create new substitutes
(e.g.. innovations by some with soy and goats' milk were publicized on

Oahu). or they may try to change market structure to increase their

122

options. Had the industry not lost the confidence of consumers and
thresholds not been crossed. these courses of action may not have been
explored. How much awareness of product contamination is necessary to
change consumption was not studied here. It is believed a threshold
may have been quickly crossed given four recalls in less than two
weeks.

The distributional impacts of this study are many. Of particu-
lar interest to producers is to determine a shift in market shares
between Meadow Gold. which received much unfavorable publicity. and
Foremost. Likewise. there may have been a shift in consumption away
from whole to low-fat milk. The extent of this is unknown. Spill-
over losses to dairymen on the island of Hawaii occurred in response to
the Oahu incident even though Hawaiian cows were never contaminated
with heptachlor. Determining the extent and duration of losses to them
requires price data from the major retailers on the island since no
retail price survey is conducted. A problem with such a study is
measuring consumer awareness of the contamination since television
broadcasts originate on Oahu. but the Hilo. Hawaii. newspaper informed
island residents that the contamination was limited to Oahu. Consumers
may have received mixed information.

A related topic worthy of investigation is the incident's
effect on sales of substitutes and complements. Knowing the benefits
that accrued to Real Fresh. Inc.. which sold sterilized canned whole
milk on Oahu. and to producers of other substitutes would help

understand the ripple effect of these incidents through the market.

123

This is particularly interesting in the Oahu incident because the dairy
processors also produced fruit nectar. the leading substitute to

milk. Another cost stemming from food contamination is the cost to
producers of complements of the contaminated product. It is unknown
how sales of breakfast cereals were affected.

This study of consumer response to food contamination quanti-
fies the sales 1055 from a contamination incident. It indicates that
producers do suffer from consumer awareness of the problem which should
provide them with incentive to observe food safety regulations. This
incentive may be inadequate for market forces alone to enforce food
safety regulations unless producers can fully capture the benefits of
observance of such regulations or fully pay the costs of their viola-
tion. Several factors may influence the sales 1055. including market
structure and government action (or lack of it) to minimize public
health risks. Product contamination may affect a change in market
structure.1naking imperfectly competitive markets more competitive.
Attempts to counter bad publicity stemming from the contamination may
not help sales recovery. Future research should include an examination
of threshold levels of awareness needed to change purchasing behavior
and study of the distributional aspects of these incidents. Producer
awareness of market incentives to observe food safety regulations in

conjunction with government inspection will help ensure food safety.

APPENDICES

12')

APPENDIX ONE

THE DATA

125

APPENDIX ONE

THE DATA

Confirmation

Monthly data on Oahu milk production and utilization are
available only for aggregate forms of utilization. .After the contami-
nation. Class I utilizatigngas corrected by several factors. First.

\/the quantity of imports. recorded by the Division of Milk Control. was
~——.,—_——_—.....————-—~

added. Second. route returns (milk returned from stores and othe:

‘—

 

 

outlets) were estimated from data provided by one processor. It was

W,

._.._-—~—_.__.M v—._

assumed route returns reflegtgthejmgrkethshareshgfueachmprgcessog.
Hence Meadow Gold with 59.5 percent of the pre—contamination market
accounted for the same percentage of route returns. Foremost accounted
for 40.5 percent. In those months when estimated returns exceeded the
allowance for plant shrinkage and route returns. as set by the Division
of Milk Control. Class I utilization was decreased by the excess
returns. .

Schools were supplied with imitation milk in the fall to avoid
retail fresh milk shortages. To properly reflect quantity demanded.
estimated school consumption of fresh milk. had supplies been ample.
was added to utilization data. Schools usually received 7.000 gallons

of milk a day (Harpham. 9/25/80). The numbers of school days (excluding

holidays) in which imitation milk was served were 3 in September.

126

127

20 in October. 18 in November. and 13 in December. Calculation of

total fresh fluid milk consumption is shown in Table A.1.

Emulation

On a given day. about 10 percent of the people in Hawaii are
tourists. Hence. dividing consumption by resident population yields
incorrect per capita figures. Tburist numbers vary with peaks in July
and August. and a lesser peak in March. ‘To estimate the number of

' tourists on Oahu by month. the number of westbound overnight and_longer

-——_...-—

 

 

visitors to Oahu and their average length of_stay on Oahuflwerewootained'

“...-r

 

_._.—._.__ WM ""

W-
; __—-.._e-—-—-'

from the Hawaii Visitors Bureau. These data reflect the number of

...n .- ‘Méw

 

...-o-fl—b
—-u.--
...—-

tourists from North America (about three-quarters of all tourists)
arriving on Oahu and staying at least overnight. The average length of
stay on Oahu for visitors from Asia or Oceania was unavailable. so the
number of such tourists was not considered. It was assumed such
tourists are not heavy milk drinkers. ‘To calculate the average number
of tourists on Oahu each month. the following equations from the Hawaii
Visitors Bureau were employed:

Number of Visitors per Month x Average Length of Stay (days) =
Number of Visitor-Days per Month

Number of Visitor-Days per Month 9 Days per Month =
Average Daily Visitor Census

i,//This was then added to monthly Oahu resident population. ghich was

 

interpolated from annual data.

 

 

128

moan
Retail rices are sampled weekly on Oahu. E92511_£ol_data_wer -

 

 

 

V ' .— “>—

Agriculture: and QWWE

”M ”NH-“‘4

weeklyi'Retail Food Price Guide. Monthly prices were calculated from

the weekly surveys .EI eight stores for half-gallons of whole and Lilled

Wm- NW... --4—-.r—--——.._ (“hf—tn“ -v—-.-- ..._,_._.A.-.———-¢——.,_'_r “H"HNMWF—m—T

 

 

 

 

 

 

milk. and fruit nectar. a l4:quart package of nonfat dry_ milk. 46

. .-..~—--" M...
...-.4 _‘-.—- fl‘— ...... L._,__.-.

....._._-

ounces of canned fruit drink. and 12 ounces of canned soda. Often.

Wm-.—-.———-———t W .. ‘7

 

 

- *‘PH-v— .— q—‘_._,—'--—-_HM.J ~_—-

My

only one observation was available_per month; when there were more. an

Mu...“ ..

 

—_.._-.... u

 

average_wasmused. 6925-19.9939 were filed by interpolation.

Gaps were minor except for nonfat dry milk and fruit drink
when for 19 months (March 1979-October 1980) prices of an eight-quart
package of nonfat dry milk and 12 ounces of fruit drink were surveyed.
For nonfat dry milk. the 14-quart non-sale price before the interrup-
tion in the series was compared to the eight-quart non-sale price at
the beginning of the gap. An index was calculated and the eight-quart
price was adjusted to approximate the l4-quart price. A five-month gap
remained for which prices were interpolated. Similarly for fruit
drink. the 46-ounce non-sale price in October 1980 was compared with
the non-sale 12-ounce price in the same month to devise an index by
which to adjust the lZ-ounce price.

