SHRIVER

L ‘

 

AH

 

'9

 

   
      

 

_ -1 fa ’. fl _
- J , f— : ‘
' "15-; 3: ..

I

7,

 

 
  
  
   
   
  

t.) W
. 6"; aka ;Ir|y

   
  
 

Wm" 50.. “if" 4 .’;3173}::":';
‘ . . . lewd}.
.-;3F.‘- ‘fi't-r

" "' '.A ’ P" U
' I'M ‘ -\ M
: . . ‘ mtg .. w ‘

1_--1

w

 
  
  
 
 

  
  
 
 

  

 
    
 

 

 

fllllllltllllllllfllllilllHlllHlHl.|Hl||1|l\HIHHNW “BM"

3 1293 00074 158

Michigan State
University

 

 

This is to certify that the

thesis entitled

Enforcement of Federal Standards for
Chemical Residues in Meat: An Evaluation

presented by

Ann Lynch Shriver

has been accepted towards fulfillment
of the requirements for

 

 

Master of Science degree in Agricultural Economics

z’fcéefl/j ifl/IWA!
L/
Major professor

Dr. Eileen van Ravenswaay

[MNe May 30, 1984

 

0-7639 MSU is an Affirmative Action/Equal Opportunity Institution

 

 

thESI.) RETURNING MATERIALS:

Place in book drop to
LIBRARJES remove this checkout from
JIEIESIEIL your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

 

 

 

 

W".

 

 

ENFORCEMENT OF FEDERAL STANDARDS FOR CHEMICAL
RESIDUES IN MEAT: AN EVALUATION

By

Ann Lynch Shriver

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

“73/8 L/Q'N

ABSTRACT

ENFORCEMENT OF FEDERAL STANDARDS FOR CHEMICAL
RESIDUES IN MEAT: AN EVALUATION

By
Ann Lynch Shriver

TWO federal government agencies regulate levels of chemical residues
in meat. How well does the enforcement system operate to protect con-
sumers, and what does it cost?

This thesis applies concepts of deterrence developed in economic
enforcement literature to evaluate the cost effectiveness of this enforce-
ment system. Standard operating procedures used by enforcers to create,
sort, and act on information about residues will be described. Resource
allocation among enforcement problems is assessed. Effects of the system
on violation rates are examined using econometric models.

The research indicates that direct costs are rarely imposed on vio-
lators. The U.S. Department of Agriculture's Food Safety and Inspection
Service finds condemnation of violative carcasses difficult and costly
due to current chemical testing requirements. Legal constraints make
prosecution of violators difficult for the Food and Drug Administration.
Consequently, increased sampling is uncorrelated with reductions in vio—

lations. Policy implications are also examined.

ACKNOWLEDGEMENTS

The contribution of Dr. Eileen van Ravenswaay to this research in
its design and implementation has been invaluable. A debt of gratitude
is owed her for her high standards and encouragement along the way.

Thanks are also due to committee members Jim Shaffer and Warren Samuels
for their constructive comments. Undying gratitude for his contributions
also goes to fellow graduate student Bill Rockwell who generously provided
many hours of advice and expertise.

Many individuals within the USDA assisted greatly through the provi-
sion of data. including Lecta Hensley and Mildred Macabee. John Spaulding
was extremely generous with his time and immense store of knowledge.

Clark Burbee of ERS provided important data and unflagging interest and
support.

The author wishes to thank the USDA and the Department of Agricul-
tural Economics at Michigan State University for their generous support
of this research.

Finally, without the confidence nurtured by the support. patience,
and encouragement of one's family. nothing could be accomplished. These
were provided unstintingly by my parents. brothers and sisters. and hus-

band Larry Lev.

ii

TABLE OF CONTENTS

Page
LIST OF TABLES .......................... vi
LIST OF FIGURES ......................... vii
LIST OF ABBREVIATIONS USED IN THESIS ............... viii
CHAPTER ONE: INTRODUCTION .................... l
1.1 Introduction ...................... 1
l.l.l Federal Standards for Chemical
Residues in Meat ................ 3
1.1.2 Enforcement of Tolerances for Chemical
Residues in Meat ................ 4
1.1.2.1 Monitoring and Surveillance ...... 5
1.1.2.2 Follow-up ............... 7
l.2 The Enforcement Problem ................ 9
1.2.1 Research Procedures ............... 12
1.2.1.1 Economic Theories of "Optimal"
Enforcement .............. 13
1.2.1.2 The Objective Function of the
Agency ................. 16
1.3 Plan of the Thesis ................... 18
CHAPTER TWO: THEORY ....................... 20
2.1 Introduction ...................... 20
2.2 Standard Operating Procedures ............. 23
2.3 "Rational" Enforcement ................. 27
2.3.1 Theories of Rational Enforcement ........ 27
2.3.2 Risk Attitudes and Compliance .......... 34
2.3.3 Limitations to "Rational" Enforcement ...... 37
2.3.4 Other Factors in the Compliance Decision
Affecting Optimal Enfbrcement .......... 38
2.3.4.1 Standard Setting ............ 39
2.3.4.2 Ethics ................. 40
2.3.5 The Infbrmation Problem and "Optimal"
Enforcement ................... 42
2.3.5.1 The Effects of Uncertainty on
Agency Decisions ............ 42
2.3.5.2 The Effects of Uncertainty/
Information Costs on Compliance
Decisions ............... 44
2.3.5.3 Implications of the Information
Problem for Program Evaluation ..... 46
2.3.6 Implications of a Normative Evaluation
Criterion .................... 48

iii

2.4 Conclusions ...................... 50
2.4.1 Factors in Violative Behavior .......... 51
2.4.2 "Efficient" Enforcement ............. 52
2.4.3 Effective Enforcement .............. 54

CHAPTER THREE: RESULTS PART 1: THE ENFORCEMENT STRUCTURE. . . . 56

3.1 Introduction and Methods ................ 56

3.2 FSIS Enforcement Programs ............... 61
3.2.1 Meat Inspection: Slaughter/Processing

Level ...................... 61
3.2.1.1 Surveillance Sampling Procedures:
Responsibilities at the Plant ..... 71
3.2.1.2 Implications .............. 75
3.2.2 Regional Level ................. 79
3.2.2.1 Routine Follow-up ........... 79
3.2.2.2 Contamination Response System ..... 86
3.2.2.3 Other Surveillance Programs ...... 87
3.2.3 National Level ................. 88
3.2.3.1 Monitoring Phase ............ 89
3.2.3.2 Surveillance Phase ........... 95
3.2.3.3 Objectives of Surveillance:
Information and Enforcement ...... 107
3.2.3.4 Enforcement and Information:
The Importance of Labelling ...... 109
3.2.4 Producer Education Programs ........... 110

3.3 FDA Responsibilities .................. 113

3.4 Conclusions ...................... 116
3.4.1 The Probability of Detection .......... 117
3.4.2 The Criteria for Surveillance .......... 119

CHAPTER FOUR: RESULTS, PART 2: FSIS RESOURCE ALLOCATION
AND ITS EFFECTIVENESS ...................... 123

4.1 Introduction and Methods ................ 123

4.2 Enforcement Program Resource Allocation ........ 126
4.2.1 The Monitoring Data ............... 127
4.2.2 Trends Among Substances and Substance

Groups ..................... 134
4.2.2.1 Hormone Monitoring ........... 134
4.2.2.2 Extra-Monitoring Enforcement

Programs for Hormones ......... 136
4.2.2.3 Sulfa Monitoring ............ 136
4.2.2.4 Extra-Monitoring Enforcement

Programs for Sulfa ........... 138
4.2.2.5 Antibiotic Monitoring ......... 141
4.2.2.6 Other Enforcement Programs for

Antibiotics .............. 144
4.2.2.7 Chlorinated Hydrocarbon Monitoring. . . 145
4.2.2.8 Extra-Monitoring Enforcement

Programs for CHC's ........... 150
4.2.2.9 Organophosphate Monitoring ....... 151

iv

Page

4.2.2.10 Trace Element Monitoring ....... 152
4.2.2.11 Conclusions .............. 153

4.2.3 Monitoring Sample Sizes and Statistical
Validity .................... 154
4.3 The Relationship Between Sampling and Violations. . . . 157
4.3.1 Introduction: Description of the Model ..... 157