Estimating the retail price of fruit nectar over a 20-month gap
(February 1975-0ctober 1980) was more difficult. The deflated price of

fruit nectar (DPFN) was regressed on the fruit and vegetable component

of the Honolulu CPI (CPIFV). Regression results were:

129

DPFN = 0.480 + 0.280 CPIFV
(0.0618) (0.0224)

T22 = 0.735 0.11. = 1.266

Figures in parentheses are standard errors.

Although serial correlation was present. the estimators are unbiased.
This equation was then used to estimate fruit nectar prices over the
period of missing observations.

It should be noted that the Honolulu CPI is not seasonally

 

adjusted. The Bureau of Labor Statistics does not adjust it because it

m ...me run—m- 3' """'"

is based on a small sample. so correcting for seasonality may introduce

 

 

as much error as not adjusting it.

130

 

 

mo~.oom.o o om:.~m msh.mmm.o 2:3
:am.~om.m o mm~.m: mnm.o_s.m ><z
m:_.m~o.w o o m:_.m~o.w ¢a<
N~_.oss.m o o N~_.o::.m ¢<z
mam.oom.m o o mmm.oom.e mud
wam.~:o.~ o o wsm.~:o.~ z<s mam.
Nmo.mm_.~ oom.~m~ omo.mm o ~oo.mmm.o own
mum._mo.~ oom.mmo._ o oo_.~_ mm:.o_o.o >oz
.ms.s:n.~ ooo.:o~._ om:.m_ mmo._: smo.me_.o Poo
mom.~mm.o oom.om_ o o moo.~o:.o mum
mmo.mmo.o o o mmo.mmo.w u=<
moo.m_m.m o mom.m mom.-m.m 4:5
mgm.~:m.m s__.mm mno.o_ _sm.~_m.m 23a
”mo.m__.s -_._o o “.m.~mo.s ><x
nmm.oe:._ Nom.mo. _:N.ms mom.:mm._ ¢a<
osm.mas.s omm.o~ o o-.-m.: «<2
voucmeoo >_ao:m mccauox

>u_ucm:d .oo;um mucous. ouaoz co_um~___u:

noume_umu __mu mmouxm _ awn—u

 

.mmm_ onus-Nam. coco: .Amocaoov co_uaE:mcoo ooumamo<uu.e;‘o_nmh

131

Table A.2,--Consumption. calendar composition adjustnent factors. and

milk dumped.

 

 

Fluid Milk Calendar Milk
Consumption 3 Composition Dumped a
(Oz./Person/Day) Adjustment Factor (Pounds)
1977 JAN 5.97342 0.9837 0
FEB 5.80994 1.0000 0
MAR 5.83109 1.0077 0
APR 5.56503 1.0162 0
MAY 5.90165 0.9790 0
JUN 4.83096 1.0055 0
JUL 4.93536 0.9844 0
AUG 4.92175 1.0090 0
SEP 6.07578 1.0157 0
OCT 5.86075 0.9837 0
NOV 5.90973 1.0011 0
DEC 5.12989 1.0223 0
1978 JAN 5.85154 0.9823 0
FEB 5.58569 1.0000 0
MAR 5.64589 1.0129 0
APR 6.17028 0.9751 0
MAY 5.85454 1.0099 0
JUN 5.03486 1.0147 0
JUL 5.03781 0.9832 0
AUG 5.07947 1.0094 0
SEP 5.94753 1.0126 0
OCT 5.93240 0.9823 0
NOV 5.73423 1.0057 0
DEC 5.63509 0.9833 0
1979 JAN 5.67742 1.0087 0
FEB 5.83686 1.0000 0
MAR 4.93947 1.2010 0
APR 6.13927 0.9771 0
MAY 5.84254 1.0098 0
JUN 5.32830 1.0128 0
JUL 5.16711 0.9839 0
AUG 5.10002 1.0122 0
SEP 5.75965 0.9743 0
OCT 5.48121 1.0087 0
NOV 5.13706 1.0167 0
DEC 5.70715 0.9817 0

Table A.2.--Cont1nued.

132

 

 

Fluid Milk Calendar Milk
Consumption a Composition Dumped a
(Oz./Person/Day) Adjustment Factor (Pounds)
1980 JAN 5.71371 1.0064 0
FEB 5.72876 1.0485 0
MAR 5.87161 0.9821 0
APR 5.92405 1.0022 0
MAY 5.73991 1.0187 0
JUN 5.21007 0.9785 0
JUL 4.94494 1.0064 0
AUG 5.16567 0.9861 0
SEP 5.60939 1.0118 0
OCT 5.56231 1.0140 0
NOV 5.61365 0.9737 0
DEC 5.06627 1.0096 0
1981 JAN 5.45618 1.0187 0
FEB 5.58524 1.0000 0
MAR 5.25543 0.9831 0
APR 5.69500 1.0025 0
MAY 5.69917 0.9861 0
JUN 4.78825 1.0118 0
JUL 4.89074 1.0140 0
AUG 4.96829 0.9821 0
SEP 5.52542 1.0022 0
OCT 5.27889 1.0187 0
NOV 5.20056 0.9785 0
DEC 4.88222 1.0064 0
1982 JAN 5.13185 0.9840 0
FEB 5.08804 1.0000 0
MAR 2.59754 1.0119 3.155.170
APR 0.88405 1.0153 8.019.060
MAY 2.50146 0.9822 5.938.650
JUN 3.22787 1.0031 3.590.020
JUL 3.39667 1.0178 2.825.120
AUG 3.63138 0.9846 3.137.140
SEP 4.07963 1.0018 2.009.870
OCT 4.42498 0.9840 2.666.900
NOV 4.26711 1.0148 1.958.750
DEC 4.20177 1.0124 1.511.190

133

Table A. 2. "Cont i nued .

 

 

Fluid Milk Calendar Milk

Consumption Composition Dumped
(Oz./Person/Day)a Adjustment Factorb (Pounds)a
1983 JAN 4.61417 0.9822 1.165.160
FEB 4.52604 1.0000 811.148
MAR 4.90825 1.0071 263.690
APR 4.86581 1.0141 61.580
MAY 5.00106 0.9846 31.820

JUN 4.09559 1.0018 0

 

aBased on data from the Hawaii Department of Agriculture, Division
of Milk Control. Consumption adjusted as described in Appendix.

bSource: Various issues of the Federal Mi1k Order Market Statis-
tics (FMOS--221, 233. 293, 258, 285).

 

130

Table A,3..--Milk price. fruit nectar price. and consumer price index.