4.3.3.1 Alternative Model Specifications . . . 159
4.3.2 The Results of Ordinary Least Squares

Regression ................... 161
4.3.3 Interpretation of the Results .......... 164
4.3.4 The Limitations of Regression Models ...... 167
4.4 Conclusions ...................... 169
CHAPTER FIVE: REVIEW OF CONCLUSIONS AND POLICY IMPLICATIONS. . . 171
5.1 Introduction: Objectives ............... 171
5.2 Enforcement Programs .................. 171
5.2.1 Defining the Residue "Problem" ......... 171
5.2.2 The Costs of Residue Violations ......... 173
5.2.3 Enforcement via the "Market" .......... 174
5.2.4 Indirect Enforcement: Persuasion vs.
Penalties .................... 175
5.2.5 The Value Judgements of Enforcers ........ 176
5.3 Policy Implications .................. 178
5.3.1 Imposition of Real Costs on Violators ...... 178
5.3.1.1 Tracing ................ 178
5.3.1.2 A Change in Sampling Approach ..... 179
5.3.1.3 Enabling the System to Impose
Penalties ............... 180
5.3.2 Statistical Assistance ............. 181
5.3.3 Improving Technology .............. 181
5.4 Limitations of the Research .............. 182
5.4.1 The Costs and Benefits of Proposed Changes . . . 182
5.4.2 The True Determinants of the Compliance
Decision .................... 183
5.4.3 The Role and Activities of FDA in Enforcing
Tolerances ................... 183
APPENDIX ............................. 184
BIBLIOGRAPHY ........................... 187
LIST OF GOVERNMENT OFFICIALS AND INDUSTRY REPRESENTATIVES
INTERVIENED ........................... 192

LIST OF TABLES

Table Page
3.1 Percentage of Meat Animals Sampled at Slaughter ..... 70
3.2 Number of Samples Collected Under National Residue

Monitoring Program 1973-1982, Selected Species. 91-

by Calendar Years .................... 93
3.3 Factors in FSIS Resource Allocation Decisions:

Initiation of Surveillance ............... 97
4.1 Residue Monitoring Sample Sizes by Substance 128—

Group and as a Percentage of Total 1973-1983 ...... 131
4.2 Percent of Samples Tested for Selected CHC's

Which Showed Positive Results .............. 146
4.3 Alternative Model Specifications ............ 162

vi

0000030000

.4a
.4b

LIST OF FIGURES

Creation and Regulation of Residues in Meat ......

Food Safety and Inspection Service, U.S. Department

of Agriculture .....................
Form 6000-2 ......................
List of Substances Tested for by FSIS, With Codes . . . .
Form 6000-1, Side One .................

Form 6000-1, Reverse ..................

Letter to Producer Informing Her that Testing

Revealed a Below Tolerance Level Residue ........

Letter to Producer Informing Her that Testing

Revealed a Violative Level of Residues .........

Letter Informing State Department of Agriculture

of a Violation .....................

Letter Informing a Producer who is a Previous
Violator that Results of Follow-up Surveillance
Demonstrate the Continued Presence of a Residue

Problem ........................

Percent of Total Monitoring Samples Tested for

Selected Substances or Substance Groups ........

Violation Rates for Selected Substance Groups .....

Antibiotics in Cows and Calves: Sample Size

and Violation Rate ...................

Percent Positive vs. Sample Size: CHC in Cows .....

vii

Page
8

6O
67
68
73
74

82

83

85

133
135

143
147

BVM
CHC
DES
EPA
FDA

FFDCA
FIFRA
FSIS
FSQS
GAO
HHS

MPI
0P
OTA
REPD

SOP
STOP
SST
USDA

LIST OF ABBREVIATIONS USED IN THESIS

Bureau of Veterinary Medicine of FDA
Chlorinated Hydrocarbons
Diethylstylbestrol (hormone)
Environmental Protection Agency

Food and Drug Administration of the U.S. Department of
Health and Human Services

Federal Food, Drug and Cosmetic Act

Federal Insecticide, Fungicide and Rodenticide Act

Food Safety and Inspection Service of USDA (formerly FSQS)
Food Safety and Quality Service

Government Accounting Office of the U.S. Congress

U.S. Department of Health and Human Services (formerly
Health, Education and Welfare)

Meat and Poultry Inspection Division of FSIS
Organophosphate Pesticides
U.S. Congress Office of Technology Assessment

Residue Evaluation and Planning Division of FSIS (commonly
referred to as "Residue Division")

Standard Operating Procedure
Swab Test on Premises
Sulfa STOP

U.S. Department of Agriculture

viii

CHAPTER ONE
INTRODUCTION

1.1 Introduction

The existence of potentially harmful chemical residues in food for
human consumption is a source of increasing concern to government and
consumers, and a continuing focus of media attention. The U.S. General
Accounting Office (GAO) has indicated concern particularly over the
amount of meat containing potentially harmful drug. hormone, and environ-
mental contaminant residues which reaches consumers. It has estimated
that the proportion of meat reaching U.S. consumers containing dangerous
and illegal residues has averaged 7.4% between 1979 and 1981, and reached
as high as 14.3% in earlier periods (1974-1976).]

Concern over chemical residues has arisen over their known or sus-
pected potential for posing a threat to human health. Some of the sub-
stances in question are linked with the incidence of cancer, birth de-
fects. and mutations. Others cause severe allergic reactions among sen-
sitive individuals at minute dosages. Use of antibacterials in animal

feed is also felt by an increasing number of scientists to contribute to

 

1U.S.. Congress. General Accounting Office. Monitoring and Enforcigg
Food Safety--An Overview of Past Studies, GAO/RCED-83-153, September 9
1983. p. 49. These rates do not correspond with USDA calculated "viola-
tgontrates"; this issue will be discussed in more detail in a later
c ap er.

the creation of drug resistant bacteria which can infect humans, and
prove difficult to treat with commonly used drugs.1

Several important questions are raised by the GAO's findings con—
cerning the allocation of public resources to the enforcement of federal
standards on chemical residues in meat. Is the apparently high level
of violation of these standards due to ineffective design of enforce-
ment programs, or is the level of resources allocated to such programs
merely inadequate?

In order to answer these questions. this thesis examines existing
federal programs for enforcing standards for chemical residues in meat
and the enforcement outcomes produced. We will describe the programs,
enforcement tools and resources, and factors in the enforcement environ-
ment which affect or limit agency control over outcomes. Theories of
enforcement will be examined which describe effective enfbrcement stra-
tegies. and prescribe the structure of incentives which the agency
should strive to create. Programs implemented by the Food Safety and
Inspection Service of the U.S. Department of Agriculture (FSIS and USDA)
to control the level of residues present in meat will be assessed ac-
cording to this standard.

The remainder of this chapter describes the nature of the standards
whose enforcement is in question, the jurisdictions and responsibilities
of the set of agencies involved, the enforcement problem faced by these

agencies, and the procedure to be used in this thesis for answering the

 

1
U.S.. Congress, Office of Technology Assessment (OTA). Dru s in
19791.

Livestock Feed. (Washington, D.C., Government Printing Office.
pp. 5-6. See also: Eliot Marshall, "Scientists Endorse Ban on Anti-
biotics in Feeds," Science. Vol. 222 #4624, Nov. 11 1983.

questions posed. Also. definitions and explanations of some terms will
be offered in order to facilitate the reader's comprehension of certain
aspects of agency enforcement duties. although detailed descriptions of

their operationalization will be delayed until Chapter Three.

1.1.1 Federal Standards for Chemical Residues in Meat

Not all residues occurring in meat are illegal or considered harm-
ful. It is the responsibility of the Food and Drug Administration (FDA)
of the Department of Health and Human Services (HHS) to issue regula-
tions defining the levels of drug residues and environmental contaminants
which will be tolerated in meat subject to Federal inspection.1 FDA
must also approve drugs for safety and efficacy before they can be used
in food producing animals, and ensure that animal feeds comply with
legislated limits on levels of drugs. pesticides, and environmental
contaminants which may occur in the feeds.2

In carrying out its responsibilities, FDA may set "tolerance" levels
and "withdrawal periods" for substances. A tolerance is the maximum
residue of a substance permitted in food, usually measured in parts per
million or per billion. A withdrawal period is the length of time during
which a drug or pesticide may not be intentionally used on an animal be-
fore it is slaughtered. If an animal grower complies with a legislated

and labelled withdrawal period for a particular drug (and it is not re-

ceived by the animal from some other source). chemical analysis of the

 

1Federai Food. Drug, and Cosmetic Act (FFDCA), U.S. Code, Vol. 21,
Sec. 301 et seq, ‘1938 .