 

 

Deflated Deflated Honolulu
Retail Fluid Fruit Consumer
Milk Price Nectar Price Price Indeg

(5/1/2 0:11.13 (571/2 Gal.)a (1967=1.00)
1977 JAN 0.649038 0.600962 1.664
FEB 0.645161 0. 603345 1 . 674
MAR 0.647653 0.594177 1 .683
APR 0. 650888 0.5 97633 1. 690
MAY 0.641932 0.594817 1.698
JUN 0.651026 0.545455 1. 705
JUL 0.653061 0.530612 1.715
AUG 0.643 852 0.551044 1. 724
SEP 0. 640138 0.582468 1.734
001 0.638298 0.580794 1.739
NOV 0.636468 0.579128 1 .744
DEC 0.634648 0.577473 1.749
1978 JAN 0.628895 0.577904 1.765
FEB 0.623596 0.606742 1.780
MAR 0.639955 - 0.612131 1.797
APR 0.639471 0.622933 1.814
MAY 0.635268 0.629792 1 .826
JUN 0.630778 0.625340 1.839
JUL 0.633117 0.600649 1.848
AUG 0.63 0388 0.5 81897 1. 856
SEP 0.625000 0.571581 1. 872
001 0.625000 0.577331 1. 888
NOV 0.621053 0.605263 1 .900
0150 0.616832 0.601150 1.913
1979 JAN 0.603715 0.593395 1.938
FEB 0.606524 0.606524 1. 962
MAR 0.625000 0.630040 1 .984
APR 0.612855 0.597907 2.007
MAY 0.612043 0.587364 2.026
JUN 0.616438 0.621331 2.044
JUL 0.626822 0.612245 2.058
AUG 0.627413 0.593629 2.072
SEP 0.627395 0.589080 2.088
001 0.627078 0.579572 2.105
NOV 0.625294 0.578279 2.127
050 0.619181 0.558659 2.148

Table A. 3.--Continued .

135

 

 

Deflated Deflated Honolulu
Retail Fluid Fruit Consumer
Milk Price a Nectar Price Price Index

(5/1/2 Gal.) (571/2 061.)a (1967:100)
1980 JAN 0.610652 0.550964 2.178
FEB 0.602082 0.543232 2.209
MAR 0.611062 0.544157 2.242
APR 0.602463 0.532102 2.274
MAY 0.606860 0.540897 2.274
JUN 0.606593 0.545055 2.275
JUL 0.603147 0.537587 2.288
AUG 0.643199 10.547588 2.301
SEP 0.636833 0.546472 2.324
001 0.639386 0.575448 2.346
NOV 0.641734 0.611985 2.353
DEC 0.639560 0.614147 2.361
1981 JAN 0.629954 0:600751 2.397
FEB 0.678175 0:591862 2.433
MAR 0.668829 ' 0.579651 2.467
APR 0.660000 0.560000 2.500
MAY 0.656325 0.536993 2.514
JUN 0.648734 0.534019 2.528
JUL 0.647821 0.530035 2.547
AUG 0.643024 0.518316 2.566
SEP 0.647538 0.387747 2.579
001 0.644042 0.528346 2.593
N0v 0.645286 0.568006 2.588
DEC 0.646535 0.530391 2.588
1982 JAN 0.641814 0.511145 2.602
FEB 0.637405 0.515267 2.620
MAR 0.635948 0.521706 2.626
APR 0.634258 0.516521 2.633
MAY 0.597295 0.503381 2.662
JUN 0.617100 0.371747 2.690
JUL 0.616642 0.523774 2.692
AUG 0.619896 0.530809 2.694
SEP 0.613294 0.495777 2.723
001 0.606831 0.505087 2.752
NOV 0.612844 0.554128 2.725
DEC 0.611338 0.566877 2.699

136

Table A.3.--Continued.

 

Deflated Deflated Honolulu
Retail Fluid Fruit Consumer
Milk Price Nectar Price

a Price Indeg

(5/1/2 Gal.) (5/1/2 Gal.) (1967=100)

 

1983 JAN 0.583486 0.535780 2.725
FEB 0.600000 0.563636 2.750
MAR 0.586957 0.514493 2.760
APR 0.566787 0.516245 2.770
MAY 0.564047 0.491266 2.748
JUN 0.557594 0.502568 2.726

 

aFrom the Hawaii Department of Agriculture, Market News Branch,
and the Honolulu Advertiser weekly “Retail Food Price Guide.“

bBased on data from U.S. Department of Commerce, Bureau of

Labor Statistics, CPI Detailed Report, various issues.

137

Tablethr-Hawaii state personal income. resident population. and per
capita personal income.

 

 

Nominal State Deflated
Personal Incomea Resident Per Capita
(Millions<yf0ollars) Populationb Personal Income
1977 JAN 6548 909000 4329.04
FEB 6653 910167 4366.58
MAR 6758 911333 4406.13
APR 6787 912500 4401.07
MAY 6815 913667 4392.79
JUN 6844 914833 4387.77
JUL 6894 916000 4388.46
AUG 6945 917083 4392.65
SEP 6995 918167 4393.56
OCT 7098 919250 4440.20
NOV 7200 920333 4485.81
DEC 7303 921417 4531.64
1978 JAN 7345 922500 4511.08
FEB 7388 923583 4493.98
MAR 7430 924667 4471.52
APR 7490 925750 4460.16
MAY 7549 926833 4460.54
JUN 7609 927917 4458.99
JUL 7713 929000 4492.68
AUG 7816 930750 4524.53
SEP 7920 932500 4537.02
OCT 8004 934250 4537.76
NOV 8087 936000 4547.35
DEC 8171 937750 4554.84
1979 JAN 8231 939500 4520.66
FEB 8291 941250 4489.55
MAR 8351 943000 4463.60
APR 8423 944750 4442.25
MAY 8496 946500 4430.52
JUN 8568 948250 4420.54
JUL 8665 950000 4432.00
AUG 8761 951583 4443.42
SEP 8858 953167 4450.78
OCT 8938 954750 4447.32
NOV 9018 956333 4433.37
DEC 9098 957917 4421.64

Table A.h.--Continued.

138

 

 

Nominal State Deflated
Personal Incomea Resident Per Capita
(Millionscfi:DOIIarS) Populationb Personal Income
1980 JAN 9276 959500 4438.72
FEB 9453 961083 4452.59
MAR 9631 962667 4462.31
APR 9651 964250 4401.41
MAY 9672 965833 4403.76
JUN 9692 967417 4403.71
JUL 9785 969000 4413.48
AUG 9878 969833 4426.45
SEP 9971 970667 4420.10
OCT 10081 971500 4423.16
NOV 10191 972333 4454.30
DEC 10301 973167 4483.28
1981 JAN 10362 974000 4438.30
FEB 10424 974833 4395.03
MAR 10485 975667 4356.10
APR 10547 976500 4320.33
MAY 10608 977333 4317.43
JUN 10670 978167 4314.94
JUL 10774 979000 4320.81
AUG 10877 980225 4324.41
SEP 10981 981450 4338.33
OCT 11057 982675 4339.35
NOV 11134 983900 4372.56
DEC 11210 985125 4405.45
1982 JAN 11239 986350 4379.15
FEB 11268 987575 4354.87
MAR 11297 988800 4350.71
APR 11370 990025 4361.78
MAY 11442 991250 4336.21
JUN 11515 992475 4313.13
JUL 11589 993700 4332.27
AUG 11664 994975 4351.49
SEP 11738 996250 4326.91
OCT 11747 997525 4279.12
NOV 11756 998800 4319.31
DEC 11765 1000080 4358.69

139

Table A.h.--Continued.

 

 

Nominal State Deflated
Personal Incomea Resident Per Capita
(Millionscfi Dollars) Populationb Personal Income

1983 JAN 11853 1001350 4343.86
FEB 11940 1002630 4330.45
MAR 12028 1003900 4341.04
APR 12102 1005180 4346.46
MAY 12176 1006450 4402.46
JUN 12250 1007730 4459.32

 

aBased on data from U.S. Department of Commerce, Bureau of Economic
Analysis, Survey of Current Business, April 1983, October 1983, and
Jan 1984.
‘\

 

bB}ased on data from Hawaii, Department of Planning and Economic
Deve10pment, State of Hawaii Data Book 1983.