2U.S., Congress, Government Accounting Office, Problems in Prevent-
in the Marketing of Raw Meat and Poultry Containing_Potentially Harmful
Res dues, HRD-79-10. April 17 1979. pp. 18-19.

animal tissue should not reveal residues of that drug in excess of the
tolerance level. A tolerance is not necessarily a "safe” level, but
one which has been found to pose an "acceptable" risk to human health,
the type and level of any toxicity having been taken into account.

While FDA sets tolerances for drugs and toxic substances in food,
the Environmental Protection Agency (EPA) is responsible under the
Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA)1 for setting
tolerances for those substances. EPA also approves their use on animals

and sets withdrawal periods.

1.1.2 Enforcement of Tolerances for Chemical Residues in Meat

The third, and most important (from the perspective of this thesis)
agency sharing responsibility for the control of residues in meat is the
Food Safety and Inspection Service Division (FSIS) of USDA. This sec-
tion will delineate FSIS's jurisdiction from that of the other agencies,
and briefly introduce the "monitoring" and "surveillance" programs
through which it implements its enforcement responsibilities.

While FDA is responsible for setting tolerances and for enforcement
at the farm level, USDA has historically held inspection jurisdiction
over marketed raw and processed meat and poultry products. That is,
USDA's responsibilities begin at the point of slaughter and continue
through the processing stages, while on-farm inspections of live food
animals and the conditions under which they are raised are carried out
by other agencies--usua11y FDA, state departments of agriculture, or

EPA (in the case of toxic substance contamination). This division of

 

1Federal Environmental Pesticide Control Act, U.S. Code, Vol. 7,
Sec. 36, 1972.

responsibility has important implications for enforcement of chemical
residue tolerances which will be discussed later in this chapter and
throughout the remainder of the thesis.

USDA has inspected slaughter and meat processing facilities for
sanitation and wholesomeness of the product since 1906, under the
Federal Meat Inspection Act (34 Stat. 669). Formal testing for chemical
residues, however, did not begin until 1967, when a small testing pro-
gram for antibiotics and chlorinated hydrocarbons was initiated with
the analysis of some 1,000 samples of meat and poultry. Monitoring for
these substances has been broadened and restructured since then to com—
prise the collection and analysis of 25,000 or more samples annually

for 50-70 substances.1

1.1.2.1 Monitoring and Surveillance

Two levels of sampling and testing are used by FSIS in performing
chemical analyses of meat and poultry. The monitoring phase is designed
to randomly select among slaughterhouses subject to federal inspection
a group from which in-plant inspectors will collect samples of animal
tissue. Samples are sent for analysis to one of three FSIS laboratories
located in San Francisco, St. Louis, and Athens (Georgia). The results
of these analyses are compiled at FSIS headquarters in Washington, D.C.,
and used to give the agency a picture of the "violation rate" for each
of the substances tested for, and of the overall "violation rate." By

violation rate we mean the percentage of all animals sent to slaughter

 

1Each individual sample may be analyzed for a particular substance
or a group of related substances; that is, not every sample is analyzed
for all of the 70 substances FSIS has the capacity to test for.

at federally inspected establishments which contains illegal (above
prescribed tolerance limits) residue levels.

If FSIS has reason to suspect that any food animal brought to
slaughter contains illegal residues, it may switch over to a "surveil-
lance" phase. These reasons include previous vidlations on the part of
the producer of the animals, characteristics of the animals themselves
(such as visible injection sites or staggering), or the existence of a
known point source of chemical contamination in the area where they were
produced. Surveillance for chemical residues implies a change in sam-
pling intensity and/or frame; that is, particular animal growers, man-
agement systems, states, or regions may be selected for stepped up sam-
pling when one of FSIS's information sources indicates the existence of
a residue problem.

The objective of the monitoring system is, as mentioned above, to
give FSIS a general picture of the amount of meat which reaches consumers
containing Violative residues. Similarly, the objective of surveillance
is to provide FSIS with more detailed information about the source and
extent of a problem which has been turned up, either by the monitoring
system or some other source of information. However, these programs
also serve enforcement functions since the information that they gener-
ate may be used fur follow-up investigations designed to remove any
source of potentially dangerous residues in meat and to prevent such a
source from recurring or continuing to contaminate the food supply, by

condemnation of tainted carcasses.

1.1.2.2 Follow-up

Investigation, upon which further enforcement action rests, moves
back within the jurisdiction of the FDA. Violative residues usually
indicate a faulty production practice on the part of the animal grower.
Since USDA's inspection responsibilities are limited to slaughter and
processing facilities, it must provide FDA With data derived from the
samples taken at slaughter. Regional offices of FSIS attempt to trace
test results back from the point of slaughter to discover the origin of
any animal discovered to be Violative, in order to facilitate FDA's on-
farm investigation, but are not always successful.1

If the producer can be traced, FDA officials may visit the farm
to investigate in person, or conduct an inquiry by letter. If the
source of the residue can be identified, FDA may either let the pro-
ducer off with a warning to discontinue the practice which is causing
the Violative residues in her animals, or seek a consent order from the
courts to stop her from doing so. If the producer then violates the
consent decree, FDA may bring suit in the courts for violation of the
decree. This is not an option which FDA has as yet chosen to exercise
in a residue case.2

In order to provide a clear picture of the flow of events and di-

vision of responsibilities in enfbrcing tolerance levels for chemical

residues in meat, a diagram is provided (see page 8, Figure 1.1). To

 

1Interview with Dr. F. R. Rellosa, U.S.D.A. Northeast Regional
Office, September 12,1983. For more details on interviews, please re-
fer to List of Interviews, following Bibliography.

2Interview with Mr. Homer Ransdell, F.D.A. Compliance Division,
September 14,1983.

aemz cw mwzuwmom mo cowuepsmox was coeueosu

F._ eraser

 

mmauwmwt m>wuepow> monquoL o>wuepop> o:
msozm memxpecm xLoueLoamo .am mzo;m memxpoeo atouegonuo .um

 

          

cwezmcou o» >F~e=_w use
soprano; .Lo—ammpocz op upon new: .c

 

 
     

smuacosa op count»
Fmewcm m>wuo~om> .m

        

 

 

  

umpmamwsll
“a: Peewe<

Seance; m:_gouwcoe

.cowue_0w> co to; :Pstowcm
Eocene cw uwFaEem peewc<

<am: an smuump ucmm smuauoga
.<oa op uw_Paa=m coopestoecH .N

 

   

   
 
 

.am

  
       
  

 

 

 

 

m L om mzuwmmc mo cowue:
-ms+_o ucm cowuemwumw>cw
Eco» :o ”a: mzoF_oe <ou .mm

Lopgoampm o»
use» Feswce =o>vpepov>= .~

 
 

 

 

  

 

 

 

 

mozuvmog oveuwumon Lou

_m>m_ smuzuoeq mmauwmws mace Fes_=e com
=o>wum_o’>= mocwmov <am

m w: PM?» 0 wwamo
mw cowwao_wwW:wuw o: =w>wpepor>= mmepmwu <oa
:5 30:8 8e 88 «E .8] kekamaae e

 

 

sea. mmwwmae
s was mouumsu covuuu connect; .—

 

summarize the set of agency actors and their roles: EPA sets tolerances
for pesticides and other toxic substances and environmental contaminants
under FIFRA. FDA approves and sets tolerances for food animal drugs,
and performs on-farm follow up investigations when evidence exists of

a production related source of Violative residues. FSIS, a division of
USDA, inspects food animals from slaughter through processing stages

for chemical residues, and provides infbrmation about its results to

FDA for follow-up investigation.

1.2 The Enforcement Problem

In order to understand the way in which FSIS has chosen to imple-
ment its enforcement structure and to make any judgement about whether
its program meets some standard of "effectiveness," we need to under-
stand something about the producer behavior the agency seeks to control.
What is it that FSIS has to overcome in order to ensure a "safe" or
uncontaminated meat supply to the consumer? Put differently, why do
residues occur--what types of producer actions can create Violative
residues?