140

Table A.5.--Visitors to Oahu. average length of stay. and average daily
visitor census.a

 

 

Westbound Average Average
Overnight and Length of Daily Visitor

Longer Visitors Stay Census

1977 JAN 197940 6.03 38502.5
FEB 188767 6.06 40854.6
MAR 211506 5.77 39367.4
APR 186099 5.66 35110.7
MAY 162989 5.73 30126.7
JUN 203772 6.06 41161.9
JUL 218362 6.08 42827.1
AUG 228109 5.88 43267.1
SEP 160662 5.54 29668.9
OCT 178760 5.94 34252.7
NOV 162717 5.49 29777.2
DEC 195630 6.32 39883.3
1978 JAN 197674 5.96 38004.4
FEB 193618 6.02 41627.9
MAR 226525 5.71 41724.4
APR 185866 5.66 35066.7
MAY 170781 5.55 30575.3
JUN 217595 6.01 43591.5
JUL 238373 6.15 47290.1
AUG 235232 5.79 43935.3
SEP 186228 5.47 33955.6
OCT 213348 5.74 39503.8
NOV 213608 5.71 40656.7
DEC 216045 6.25 43557.5
1979 JAN 232118 6.12 45824.6
FEB 225847 5.84 47105.2
MAR 236688 5.70 43520.1
APR 185714 5.54 34295.2
MAY 161154 5.65 29371.6
JUN 216684 6.10 44059.1
JUL 247185 6.21 49516.7
AUG 262311 5.87 49669.9
SEP 188217 5.69 35698.5
OCT 201564 5.78 37581.9
NOV 182581 5.55 33777.5
DEC 202654 6.14 40138.6

141

Table A.S.--Continued.

 

 

Westbound Average Average
Overnight and Length of Daily Visitor

Longer Visitors Stay Census

1980 JAN 194093 5.98 37441.2
FEB 195294 5.87 39530.2
MAR 224540 5.68 41141.5
APR 188490 5.52 34682.2
MAY 177902 5.50 31563.3
JUN 212831 5.88 41714.9
JUL 263799 6.23 53015.1
AUG 238713 5.79 44585.4
SEP 154597 5.53 28497.4
OCT 176023 5.62 31911.3
NOV 172714 5.47 31491.5
DEC 199741 6.24 40205.9
1981 JAN 181273 6.09 35611.4
FEB 195612 5.80 40519.6
MAR 196019 5.58 35283.4
APR 187020 5.69 35471.5
MAY 183508 5.54 32794.7
JUN 225276 6.25 46932.5
JUL 229850 6.27 46489.0
AUG 257381 5.98 49649.6
SEP 171610 5.79 33120.7
OCT 190622 5.63 34619.4
NOV 178100 5.74 34076.5
DEC 202206 6.60 43050.3
1982 JAN 205932 6.23 41385.7
FEB 218725 5.86 45776.0
MAR 247247 5.68 45302.0
APR 216120 5.61 40414.4
MAY 200076 5.77 37240.0
JUN 250069 6.05 50430.6
JUL 267627 6.07 52403.1
AUG 264963 5.75 49146.4
SEP 194847 5.53 35916.8
OCT 214040 5.54 38251.0
NOV 218054 5.35 38886.3
DEC 204955 5.93 39205.9

142

Table A.5.--Continued.

 

 

Westbound Average Average
Overnight and Length of Daily Visitor
Longer Visitors Stay Census
1983 JAN 195057 5.62 35361.9
FEB 214920 5.51 42293.2
MAR 250658 5.48 44309.9
APR 191806 5.47 34972.6
MAY 209499 5.62 37980.1
JUN 279346 6.06 56427.9

 

a O O O O
Source: Hawaii Visutors Bureau.

143

Table A.6.--Oahu resident population and de facto population.

 

 

Oahu
Resident De Facto
Populationa Population
1977 JAN 729900 768403
FEB 730550 771405
MAR 731200 770567
APR 731850 766961
MAY 732500 762627
JUN 733150 774312
JUL 733800 776627
AUG 734342 777609
SEP 734883 764552
OCT 735425 769678
NOV 735967 765744
DEC 736508 776391
1978 JAN 737050 775054
FEB 737592 779220
MAR 738133 779857
APR 738675 773742
MAY 739217 769792
JUN 739758 783350
JUL 740300 787590
AUG 741442 785377
SEP 742583 776539
OCT 743725 783229
NOV 744867 785524
DEC 746008 789565
1979 JAN 747150 792975
FEB 748292 795397
MAR 749433 792953
APR 750575 784870
MAY 751717 781089
JUN 752858 796917
JUL 754000 803517
AUG 754917 804587
SEP 755833 791531
OCT 756750 794332
NOV 757667 791444
DEC 758583 798722

144

Table A.6.--Continued.

 

 

Oahu
Resident De Facto
Populationa Population
1980 JAN 759500 796941
FEB 760417 799947
MAR 761333 802475
APR 762250 796932
MAY 763167 794730
JUN 764083 805798
JUL 765000 818015
AUG 765233 809818
SEP 765467 793964
OCT 765700 797611
NOV 765933 797425
DEC 766167 806373
1981 JAN 766400 802011
FEB 766633 807153
MAR 766867 802150
APR 767100 802571
MAY 767333 800128
JUN 767567 814499
JUL 767800 814289
AUG 768417 818067
SEP 769033 802154
OCT 769650 804269
NOV 770267 804343
DEC 770883 813933
1982 JAN 771500 812886
FEB 772117 817893
MAR 772733 818035
APR 773350 813764
MAY 773967 811207
JUN 774583 825014
JUL 775200 827503
AUG 776108 825254
SEP 770017 805934
OCT 777925 816176
NOV 778833 817719
DEC 779742 818948

Table A.6.--Continued.

14S

 

 

Oahu
Resident De Facto
Populationa Population
1983 JAN 780650 816012
FEB 781558 823851
MAR 782467 828776
APR 783275 818248
MAY 784283 822263
JUN 785192 841520

 

3Based on data from Hawaii Department of Planning and Economic
Development, State of Hawaii Data Book 1983.

APPENDIX TWO

CONTENT ANALYSIS OF OAHU NEWSPAPER ARTICLES

146

APPENDIX TWO

CONTENT ANALYSIS OF OAHU NEWSPAPER ARTICLES

The purpose of this study was to determine how consumer aware-
ness of milk safety affects fluid mi1k demand. 'To this end. variables
representing information presented to the public about the contamina-
tion incident were included in the model.

Awareness of milk safety has three components: awareness of
milk quality. the level of state protection of consumer health. and
processor integrity. Milk quality refers to its physical characteris-
tics. especially the presence or absence of harmful organisms or chemi-
cals. The level of state protection of consumer health refers to the
efforts and ability of the government to ensure a safe milk supply and
keep adulterated products off the market. Processor integrity reflects
concern and action taken by the processors to ensure safe milk. Aware-
ness of milk safety may arise from friends. family. and the mass media.
This study measured the amount and direction of information (positive.
negative. or neutral) presented to the public by the mass media as a
proxy for that from all sources.