An important source of chemical residues in meat is low or "sub-
therapeutic" application of drugs mixed with feed, in pellets under the
skin, ear implants, or the water supply of food animals. This may be
done as nutritional supplementation, disease treatment, disease preven-
tion, to promote rapid weight gain, and/or to improve feed efficiency.
Drugs used for these purposes include antibacterials such as antibiotics
(e.g. tetracyclines and penicillins) and sulfas (sulfamethazine, sulfa-
thiazole), as well as estrogenic compounds such as diethylstilbestrol

(DES) (now banned for this use) and zeranol (the active ingredient in

10

Ralgro). Although the precise way in which these substances work to
promote weight gain is not yet fully understood or agreed upon, almost
all poultry, 90% of swine and veal calves, and 60% of cattle produced

1 When withdrawal periods are not observed

in the U.S. receive them.
for these medications, either intentionally or unintentionally, viola-
tive residue levels in animal tissue may result.

Accidental feed contamination incidents have also been a common
cause of Violative residues. Well publicized examples include the P88
(polybrominated biphenyl) incident in Michigan, when fire retardant
chemicals were accidentally mixed with cattle feed, and again in 1979
when an unused electrical transformer leaked PCBs (polychlorinated
biphenyls) into poultry feed ingredients which were subsequently dis-
tributed throughout several western states. The presence of contaminants
in groundwater consumed by livestock may also result in illegal residues.
Failure to use good manufacturing practices suggested by EPA with re-
gard to timing and extent of pesticide use, carelessness and improper
storage of toxic substances may all have similar results.

The characteristic common among all of the three classes of contam-
inants of meat--drugs, pesticides, and other toxic substances--is that
the source of the residue is generally production system related. That
is, the source is the practices of the animal grower;feed manufacturer,
or dealer, rather than those of the slaughterer, where the Violative
residue is discovered. Further, a source such as feed containing medi-

cation or toxic substances is likely to have affected an entire lot of

 

1OTA, Drugs in Livestock Feed. pp. 3-4.

11

animals. While only one of the lot may have been randomly chosen for
sampling, a single Violative result may indicate that many other animals
are affected.

The only way that the exact magnitude of an individual residue
problem can be determined is to trace the producer through slaughter-
house purchase records--if they are kept--and perform an investigation
on that producer's premises which may indicate the number of animals
potentially containing the Violative residue. The tracing process is
difficult in the absence of a comprehensive animal identification sys-
tem; in addition, during the ten day period required to complete the
chemical analysis and communicate the results such animals are likely
to have been sold, slaughtered, broken up into retail cuts, and con-
sumed.1

We may therefore define the "enforcement problem" as follows: how
can the enforcement agency best affect producer decisions? Further we
can narrow down the broad field of producer decisions to the category
of decisions which affect residue levels in the meat which they produce.
Since the entirely accidental contamination incident is beyond the realm
of the producer's decisions, it is also beyond the power of the agency
to affect. (By "accidental," we refer only to those incidents which
the producer has no control over. This does not include incidents
caused by carelessness, inattention, or ignorance of regulations and

labelling instructions for agricultural chemicals, which could be

 

1The exception to this general rule occurs when visible evidence
of contamination exists on a carcass, which can lead to condemnation.
See Chapter Three, section 3.2.1.

12

prevented by producer actions-~we will adhere to the tenet that "igno-
rance of the law is no excuse.") Such producer decisions will be termed
"compliance decisions," because they involve a choice about whether or
not to comply with or obey laws, regulations, and government suggested
practices. From a broader perspective, the prevention of chemical
residues in meat is a question of the type, how can compliance best be

created by an enforcement agency?

l.2.l Research Procedures

The general objective of this research is to arrive at an answer
to some of the questions posed above. In coming to conclusions about
whether FSIS behavior and resource allocation decisions are efficient
or effective, the first step is to develop criteria or standards of
efficiency and effectiveness; in other words, we require a picture of
what effective enforcement programs would look like if they did exist.

In order to develop such a standard of what effective enforcement
should be, both normative and descriptive theories of regulatory en-
forcement and decision making will be examined. Normative economic
theories have focused on issues such as defining what optimal levels of
enforcement are when: (1) social costs of enforcement and the costs
and benefits of violation to various members of society are taken into
account; and (2) a marginal analysis approach is taken. Descriptive
theories of organizational decision making provide useful clues to ac-
tual agency goals--rather than making assumptions about what these goals
are or should be--making it possible to examine the implications of

existing agency practices.

13

Our interest in the decision making processes of both producers
and the enforcement agency stems from the fact that it is the interrela-
tion of these groups' decisions which result in the chemical residues
found in meat. Agency decisions form the structure of enforcement faced
by producers. Conversely, producer decisions about whether or not to
comply with regulations are what the agency is trying to affect through
its deterrence strategies. Enforcement outcomes result from the crea-
tion of incentives by FSIS and FDA which influence producer behavior,
presumably changing it in a way desired by the agencies (as representa-
tives of the "public interest”). To identify the appropriate incentives
we need to understand how producers make drug-use decisions. To ob-
serve the incentives which actually prevail we need to understand
agency decisions. The economic theories of regulatory enforcement and
behavioral decision theories mentioned above will serve as a vehicle
to predict and describe these decision processes and their results.
The usefulness and validity of these theories when applied to the en-

forcement situation for chemical residues in meat will also be assessed.

l.2.l.l Economic Theories of "Optimal" Enforcement

Economic models of producer compliance decisions conclude that
the decision to comply with laws is made when expected benefits of such
compliance exceed the expected costs of taking the necessary actions to
bring oneself into compliance. These "expected benefits" can, conversely,
be regarded as avoiding the expected costs of noncompliance; in other
words, avoiding the payment of penalties or fines which might be in-
curred if one were detected in the act of noncompliance by the enforc-

ing agency. Where illegal residues in meat are the outcome, the

14

compliance decision may be about whether or not to withdraw a lot of
animals which is soon to be slaughtered from a medicated feed at the
stipulated time. The producer's expected costs of noncompliance (i.e.,
failure to withdraw the animals from the medicated feed) are the pro-
duct of the value of the fines or punishment which could be imposed,

and the probability that they will be imposed. The expected benefits
of violating residue standards are reduced mortality and morbidity dur-
ing the final period before slaughter, as well as any extra weight gain
among the animals in the lot attributable to the medication in the feed.
Therefore appropriate incentives for the agency to create are those
which will offset the producer's perceived expected gains from viola-
tion. In achieving this incentive structure, the agency may either
raise expected costs of violation of the law concerning the withdrawal
period for the medicated feed, or lower the expected benefits of violat-
ing (e.g., by imposing a tax on drug use), until it observes (or pre-
dicts) that expected costs of the Violative act exceed the benefits.1
The economically "rational" producer will then be motivated to comply,
and the agency will have achieved its broad objective of creating com—
pliance.

Since "expected costs" are the product of a probability and a po-
tential penalty, raising either element should affect producers in the
desired manner. In the residue case, the probability of penalties being
imposed upon violators is dependent upon successful outcomes of a series

of enforcement actions, carried out by more than one agency. The

 

1Current law prevents the imposition of such a tax and substitutes
have therefore not been developed.

15

violative animal must first be chosen for sampling and chemical analy-
sis. The analytical procedure must then successfully demonstrate the
presence of an above tolerance residue. If both of these events occur,
the agency must then decide to act or follow up on the information it
has, and succeed in both tracing the Violative producer and imposing
whatever penalty it decides is appropriate.

Available enforcement tools then fall into several broad catego-
ries, namely those which can affect: (1) the level of the penalty im-
posed on violators; (2) the probability of detecting the violation and
imposing the penalty; and (3) the perceptions held by the regulated
group of both of these things. The third category is important because
even if enforcers were able, for example, to raise the probability to
.99 of detecting a violation, unless the potential violator perceives
that the probability is high, she will not be motivated to act in the
way desired by enforcers. Furthermore, in making resource allocation
decisions, enforcers must operate under some assumptions or theory
about the relative responsiveness of the regulated group to the level
of penalty versus the probability of detection.