Selection of the appropriate media was limited by the data
available. 0f the three major Oahu television networks. one cooperated

by approximating the monthly number of news reports relating to milk

147

148

contamination aired from March 1982 through June 1983. as well as
reports about the importation of Safeway milk from March through June
1983. .Another station provided data only through 1982 while a third was
uncooperative. These reports could not be coded whether they reported
milk safety positively. negatively. or otherwise. ‘The two major Oahu
newspapers. WW and MW
provided photocopies of articles related to milk and the contamination
incident. Reports in the Final edition of the.AdveLtisen and the Home
edition of the.§tan;Bu11§tin were used since they are for home delivery
on Oahu and street sales on Oahu and other islands. Relevant articles
were those that mentioned "milk." "dairy." "heptachlor." or other con-
taminants (e.g.. antibiotics. DDT) in the headline or first two para-
graphs. An exception to this rule were those articles on dairy
management-labor negotiations and federal mi1k fee assessments. These
related more to business and labor unions than to milk safety and were
excluded.

The analyst cannot assume that an article's content will affect
all persons similarly. Hence. one cannot code the articles in terms of
their effects on consumers (e.g.. creates doubts about. or enhances
confidence in milk safety). One can code the articles based on whether
they present positive or negative information about milk safety more
confidently. and this was pursued.

Articles were coded conservatively. from the viewpoint of a
concerned consumer who accepts federal "action" (permissible) levels as

safe until even federal agencies question their adequacy. Hence. the

149

announcement that heptachlor levels fell below the action level was
coded positively. .An exception to this was two articles announcing the
discovery of DDT in nonviolative levels inuanG DDT is a more
familiar poison. whose presence in even small amounts was considered
negative information.

Mention of current wrongdoing was coded negatively: the direction
of reports of past wrongdoing was more equivocal. Reports of unethical
behavior ten years in the past probably do not affect awareness of
current food product safety. Mention of such behavior within the past
six to twelve months may affect awareness of food safety. but its
specific effect is uncertain and varies between individuals. Mention
of past contamination and wrongdoings by State officials or processors
more than one month ago was not considered negative information. (The
problem was identified and presumably correctedd If. however. wrong-
doings which had not been corrected were reported. it was coded nega-
tively.

When uncertain about the direction of an article. each sentence
was analyzed to determine whether it was positive. negative. or neutral
information. Respective weightings were +1. -1. and zero. 'The
article's headline was double counted. The sum of all coding deter-
mined how the article was coded. It was recognized that actions and
statements by certain sources have more credibility and impact than
those of others. Unfamiliar with Hawaiian officials. I did not attempt

to account for this.

150

The following are the coding rules used. .An article was coded
as positive media coverage. or positive information about milk quality
by mentioning:

-levels of contaminants below the action level that were pre-

viously above.

-how the presence of the contaminant was not harmful.

-government announcement that the crisis was over.

-government announcement that the dairy farms were clear of

contamination.

-others' positive attitudes toward local milk.

Negative information about milk quality was mention of:

-current presence of contaminant in Oahu milk above the

action level.

-dangers of the contaminant or uncertainty of the dangers.

-questioning of the adequacy of the action level by

federal agencies.

-calls for a lower action level.

-product recall and banning on airlines.

-revelation of longer exposure period than originally suspected.

-doubts. suspicions of milk quality (e.g.. about the length of

exposure to contaminant).

-others' concerns about milk quality.

-presence of DDT. even below the action level.

151

Positive media coverage. or positive information concerning the
current level of state protection of consumer health by ensuring safe
milk supplies. was mention of the following:

-precautionary action taken to protect the public (e.g.. recall

when the health director suspected wrongdoing by Meadow Gold).

-the first mi1k recall for heptachlor (other recalls contra-

dicted earlier assurances of safe mi1k supplies).

-recalls for antibiotics.

-testing and improvement of laboratory facilities.

-invitation of outside help (e.g.. FDA expert).

-an investigation into the incident.

-lowering of the action level.

-state officials' certainty of milk safety (first time only).

-resignations ("The rascals are gone."). I
Negative media coverage of the level of state protection involved:

-government allowance of violative levels of contaminants.

-government interest more in industry welfare than public

safety (e.g.. delay in recalling mi1k. This did not include
denial of Safeway's license in 1983. which was seen as more a
matter of protecting island business than a health matter.).

-revelation of past incompetence. or mixed priorities with no

indication of correction.

-government statements proven false. contradictory lab results

from other labs.

152

-inaction. indecision. confusion on the part of the State.

-accusations of a cover-up.

Positive information about processor integrity was:

-openness and cooperation with state officials.

-openness with public (e.g.. creating a telephone answer line).

-voluntary action to ensure milk safety (e.g.. independent
testing).

-clearance from investigations.

Negative information was:

neutral

-lack of openness and cooperation with state officials.

-destruction of records.

-illegal or intentional use of contaminated milk within the
past month.

-suspicions. investigations. and accusations made about
both processors.

If only one processor was criticized. the article was coded
for processor integrity.

Each article was coded positive. negative. or neutral with

respect to overall milk safety. Positive (negative) information about

milk safety was positive (negative) information about milk quality.

level of state protection. and processor integrity. Sometimes there

was combination of positive and negative components of milk safety.

Information on milk quality was of prime importance in determining the

direction of the article with respect to milk safety. If milk on the

shelf was known to be contaminated. awareness of milk safety would have

153

been negative. regardless of the level of State protection and proces-
sor integrity. ‘The level of government protection was of secondary
importance and overrode information about processor integrity. If the
article reported unethical processor behavior and State action to
correct it. the article was coded positively for overall milk safety.
To reflect the prominence editors give a story in terms of
placement within the paper. position on the page. size of headline and
length of story. Budd (1964) developed an "Attention Score." This was

used (with slight modifications) to reflect the prominence of the

article. and hence attention given it by the reader. Items were weighted

as follows (taken from p. 260):

1. One point is assigned to any article with a headline two
columns or more in width. except that an article carrying a
headline that occupies horizontally more than half the number
of columns of the page is assigned two points. [One point is
assigned stories whose headline is to the side of the story.]

2. One point is assigned to any story appearing above the fold or
above the measured center of any page. To be considered above
the fold. the first line of the body text of the story has to
appear above the fold.

3. One point is assigned to any article occupying three-fourths of
a column or more (based on the column length of the newspaper
concernedL For purposes of assigning the attention score.
pictures accompanying articles are measured as part of the
over-all length of the story.

4. One point is assigned for any article appearing on page one.
the editorial page. or the front page of a section.

Hence an article may be scored between zero and five. Pictures
and cartoons without an accompanying article are scored in the same
manner: Pictures and cartoons accompanying an article are con-
sidered part of the story and are included in measurement of the
article's overall length. as are the headlines.

Scores and confidence in them are given in Table 8.1.

154

Table B.l.--Components of media coverage.a

 

Average Average Average
NM Confidence PM Confidence NEUM Confidence TM

 

 

1982 MAR 92 3.5 34 3.5 52 3.6 178
APR 122 3.6 60 3.6 54 3.7 236
MAY 14 3.6 33 3.6 30 3.6 77
JUN 6 3.4 5 3.4 13 4.0 24
JUL 10 3.7 9 3.9 14 3.2 33
AUG 85 3.4 11 3.2 8 3.8 104
SEP 25 3.5 28 3.5 12 4.0 65
OCT 6 3.5 12 3.5 7 3.9 25
NOV 0 O O 0
DEC 18 3.8 9 3.8 16 3.5 43

1983 JAN 8 3.3 l 4.0 23 3.9 32
FEB 8 3.3 5 3.7 19 3.7 32
MAR 0 O 38 3.9 38
APR 13 3.5 5 4.0 38 3.8 56
MAY 16 3.6 14 3.9 84 3.9 114
JUN 10 3.7 4 3.8 36 3.9 50
aNM = Negative media
PM = Positive media
NEUM = Neutral media
TM = Total media

REFERENCES

155

REFERENCES

Arndt. Johan. "Perceived Risk and Word-of-Mouth Advertising." In 31:15

Iakingendlnionmaflenflendlinginflonsumenflenuiot. pp. 389-316.
Edited by Donald F. Cox. Boston: Harvard University. 1967.

Bartlett. Randall. Economic. Eounoanons .oi Eolitieal Bowen. New
York: The Free Press. 1973.

Beck. Melinda. with Hager. Mary. "EDB: A Cancer Scare." .Newsweek.
13 February 1984. pp. 23-24.

Berel son. Bernard. Content Anabel: in Communication Reseanoh.
Glencoe. 111.: The Free Press. 1952.

Boehm. William T. "The Household Demand for Fluid Milk in the United
States With Regional Consumption Projections Through 1990."
Research Div. Bull. 120. Blacksburg: Virginia Polytechnic
Institute and State University. December 1976.

Boylan. Paul. In U.S. Congress. House. Committee on Agriculture.

Miscellaneous. fleaninosfleioneineflomitieeonAonlenltnne
9.0 11.8. 65.25.. 9lst Cong.. lst Sess.. Dec. 8. 1969.

Brown. Joseph 0. "Effect of a Health Hazard 'Scare' on Consumer

Demand." Amenioan Jounnal of AonieuJinnel Economics 51 (August
1969): 676-78.

Budd. Richard W. "Attention Score: A Device for Measuring News 'Play.'"
Journalism manteflx (Spring 1964): 259-62-

Budd. Richard W.; Thorp. Robert K.: and Donohew. Lewis. .antent

.of Communications. New York: The Macmillan Company.
1967.

Butler. J.. and Thompson. S. R. "On the Selection of Alternative Lag
Models." Michigan State Univ.. Ag. Econ. Staff Paper 179-42
(August 1979).

Chaffee. Steven H. "Mass Media Effects: New Research Perspectives."

In Mass Conmunieetien Reflex leannook. lol. .1. PP- 77408-
Edited by G. Cleveland Wilhoit and Harold de Brock. 1980.

156

157

Chen. Edwin. ‘EBB;.An.Amenican.Inagedy. Englewood Cliffs. N.J.:
Prentice-Hall. Inc.. 1979.

Clark. Darral G. "Econometric Measurement of the Duration of Adver-

tising Effect on Sales." .1ounnal.g£.Manket1ng.Beseanch 13
(November 1976): 345-57.

Cohen. Gary. "United States Department of Justice Post-Hearing Memo-
randum and Findings of Fact." Presented to the Hawaii Board of
Agriculture. Nov. 4. 1983.

Cohen. Gary. and Eisenstat. Philip M. "Pre-Hearing Memorandum of the
United States Department of Justice." Presented to the Hawaii
Board of Agriculture. 1983.

Consumer Nutrition Center. "Food Consumption and Dietary Levels of
Households in Hawaii. Winter 19783' Nationwide Food Consumption
Survey 1977-78. Science and Education Administration (USDA).
Preliminary Report No. 4. April 1981.

Cook. Hugh L.. et a1. IneflaitxfiuooeetonofAmenioenAonieultnne:
.and Vertical Coondination. North Central Regional
Research Publication 257. University of Wisconsin-Madison.

November 1978.

Cox. Donald F. "Risk Handling in Consumer Behavior--An Intensive

Study of Two Cases." In Risk liking and 1111913111120
in Consumer Beneitiot. pp. 34-81. Edited by Donald F. Cox.
Boston: Harvard University. 1967.

Cusmano. Donald R.. and Richey. Marjorie H. "Negative Salience in
Impressions of Character: Effects of Extremeness of Stimulus

Information." .Esyghonomig Science 20 (1970): 81-83.

"Dairy Farmers' Losses Hit at Least 3600.000." .flgnolulu Stan:
mm, 14 May 1982' p0 A-go

Donoho. Harry R. "Average Cost of Milk Production." (Study requested
by the State of Hawaii Department of Agriculture. Division of
Milk Control.) Dec. 11. 1980.

Fiske. George. and Chandran. Rajan. "How to Trace and Recall

Products." .flanxand Business Bexiew (November-December 1975):
90-%o

Folkes. Valerie 8. "Consumer Reaction to Product Failure: An Attri-

butional Approach." .Jounnal of ConsumeL.BeseaLch 10 (March
1984): 398-409.

158

Garrod. Peter V.. and Ching. Chauncey T. K. "Beef Marketing in
Hawaii: An Overview." Information Text Series 009. Hawaii
Institute of Tropical Agriculture and Human Resources. Univer-
sity of Hawaii. August 1982.

Grocery Manufacturers of America. Inc. .Guidelines.£o:.Enoduct Recall.
Washington. D.C.: Grocery Manufacturers of America. Inc.. March
1974.

Gujarati. Damodar. .Baslg.£cgnome1n195. New York: McGraw-Hill
Book Company. 1978.

Hamilton. James L. "The Demand for Cigarettes: Advertising. the Health
Scare. and the Cigarette Advertising Ban." .Bexiew.of.£gonomios
.and.Steiieiios 54 (November 1972): 401-11.

Harpham. Anne. "Supply Expected to Exceed 50% of What's Usual."
.HQannlu.AdleEti§e£. 20 March 1982. pp. A-1 and A-4.

. "New Testing Method for Milk to Be Aired." .flgnglulu
W, 30 Maf‘Ch 1982’ p. A-30

.________. "Milk Products Will Be Tested Before Reaching Store
Shelves." .Honolulu.Adxeniisen. 6 April 1982. p. A-4.

. "Clark: Weren't Prepared for Milk Crisis." .flgnolulu
Meniset. 21 May 1982. p. A-3.

.________. "Meadow Gold Milk Diverted to Foremost." .flQannlu.Ag1§n:
.tisen. 25 September 1982. p. A-9.

_. "Foremost Buy-Back of Yogurt Revealed." Honolulu
Admniiset. 6 October 1982. p. A-3.

_________. "Safeway Milk Import Bid Denied." .Honolulu.Adxentisen.
31 March 1983. pp. A-1 and A-4.

Harpham. Anne. and Burris. Jerry. ”Special Senate Panel to Probe

Milk Woes." .Honolulu.Ader11§en. 2 April 1982. pp. A-1 and
A-4.