Assuming that enforcers wish to maximize the amount of deterrent
effect for each dollar spent on enforcement and that producers respond
more to a dollar's worth of effort used to raise the probability of
detection than they will to a dollar's worth of raising the level of
fines, the rational enforcer will allocate the marginal dollar to rais-
ing probabilities. To do this she "fight increase the intensity of sam-

pling.

16

Since the expected benefits from different kinds of violations can
be expected to vary, enforcers should accordingly vary the level of ex-
pected costs for different violations. Enforcers attempting to operate
in an economically efficient manner could choose and implement an opti—
mal level of enforcement by allocating resources so as to maximize the
deterrent created within their resource constraint. Therefore, we
should observe the "rational" agency allocating the most resources to
programs designed to raise the expected costs of violations which yield
the greatest expected benefits to producers contemplating commission of
the Violative act. Or, it should spend its resources in such a way as
to deter violations which cause the greatest harm to society: in this
way, the greatest social benefit is derived from the expenditure of the
marginal enforcement dollar. For example, such characteristics as the
level of hazard associated with a particular chemical and the ease of
deterring its improper use (that is, the less that has to be spent to
prevent each dollar's worth of harm to society) should determine the
level of spending in each case. Factors which affect the ease of de-
terring improper use include the availability of substitutes for the
substance, lack of industry opposition or even positive industry cooper-
ation with agency efforts on certain substances, a low cost testing
method, a small or concentrated industry (making tracing easier, and
facilitating intra-industry communication about costs of violations),

and other situational components.

1.2.1.2 The Objective Function of the Agency
Upon closer scrutiny, the assumption that maximization of total

"social welfare" inherent in the marginal cost/benefit calculation

17

described above is the agency's primary objective may appear unrealistic,
and if so, misleading. Agencies, as well as firms, operate within moti-
vational structures, but incentives affecting agency decision makers

may be fundamentally different than those affecting the (profit maxi-
mizing) firm.

The perspective of the individual agency decision maker differs
from that implied in the "optimal" analysis. Global optimization is
not possible from within each agency or division of an agency, nor
would there normally be any incentive for an official within an agency
to strive to achieve such an optimum. Setting optimal penalties and
probabilities of detection requires not only a normative decision about
how to measure the "social value" of crimes prevented by various en-
forcement programs, but also an across-agency comparison of these de-
terrence programs in order to prioritize the allocation of resources
between them. This procedure also implicitly requires knowledge of the
effectiveness of different programs—-that is, how much crime will be
deterred by each--in order to determine the social value of deterrence
programs. This overall goal setting is the function of Congress and is
accomplished via budget allocations and legislation controlling the
mandate of the regulatory enforcement agencies.

In assessing an individual agency's performance according to the
criteria proposed by such theories as Becker's and Stigler's (examined
in the next chapter), one must make certain adjustments to correct for
point of view. We are dealing with issues of how FSIS spends its re-
sources, not how resources are allocated between FSIS and other agencies

by Congress. In a certain sense, we must view the agency's resources

18

as a fixed constraint within budgetary periods. The issue becomes,
does the FSIS use its (fixed) resources in an effective way? If not,
why not? For this reason, we must inquire into the nature of FSIS ob-
jectives rather than simply accepting assumptions about what they
"should" be as offered by economic theorists.

Since it is not clear that "optimal" theorists' assumptions about
agencies' objectives are true, to proceed as if they were true and
assess the outcomes of agency activities accordingly would neglect im-
portant information. Therefore the first step of this research is to
delve into the empirical question of what FSIS goals actually are. In
this way, when outcomes are assessed according to the standards pro-
posed, it will be clearer why certain failures or successes are observed.
Following this procedure will make the policy implications of the re-
search clearer, because specific aspects of agency procedures which

might be improved will have been described and analyzed.

l.3 Plan of the Thesis

Chapter Two will describe in more detail the optimal theories of
regulatory enforcement which have been touched upon in this introduc-
tory chapter. It will also explain which aspects of agency behavior
need to be researched in an effort to determine what FSIS objectives
are. These include standard operating procedures for sorting informa-
tion and delegating decision making authority.

The results of this research are divided into two sections, com-
prising Chapters Three and Four. Each of these begins with an intro-
ductory statement of its objectives and the methods used to accomplish

the objectives.

19

The primary distinction between the subjects of the two results
chapters is as follows: Chapter Three describes and analyzes proce-
dures. Chapter Four examines outcomes of FSIS residue tolerance en-
forcement programs. Procedures are described in order to develop an
understanding of how FSIS has gone about setting up an enforcement
structure, and the strengths and weaknesses of certain functions. Al-
so included is a discussion of the criteria mentioned by the Residue
Division decision makers as guidelines to resource allocation.

Chapter Four uses data on monitoring and other enforcement pro-
grams to describe what the results of FSIS resource allocation have
been. Of particular interest are issues such as what the existing allo-
cations reveal about agency objectives, as distinguished from agency
officials' claims about these objectives. We will examine how these
considerations are related to issues of "social harm" and its measure-
ment.

A test of the hypothesis "the probability of detection affects the
amount of violation which occurs" is also included in Chapter Four.

The purpose of the test is to determine whether changes in the sampling
rate are an effective enforcement tool. (Since no variation in penal-
ties imposed by the enforcement structure has been observed, it will not
be possible to examine whether this aspect of expected cost has such an
effect.)

The final chapter will provide conclusions drawn from the research
about the effectiveness of FSIS enforcement programs, and the overall
enforcement system for residues. Some policy implications will also be

discussed.

CHAPTER TWO
THEORY

2.1 Introduction

Several approaches are possible to the issues involved in evalua-
tion of an enforcement program. Frequently, it is stressed that an
objective considered to be of primary importance to the evaluator must
be chosen against which to measure the success of the agency's program.
The first step in taking such an approach is to select a set of stan-
dards which reflect what the evaluator feels should be the goal of an
enforcement agency, and a method of measuring the agency's attainment
of these standards. After completion of the measurement process, the
evaluator may then make a normative conclusion about whether or not the
agency's programs have lived up to the set of standards imposed.

A second approach is to attempt to describe the methods used by
the agency to implement its programs--in other words, to discuss the
ways in which the agency makes the resource allocation decisions which
determine the various aspects of the enforcement structure facing poten-
tial violators of the law. Then, the implications of these decision
making procedures can be discussed for various subjects of interest to
the evaluator. For example, the way in which costs and benefits, poli-
tical power, or future streams of income will be distributed through
this decision making process can be assessed. This approach can be
distinguished as "descriptive," in contrast to the more "normative"

approach described above.
20

21

The objective of this chapter is to describe in some detail what
each of these approaches entail, and the strengths and limitations of
each.

The normative criterion chosen by the theorists reviewed here,
and adopted (with some reservations) for the remainder of the evalua-
tion, is based on economic theories of "optimal" or "rational" enforce-
ment. Becker, Stigler, and others analyze the meaning of the term "ef-
ficiency" when applied to the regulatory enforcement situation. En-
forcers, they claim, should act in a way similar to firm decision
makers; while firms attempt to maximize profit subject to resource con-
straints, enforcers should expend their limited resources in a way which
guarantees deterrence of the most (or some satisfactory level of) harm
to society. In order to assess the efficiency of the agency in accom-
plishing this goal, actual resource outlays and implemented programs
must be studied.

Descriptive theories of decision making point out that decisions
within an organization are frequently made according to standard operat-
ing procedures (SOP's). SOP's are methods by which organizations deal
with the many sources of information available which may bear upon de-
cisions, sort this information, and deliver it to the individuals within
the organization, who require it in order to make decisions. They also
determine which decisions are made at the various levels and divisions
within the organization. In order to assess the implications of the pro-
gram choices which are made by the agency, one must look at the process
by which such decisions are made. An examination of the SOP's of the

agency will provide this information. The purpose of this examination

22

is to describe actual enforcement agency SOP's, rather than to predict
what they will or should be.

The use of efficiency as a criterion for evaluation of enforcement
agency performance, as described in the theories reviewed here, suffers
from certain limitations. Firstly, it requires a determination of the
"value" of the social harm prevented by agency deterrence strategies.
In the meat residue case, this process may be difficult to accomplish
in a quantitatively rigorous fashion. An important issue here is, what
determines the extent of harm society is subjected to by the presence
of residues? Is it the level of toxicity of a particular substance,
or the total amount of exposure that should be considered in this
evaluation?

Secondly, in order to design programs which deter the most harmful
of violations, we require some foreknowledge of the precise reactions
of those considering violation of the law to the various elements of
an enforcement program. The attitude of the regulated group toward the
risk of detection in violation is important in this calculation, and
equally difficult to determine with precision.

Thirdly, and very importantly, the "rational" enforcement theories
neglect important structural limitations on the ability of individual
agencies to determine overall probabilities and costs of detection. A
discussion focused broadly on the total expected cost issues misses the
point that a constructive evaluation must concern itself with the pos-
sible as well as the most efficient.

For all of these reasons, the "rational" enforcement theories fail
to yield clear, precise assessment criteria or predictions of what the

evaluator will observe. In this situation, the approach will be to use

23

the conclusions of the economic theories as broad, general factors of
success according to a standard of economic efficiency, while also pro-
viding a detailed description of the decision processes by which the
enforcers within a single agency contribute to the enforcement struc-

ture (SOP's) and a discussion of their implications.

2.2 Standard Operating Procedures

To what extent are individuals capable of comprehending, processing,
and acting "rationally" on the basis of the information which is avail-
able to them? Simon (1961) claims that human decision making is "in-
tendedly rational, but only limitedly so" due to the neurophysiologi-
cal and language limitations of human beings.1 In fact, one of the rea-
sons organizations are formed is so that the limited cognitive capacity
of individuals can be overcome. Shaffer also emphasizes that people--
consumers, producers, and other actors such as regulators--develop SOP's
and similar techniques as a way of economizing on the time and effort re-
quired for many decisions, because they are incapable of accurately pro-
cessing all available information for each decision.2 Therefore an un-
derstanding of decision making in an organization (such as an enforce-
ment agency) requires knowledge of the SOP's used by the agency to uti-
lize information in making decisions.

When firms, agencies, and individuals develop SOP's in response to

limited cognitive capacity, rather than reacting with complete

 

1Herbert Simon, Administrative Behavior, cited by Oliver Williamson,
Markets and Hierarchies, (New York: MacMillan and Co., 1961). p. 21.

2James Shaffer, "Notes for a Theory of Personality for a Theory of
Consumer Behavior," unpublished, no date.

 

 

24

flexibility to every situation, they reduce the amount of time, effort,
and resources which must be devoted to finding a solution to each problem
that arises. However, although "they permit the organization to deal
more effectively with previously experienced situations...they normally
retard adjustment to strikingly different situations" (Cyert and March,
1963, p. 107). Cyert and March maintain that because organizations
which use SOP's do not adapt immediately to new situations, and because
individuals and groups within organizations have goals which conflict,
the maximum possible level of any single goal will not be achieved.
Instead, some satisfactory level of the goal will be accepted within the
organization as a target; this process is known as "satisficing" in con-
trast to the assumption often made in economic theory that firm behavior
is "maximizing" in nature.

However, the observation that SOP's are used does not require ac-
ceptance of the "satisficing" assumption per se. Use of SOP's by organ-
izations may be thought of as a method of taking into consideration the
costs of collecting and processing information into a usable form, as
well as reducing the transaction cost of decision making on a day to day
basis. If these saved costs could be calculated, it might be discovered
that use of SOP's saves more (in information processing and decision
costs) than it "costs" in reduced flexibility to react to new situations.
That is, SOP's are a way of maximizing when information is costly. If
this is the case, the assumption that organizations such as firms and
agencies strive to maximize their achievement of the goals they have set
for themselves can still be defended. (The ”goals" referred to here
are the most important elements of the organization's utility function;

for a firm this is usually assumed to be profit. For the agency, the

25

theories reviewed in the next section of this chapter maintain that the
goal should be "efficient" creation of deterrence.)

Regardless of where one stands on the issue of satisficing vs.
maximizing, SOP's provide important clues about decision making within
an enforcement agency. They provide the criteria which are used to
choose between one option and another at various levels of the agency.
Since the options chosen create the enforcement programs which affect
the regulated group, a major task of this thesis is to establish what
these criteria are in order to evaluate the success of the agency's
decision outcomes.

A further implication of the existence of SOP's as decision rules
is that the enforcement agency must be aware of what the SOP's of the
regulated group are in order to effectively create deterrence. In ef-
fect, what the agency attempts to do is to cause the firms which it
regulates to adopt procedures and practices which do not include, or
even actually prevent, Violative behavior. In the meat residue case,
this means that FSIS should try to change livestock producer manufac-
turing practices which cause excessive residue levels in meat. These
practices include failure to observe withdrawal times for medicated
feeds and careless storage and disposal of pesticides and other toxic
substances, among others.

An especially important category of producer practices for the en-
forcer is the extent to which the consequences of Violative behavior
figure into the compliance decision. For example, if the probability
of detection is .01, and the fine which will be imposed upon detection

is $1,000, the expected costs of violation are $10 (.01 x 1,000 = 10).

26

If the expected benefits of violation are less than $10, we would expect
the risk neutral decision maker to choose compliance over noncompliance.
If so, enforcement resources have been effectively allocated.

However, what would be the effect of a producer SOP or decision
rule which said, "Many low probability events exist in the environment
which might reduce my profits. The time required to find out about
all of these risks is too great; therefore, I will ignore all events
which I perceive of as having a probability of less than .10"? The
producer ignores the .01 probability of being detected in violation be-
cause it is below the threshold set by her SOP governing reactions to
low probability events (see, e.g., Kunreuther and Slovic, 1978). If
this is the case, the enforcement agency will observe no response to
a marginal change in this probability. It will find that its expendi-
ture to raise the probability of detection from .01 to .02 is ineffec-
tive and wasteful of resources.

Therefore, the agency must also concern itself with the SOP's of
the regulated industry when it designs and implements enforcement pro-
grams. It cannot know how to deter Violative behavior in the absence of
a model which determines the variables affecting compliance decisions,
and some of these variables may be SOP's. Although this thesis is pri-
marily concerned with the decision processes of the agency and not
those of producers, the existence of SOP's among the regulated group is
pointed out as a potential explanation of the successes or failures of
enforcement programs which we may observe in the process of this evalua-

tion.

27

We now turn to a discussion of "rational" enforcement theories,
which will provide general standards by which the performance of FSIS
enforcement programs can be judged. It should be emphasized that the
use of the assessment standards proposed by these theories is normative

and qualified due to limitations which will be discussed.

2.3 "Rational" Enforcement

2.3.1 Theories of Rational Enforcement

As described in the introduction, efficiency criteria are used as
a norm in this approach. That is, the enforcement agency "should" at-
tempt to obtain the maximum amount of compliance among the regulated
population given constraints on the amount of resources it possesses
and the authority it has been given. Becker (1969) provides a detailed
development of optimality conditions for the rational enforcer in
"Crime and Punishment: An Economic Approach." The behavioral rela-
tions which result in the costs of law enforcement are separated into
five categories and mathematically defined in order to determine opti-
mal levels of "p" (probability of conviction) and "f" (punishments)
which should be set by enforcers.

The first of these relationships is between the number of offenses
and their cost to society. Total damage to society is the difference
between the harm to society created by commission of crimes and the so-
cial value of the gain to offenders which results from the crimes. The
former (harm) is said to marginally increase with the number of crimes
while the latter (gain) marginally diminishes. Net "damages" include
not only measurable financial losses but also the psychic damage to vic-

tims of crimes. In the meat residue case, harm includes both the

28

potential health hazards of consumption of such chemicals and any de-
cline in product sales due to a decline in consumer confidence. There-
fore harm accrues not only to consumers, but to other producers, since
meat is often a nondifferentiated product.

The second relationship is between the cost and level of intensity
of enforcement activity, which is presumed to be positive. This im-
plies simply that costs of enforcement will rise with the number of
arrests made, modernization of enforcement technology, and other fac-
tors which represent inputs to the intensification of policing activi-
ties. In the case of meat safety, intensity of the enforcement effort
is primarily a matter of increasing the number of samples taken for a
particular substance. Other factors which increase the likelihood of
detection include expenditures on developing rapid testing procedures
which can be used at slaughter plants. Rapid testing procedures allow
the proportion of animals tested for some residues to increase at low
cost to the agency.