Harpham. Anne. and Hastings. Barbara. "New Recall--Some Cartons of 2%
Milk." .Honolulu.Adxentieen. 24 March 1982. pp. A-1 and A-4.

Harpham. Anne. and Hastings. Barbara. "Oh N01 Now Some of the
Yogurt." .flonolulu.Adxentisen. 8 April 1982. pp. A-1 and A-4.

Harpham. Anne. and Hastings. Barbara. "A New Curdle in Milk Issue--

Some Cottage Cheese Recalled." .Honolulu.Adxentisen. 15 April
1982, pp. A-1 and A-4.

159

Harpham. Anne. and Oshiro. Sandra 5. ”Let's Import 'Safer' Milk. State
Is Asked." .Honolulu.Adxentisen. 20 April 1983. p. A-4.

Hastings. Barbara. and Harpham. Anne. "Toss Out That Milk in Fridge--
It Contains Pesticide Poison." .Honolulu.Adxentisen. 19 March
1982. pp. A-1 and A-4.

Hawaii Agricultural Reporting Service. .Statistigs of Hawaiian
.Agnigultune..1982. Honolulu. July 1983.

Hawaii Department of Agriculture. Division of Milk Control. "Honolulu
Milk Shed Monthly Report." Various issues.

Hawaii Department of Planning and Economic Development. .State of
Hanan Data Book 1982. Honolulu. 1983.

. Research and Economic Analysis Division. .Eotential.lngome
.Eiieots.of.Milk.Lmoontation. Research Report 83-3. June 1983.

Hawaii Senate. Report of. the Senate .Soeoial Comittee lmsfloatino
fleotaoblot Contamination of Milk. February 1. 1983.

Heiner. R. A. "The Origin of Predictable Behavior." ‘Amenican Economic
.Bexiew (September 1983): 560-95.

Hirschman. Albert 0. E25121... 10.1.9.9... and main Responses IQ DeoJJne

.in Eirms. Cmanjzationsi and States. Cambridge: Harvard Univer-
sity Press. 1970.

Hoffer. George E.. and Wynne. A. James. "Consumer Responses to Auto

Recalls." Journal of Consume: Afiains 9 (Winter 1975): 212-18.

Hogg. Howard C. "Honolulu Market Projections for Selected Livestock
Products: Beef and Veal. Pork. Eggs. Chicken and Milk." Depart-
mental Paper 16. University Of Hawaii. Hawaii Agr. Exp. Sta.
January 1974.

Kakesako. Gregg K. "Another Recall: State Pulls Some Cottage
Cheese." .Honolulu.§tat:8ulletin. 14 April 1982. p. A-l.

Kakesako. Gregg K.: Gomes. Lee: and Morita. Stirling. "Ariyoshi
Answers Critics but Consumer Isn't Buying." ‘flonolulu.§tan:
.Bulletin. 25 March 1982. pp. A-1 and A-3.

Keir. Gerry. "40% Still Uneasy About Heptachlor in Milk." .flonolulu
.Adxenllsen. 27 August 1982. pp. A-1 and A-5.

Kessler. David A. "Food Safety: Revising the Statute." ‘Scienge.
9 March 1984. pp. 1034-40.

160

Kinnucan. Henry W. "Media Advertising Effects on Milk Demand: The
Case of the Buffalo. New York Market With an Empirical Compari-
son of Alternative Forms of the Sales Response Equation." Dept.
of Agr. Econ. A.E. Res. 83-13. Cornell University. February 1983.

Kleinfield. N. R. "Tylenol's Survival Challenge." .Neu.Io£k.Ilmes.
8 October 1982. p. 31.

Kmenta. Jan. .Elements.o1.Egonometnigs. New York: The Macmillan
Company. 1971.

Kneale. Dennis. "Tylenol Orders Fall 25% but Competitors Lack Enough
Products to Fill Market Gap." .Wall.§1£eet.lou£na1. 13 October
1982. p. 56.

Koshi. James. In U.S. Congress. House. Committee on Agriculture.

Aonioultunefinobjemsjnflauau. Beaninosbeionetbeiuooomittee
.on.0ainx.and.Eou1tnx. 96th Cong.. 2d session. Honolulu. Hawaii.
10 June 1980.

Locander. William B.. and Hermann. Peter W. "The Effects of Self-
Confidence and Anxiety on Information Seeking in Consumer Risk

Reduction." Joutnal of Manketing Beseanob (May 1979): 268-74.

Lynch. Russ. "Ways to Save Milk Industry Explored." ,Honolulu Stan:
.Bulletin. 8 April 1982. p. E-l.

McCombs. Maxwell E.. and Shaw. Donald L. "The Agenda-Setting Function
of the Press." In Ihe Emeroeme of Amenioan Eolitieal issues:

InerenoazaettingfiunotionoitbeEness. pp. 1-18. Edited by
Donald L. Shaw and Maxwell E. McCombs. St. Paul: West Publish-

McGuire. E. Patrick. ed. .Managing.Enodugt.Begalls. New York:
The Conference Board. Inc.. 1974.

McMurdo. Mary Jane. "Should Hawaii Import Fresh Mainland Milk? Yes."
mm W! 6 May 1983’ p0 A-17o

Mollett. J. A. "Economics of Market Milk Production on Oahu--
The Honolulu Milkshed." Hawaii Agr. Expt. Sta. Agr. Econ. Rept.
No. 41. December 1961.

Morison. I. G.: Kefford. N. P.: and Harada. W. T.. eds. "Dairy
Industry Analysis." College of Tropical Agriculture and Human
Resources. University of Hawaii. May 20. 1981.

 

161

Mowen. John C. "Further Information on Consumer Perceptions of

Product Recalls." In Aduanoes in Consumer Beseanob. pp. 519-23.
Edited by Jerry C. Olson. Ann Arbor: Association for Consumer

Research. 1980.

Mowen. John C.. and Pollman. Scott B. "An Exploratory Study Investi-
gating Order Effects in Reporting Negative Corporate Communica-

tions." In Advanoes .in Consumer Beseanob. pp. 215-20. Edited
by Andrew A. Mitchell. Ann Arbor: Association for Consumer

Research. 1982.

Mowen. John C.: Jolly. David; and Nickell. Gary S. "Factors Influenc-
ing Consumer Responses to Product Recalls: A Regression Analysis

Approach." In Aduanoes in Consumer Beseanob. pp. 405-407.
Edited by Kent B. Monroe. Ann Arbor: Association for Consumer

Research. 1981.

Ong. Vickie. "New Milk Recall in Waikiki Only." 'flonolulu.Adxentisen.
28 March 1982. p. A-l.

Pace. Eric. "Lingering Damage to Sales of Tylenol Is Expected." .New
.ICLK.I1me§. 2 October 1982. p. 21.

Peterson. Iver. "Reduced Levels of Dioxin Found in Michigan Fish."

.Nen.ICEK.I1mes. 17 June 1983. sec. 1. p. 1.

Rede. George. "PCB Contamination: Economic Impact." .Caoitol
.Jounnal (Salem. Oregon). October 1979. p. 1-B.