The third issue is the function describing the "supply of offenses,"
or the relationship between the number of the offenses and its explana-
tory variables. These variables affect criminals' benefit and cost
calculations. They include the probability of conviction (p), the
punishment if convicted (f), and a "portmanteau" variable for other
factors such as income available in alternative pursuits and willingness

to break the law.1

 

1In the case of chemical residues in meat, as brought out in Chap-
ter One, the probability of conviction is the cumulative probability of
at least three separate enforcement events: being chosen fbr sampling
(inspection), being discovered in violation by the laboratory analysis
(detection), and actually having fines or other punishment imposed
(prosecution).

29

The relative responsiveness of the regulated group to p and f will
be affected by attitudes toward risk. Risk preferring individuals will
react more (i.e., reduce or increase criminal activity) to a change in
probability than they will react to an equal percentage change in the
severity of the punishment. Expected utility of a criminal act for a
risk averse individual, on the other hand, is changed less by the same
change in probability. Risk neutrality implies that those contemplating
a heinous act are equally deterred by equal changes in p and f.

Since these changes in p and f represent alternative allocations
of resources, we need to know about prospective criminals' risk pre-
ferences in order to utilize resources efficiently. Empirical evidence
cited by Becker indicates that p and f do significantly affect the num-
ber of offenses, with the effect of p exceeding that of f. This would
lead us to conclude that the criminals Becker is studying--committers of
felonies--are risk preferrers in the relevant ranges of p and f.1

The nature and costs of punishments fbr the offender are described
as having a functional relationship with "social costs" of punishment.
The latter vary with type (e.g., fine, imprisonment, torture, etc.) and
duration of punishments imposed and include prisoners' upkeep, the

state's legal costs, equipment and maintenance for prisons, etc. Costs

 

1The range issue for p may be especially important when distinguish-
ing between felonious and economic crime. Since farmers as a group are
commonly perceived of as being risk averse, they may be deterred more
by a change in the punishment than the probability of detection, es-
pecially at the very low levels of probability which prevail in the de-
tection of residues in meat products. In addition, Kunreuther and Slovic
(1978) have demonstrated the tendency of some groups to simply ignore low
probability events. Factors in the creation of this low p are discussed
in Chapter Three.

30

to the offender are expressed in terms of foregone earnings (while
spending time in prison), foregone consumption, and/or any fines paid.

Optimality conditions must take into account all of the behavioral
relationships described above. If enforcement and punishment were cost-
less, the maximum deterrence could be obtained by merely raising p close
to one, and increasing f until expected costs outweigh expected gains
from crime. When enforcement is costly, as in most conceivable cases,
the social loss which occurs when resources are spent on enfbrcement
programs must be taken into account. Therefore, in an overall sense
the rational enforcer should attempt to minimize the social loss func-
tion from offenses, L = 0(0) + C(p,O) + bpf (O), with respect to p and
f. In this function, 0 represents net damages from crime, 0 the level
of offenses, C the social cost of enforcement, and b a coefficient
transforming loss to the offender into social 1055 (b will be greater
than one for long prison terms, where loss to society due to the costs
of maintenance is great).

It should be noted with reference to the meat residue case that
although the government as a whole may be responsible for the minimiza-
tion of this entire function, some aspects can be regarded as fixed by
the individual agency (FSIS) which is in charge of determining the p and
f of crimes. The decision about the size of FSIS's budget allocation
for enforcement purposes is made each year by the Congress; thus a
limit is put on C(p,0) which can result from FSIS expenditures by a
force outside the control of the enforcement agency. Another set of
decisions outside the FSIS's influence is what comprises "social harm."
In setting tolerances for the presence of chemical substances in meat,

FDA and EPA determine the amount of social harm which is acceptable.

31

FSIS's function with regard to these chemical residues is limited to
enforcing these pre-set definitions of "socially harmful" actions.
Furthermore, the "f" which can be imposed by any government agency is
defined to an extent by law. Since FSIS cannot impose civil penalties
for offenses, other methods must be used to impose costs and punishments.
It is also not within FSIS legal mandate to pursue criminal cases against
Violative producers. The response of FSIS to this constraint will be
examined in more detail in later chapters.

Becker's conclusion on the most appropriate type of punishment
is that fines should be used where possible, in amounts adequate to
cover the damages caused by the crime and the costs of apprehension.
This is because the use of other punishments such as imprisonment add
to the social losses due to crime (demonstrated in the social loss
function by the coefficient b). He does, however, allow fbr the use
of imprisonment in the case of crimes which create greater psychic than
financial damage (i.e., where monetary losses are an inappropriate
measure of total social loss due to crime).

Stigler (1970) in an essay on "The Optimum Enforcement of Laws"
expands on Becker's work and provides additional normative standards
for agency behavior and shifts the focus to "economic" or "regulatory"
crime. Important additions to Becker's original development include
the consideration of costs to private individuals or firms of compli-
ance (e.g., purchase of special equipment) in the aggregation of the
social costs of enforcement. Further, the concept of "social value of
gains to offenders" is called into question. Since society has legis-

lated against certain activities, it presumably considers gains from

32

such activities illegitimate. Deducting these from the harm created
by the offenses in order to derive total social damages therefore im-
plies a contradiction.

Stigler presents two criteria for a rational enforcement structure.1
The first is that expected penalties should rise with expected gains
from crime so that d(pSQ)/dQ §_d(l-p) SF/dQ for all 0 [where S = number
of crimes, 0 = size of crimes, F = the punishment, and p = fraction of
crimes completed successfully (p = p(E,Q) where E = expenditures on
enforcement)]. Marginal deterrence should exist, since if expected
punishments are equal for all crimes, only the largest (in dollar value)
will be committed. That is, one might as well be hung for the theft
of $1000 as for $100.

Secondly, expenditures on enforcement of regulations should yield
a decrease of offenses at the margin equal in value to the return those
resources would earn in competing uses. The social value of the "harm"
prevented should be at least as great as that of the expenditures made
in preventing it. That is, if $10 million are allocated to inspection
of meat for residue violations, at least as much "social harm" should be
prevented as if the $10 million had been allocated in another way--for

example, to inspection of meat for disease organisms.

 

1Note that the decision variables of the FSIS determine only a por-
tion of the entire enforcement structure. For example, while it can
set the probabilities of inspection and detection for residues in meat,
only FDA can follow up and impose criminal punishment upon producers
who create Violative residues. The efficiency of its resource use is
not exactly the same issue as the overall rationality of the entire
structure of enforcement facing livestock producers.

33

Stigler claims that there are several deficiencies in enforcement
agency methods with specific reference to the type of administrative/
economic crime which is the topic of this thesis. Agencies, according
to Stigler, fail to take account of the costs imposed on the regulated
group by the requirements of the law as a part of the total "social
loss" to be minimized. This has unintended and harmful redistributive
effects. In the meat safety case, the implication is that the FSIS
fails to consider the costs to the meat producer of obeying regulation
and being caught in noncompliance, which are, after all, social costs
of enforcement. (Whether Stigler's generalized complaint can be applied
to the particular case of FSIS is taken up in Chapter Four).

A second problem Stigler raises is that inappropriate measures of
agency performance are used which motivate the agency to act nonoptimal-
ly. Rather than viewing the number of prosecutions or cases undertaken
as an indication of diligence on the agency's part, the value to society
of these cases should be assessed and maximized within resource con-
straints according to specific criteria for each type of offense. The
agency responsible for enforcement of the residue tolerances for meat
should then allocate more resources to preventing the residues which
have the potential of causing the most "social harm" per incidence, ac-
cording to some definition of what comprises "social harm," assuming
that the cost of doing so would not also be higher. If it is defined
as number of human deaths, the most toxic chemicals should receive spe-
cial attention. If it is defined as the prevalence of cancer risk,
then the most carcinogenic of substances should be the target of spe-

cial surveillance. Prevention of one incident involving these substances

34

would be of as much real value as the prevention of several incidents of
a less dangerous kind, for example high antibiotic levels.

A final interesting observation made by Stigler is that apprOpria-
tion procedures provide a method of legislative control over the level
of enforcement which allows marginal adjustments to be made within an
existing structural (legal) framework. The potential for flexibility
through control of resources provides a desirable alternative to the
more difficult and complex process of changing agency mandates via
legislation.