Renaud. Bertrand M. "The Impact of Economic Growth on the Agricultural
Trade Structure of an Island Economy." Research Bulletin 150.
Hawaii Agr. Expt. Staci 19710

Richey. Marjorie H.. et a1. "Negative Salience in Impressions of
Character: Effects of Unequal Proportions of Positive and Nega-

tive Information." .Jgunnal of Social Esxohology 97 (1975):
233-410

Seibert. Donald V. "Time to Revive a Commitment to Ethics." .uew.xgnk
limes. 25 December 1893. p. B-2.

Schlenker. Anna. and Christ. Paul. "Adjusting In-Area Fluid Milk
Sales Data for Calendar Composition." U.S. Department of Agri-
culture. Consumer and Marketing Service. 28 January 1971.
(Mimeographed.)

Schuker. Raymond E.. et al. "The Impact of Saccharin Warning Label on
Sales of Diet Soft Drinks in Supermarkets." .Jgunnal.gf.flub119

Boliox and Mutation 2119531=45-55-

162

Scott. Frank 8.. Jr. "Transition in Consumer Demand for Milk in
Honolulu and Kailua." Agricultural Economics Bulletin 25.
Hawaii Agr. Expt. Sta.. January 1967.

Shaw. Eugene. "The Interpersonal Agenda." In Ibe.£mengenge.g£
AmenicanEontioallssueszInerenoazsettJnoEunetionoitne
.Eness. pp. 69-88. Edited by Donald L. Shaw and Maxwell E.
McCombs. St. Paul: West Publishing Co.. 1977.

Shulstad. Robert N.. and Stoevener. Herbert H. "The Effects of
Mercury Contamination in Pheasants on the Value of Pheasant Hunt-
ing in Oregon." .Land.£conomigs 54 (February 1978): 39-49.

Sternthal. Brian; Phillips. Lyon: and Dholakia. Ruby. "The Persuasive
Effect of Source Credibility: A Situational Analysis." ‘Euhljg

.Qoinion.0uantenlx.42 (Fall 1978): 285-314.

Strong. Edward C. "The Spacing and Timing of Advertising." ,Jounnal
of Aduentisino Beseanch 17 (December 1977): 25-31.

Sutherland. Max. and Galloway. John. "Role of Advertising: Persua-

sion or Agenda Setting?" Jounnal of Adxentisino Beseanob 21
(October 1981): 25-29.

Swartz. David G.. and Strand. Ivar E.. Jr. "Avoidance Costs Associated
With Imperfect Information: The Case of Kepone." .Land Economics
57 (May 1981): 139-50.

Thompson. Stanley R.. and Eiler. Doyle A. "A Multivariate Probit
Analysis of Advertising Awareness on Milk Use." .Canadian.lounna1
‘91.Agnigultunal.£gongm1gs 23 (February 1975): 65-73. (a)

Thompson. Stanley R.. and Eiler. Doyle A. "Producer Returns for

Increased Milk Advertising." .Amenigan.Jounna1.o£.Agnicu1tuna1
,Eggngmigs (August 1975): 505-508. (b)

Thompson. Stanley R.. and Eiler. Doyle A. "Determinants of Milk

Advertising Effectiveness." ,Amenigan‘lounnal.o£.Agnieultunal
momma (May 1977): 330-35.

Thompson. S. R.: Eiler. D. A.: and Forker. O. D. "An Econometric
Analysis of Sales Response to Generic Mi1k Advertising in New
York State." .fieaneh. Vol. 6. No. 3. Cornell University
Agr. Exp. Sta. 1976.

Tomek. William G.. and Robinson. Kenneth L. .Agnigultunal.flngdugt
.Eziges. 2nd ed. Ithaca: Cornell University Press. 1981.

"Tylenol Slayings Remain Unsolved." .Nel.IQLk.Iime§. 2 October 1983.
56C. 1' p. 130

163

U.S. Army Corps of Engineers. .Watenbonne.Commenoe.o£.the.United
.States. Part Four. Waterways and Harbors. Pacific Coast.
Alaska and Hawaii. Calendar Years 1977 through 1981.

U.S. Congress. Office of Technology Assessment. .Enxinonmenta1.Con:

.taminants.1n.£ood. Washington. 0.0.: Government Printing
Office. December 1979.

U.S. Department of Agriculture. Agricultural Marketing Service.
Eedenal Cale: Mankot Statistios (FMOS). Various issues.

U.S. Department of Agriculture. Dairy Indemnification Program. "Fiscal
Year Report of Claims Paid." 1983 and 1984.

U.S. Department of Agriculture. Food Safety and Quality Service.

ReoontontneECBlnoJoentintneWestennunitedatates.
January 1980. (a)

_______._.. ImonouingtbeESCSBesioueEnoonam. July 1980. (b)

U.S. Department of Commerce. Bureau of the Census. 1280 Census of
.Eonulation. Vol. 1. Characteristics of the Population. Chap-
ter B. General Population Characteristics. Part 13. Hawaii.
Washington. D.C.: Government Printing Office. July 1982.

U.S. Department of Commerce. Bureau of Economic Analysis. .Sunxey.o1
.Cunnent Business. Various issues.

U.S. Department of Commerce. Bureau of Labor Statistics. .921 Detailed
.Bepont. Various issues.

Van Ravenswaay. Eileen 0.. and Smith. Mark E. "Environmental Contami-
nation of Livestock." Michigan State University. forthcoming.

Waldholz. Michael. "Tylenol Maker Mounting Campaign to Restore Trust
of Doctors. Buyers." .Wall.§tneet.lounna1. 29 October 1982.
SOC. 20 p. 1.

Watanabe. June. "Meadow Gold Dumps Milk." ‘Honolulu.Stan:Bu11etin.
22 April 1982. pp. A-1 and A-24.

. ”Heptachlor Hassle Back in Limelight." .Honolulu.§tan:
.Bulletin. 2 August 1982. pp. A-1 and A-3.

Watanabe. June. and Mayer. Phil. "Allegations Denied by Meadow Gold."
.flonolulu.§tan:8ulletin. 3 February 1983. p. A-l.

164

Weinberger. Marc G.. and Dillon. William R. "The Effects of Unfavor-
able Product Rating Information." In Anyances.1n Consume:
.Beseanon. pp. 528-32. Edited by Jerry C. Olson. Ann Arbor:
Association for Consumer Research. 1980.

Weinberger. Marc G.: Allen. Cris T.: and Dillon. William R. "Negative
Information: Perspectives and Research Directions." In Anyanges

.in Consumer Beseanon. pp. 398-404. Edited by Kent B. Monroe.
Ann Arbor: Association for Consumer Research. 1981.

Witt. Stephen F.. and Pass. Christopher L. "The Effects of Health
Warnings and Advertising on the Demand for Cigarettes." .Sgottish
‘Jounna1.oi.Eolitioa1.Eoonomy,28 (February 1981): 86-91.

Yon. Bernard. and Mount. Timothy D. "The Responses of Sales to Adver-
tising: Estimation of a Polynomial Lag Structure." Dept. of Agr.
Econ. A.E. Res. 75-4. Cornell University. April 1975.