The preceding discussion has focused on normative propositions
about the "optimal" enforcement level from the agency perspective.

Some of the primary conclusions are that costs and benefits of enforce-
ment should be weighed when decisions are made about implementation of
enforcement programs. The calculus of these costs should include con-

sideration of the costs of meeting standards, the relationship between
intensity of the enforcement effort, the probability of detection/reso-
lution of a crime and enforcement costs, and the relationship between

criminals' risk preferences and effectiveness of different enforcement

strategies (changing "p" vs. "f").

2.3.2 Risk Attitude and Compliance

Policy implications of optimal enforcement theory depend on whether
potential offenders are risk neutral, averse, or preferring. If risk
neutral, the market offense supply function 0 = O(p,f,u) (where u is
the "pormanteau" variable mentioned earlier) depends on p and f equally.
Thus p could be lowered and f raised in equal proportions and total so-

cial loss reduced, since apprehension and conviction costs would also be

35

lowered by lowering p. When violators are risk neutral, the policy re-
commendation is lowering p to approach zero while raising f to a high
level, at which pf would produce the optimal O.

Breit and Elzinga (1973) also examine the implications of assump-
tions about risk preference and offer some observations about present
day managers' attitudes. They agree that the degree of risk preference/
avoidance among firm decision makers determines whether firms will com-
ply under given probabilities and costs of enforcement actions against
the firm. This occurs because the level of utility derived by the com-
pliance decision maker is a function not only of costs and benefits of
compliance, but also of the perceived amount of risk which accompanies
these choices. Therefore, a given enforcement policy entailing certain
levels of "f" and "p" for all group members will produce a different
level of "demand" for offenses between risk averse and preferring indi-
viduals. Risk averse managers will choose relatively less noncompliant
behavior when f is high and p low since they derive less utility from
such activity than risk preferrers under the same conditions.

In order to create effective deterrent, therefore, the enforcer
must determine the risk attitude of the regulated group and set p and f
accordingly. Breit and Elzinga, however, are concerned with antitrust
penalties—-whitecollar crime-~whi1e Becker refers primarily to murder,
rape, and other felonious activities. The former therefbre characterize
their "criminal" as a corporate manager conducting business as usual
(i.e., making decisions based on cost and benefit calculations).
Becker's criminal is quite a different character. The character of
compliance decision makers, and consequently the nature of their risk

attitudes, differs between "crime markets," Stigler (1970) says.

36

Differences in risk attitudes between groups of regulated individuals
can result from differing sets of pressures which exist within particu-
lar fields. Enforcement policy must adapt itself to these differences.

The recommendation of Breit and Elzinga with regard to antitrust
policy is that it should be designed with the high penalties or fines
and low probabilities of detection which will deter today's risk averse
manager. Managers of firms now seek secure, more moderate incomes and
avoid chancy situations calling for entreprenuerial risk taking, the
authors assert. Large corporations generate pressure to conform while
ownership and control are separated. Cyert and March (1963) are cited
as supporting this characterization of the manager/criminal since they
claim that avoidance or reduction of risk is a primary goal of the firm.
Although the article refers specifically to antitrust regulations, pre-
sumably the same conclusions would hold for any risk-averse regulated
group.

What are the implications of these conclusions for FSIS enforcers
of residue tolerances? The enforcer's task is to determine which levels
of p and f will best deter the group under regulation, given the risk
preferences of that group, when "optimal" enforcement is the goal. In
the situation faced by enforcers of residue tolerances in meat, the
group of concern is the growers and marketers of food animals whose
practices create the residues. If livestock producers are risk averse,
the optimal level of p will be low and f, high, when the costs of im-
plementing enforcement program are taken into account. Furthermore.
Becker has concluded that the risk averse individual will be relatively
less responsive to marginal changes in the probability of detection

than to changes in the gravity of the consequences of noncompliance.

37

It must be emphasized, however, that knowledge about risk preference
of the regulated group on the part of the enforcer is required in order
to set and implement these optimal levels of probability and cost of

detection.

2.3.3 Limitations to "Rational" Enforcement

As previously mentioned, this theory fails to take into account
some structural aspects of the bureaucracies charged with enforcement
implementation. The actual range of options open to any individual
agency enforcing particular laws or sections of laws is often not as
wise as presumed in the theory which speaks of a rational "enforcement
structure." For example, setting and imposing of penalties and deter-
mining probabilities of detection are often separated functions. The
former may be performed by the courts or otherwise legally mandated,
and the latter by an enforcing agency. Such is the case for residue
tolerances in meat. Although an analysis of the complete structure
could treat all determinants of expected costs of violation as decision
variables, an analysis of the effectiveness of FSIS Residue programs
must consider some of the variables which affect the optimality of re-
source allocation to be exogeneously determined. What the limits of
FSIS power to affect the p, f, and other variables is will become clear
in the description of the program (Chapter Three).

A second factor affecting the adoption of any "optimal" policy
suggested by economists is the incentive structure facing agency deci-
sion makers. Stigler complains that a large number rather than a high
value of cases are undertaken by enfbrcing agencies in order to improve

their prestige and budget allocations. His answer to this incentive

38

problem is to maintain that agencies "should" consider the economic
value of crimes deterred rather than just their number. Regardless of
what agencies "should" do, what they 3111 do is determined by the in-
centives they face. In suggesting changes in the criteria by which
agencies choose enforcement strategies, these incentives must be con-
sidered.

For example, if budget allocations are made on the basis of the
number of violations detected, bureaucrats may resist a recommendation
to implement low probabilities of detection. By reducing the absolute
number of violations detected with the lower probability, they may re-
duce the urgency of this issue and their own effectiveness perceived by
congressional funding allocators. It is therefore unlikely that such a
policy--optimal as it may be in terms of overall allocation of resources
--will be implemented without a facilitating change in incentives.

2.3.4 Other Factors in the Compliance Decision Affecting

Optimal Enforcement

Among factors which may be outside of the enforcement agency's
control, yet affect the observed level of noncompliance, are two which
will be examined here. The level at which the enforced standard is set
will determine how difficult or costly it is to reach. This in turn
affects the compliance decision, yet is beyond the enforcer's control
when standard setting and enforcement functions are performed by separate
agencies.

The second environmental factor is the extent to which obedience
to ethical codes provides cost-free enforcement. In the absence of
ethics promoting law obedience and social responsibility, agency moni-

toring of behavior would have to be increased.

39

2.3.4.1 Standard Setting

Regulation can be broken down into "standard setting" and "enforce-
ment" (which is comprised of both detection and penalty setting) pro-
cedures. Optimal standard setting can only be accomplished under some
knowledge or assumption about the extent to which enforcement efforts
will be effective. Viscusi and Zeckhauser (1979) describe the problem
as one in which the agency strives to maximize a goal (such as air
quality, or food safety) in the face of differing costs of approaching
the goal facing members of the regulated population. Compliance deci-
sions of a firm are based on the cost to that firm of complying vs. the
expected cost of noncompliance. The agency, on the other hand, wants
to maximize the total amount of a certain quality (for example, food
safety) present in the environment.

If the standard setting agency wishes to achieve a certain level
"x" of the quality which it regulates, yet it knows that enforcement of
that level will be less than complete, it must set the standard above
x in order to achieve that level. This is because those firms which
fail to comply in creating x will reduce the total amount of the quality
present in the environment. However, because costs of compliance to
each firm determine whether compliance will occur at all, setting stan-
dards at a very high level (which is more costly for firms to achieve)
nay not provide the maximum level of the desired quality in the environ-
ment created by the actions of all firms. If the standard is set a a
level which makes it very costly for most firms to comply, compliance,
and the level of that quality which it is the agency's objective to
achieve, will be sub-optimal. Viscusi and Zeckhauser therefore recom-

mend that standard not be set at excessively high levels; further,

40

standards should be flexible and vary between types of firms where costs
of compliance vary, for greater efficacy.

This would indicate that in cases where FDA sets tolerances at
very low levels-~through the use, for example, of large "safety margins"
to establish acceptable levels of drug residues in food animal tissues--
it can expect to observe less compliance. This practice may lower the
return on use of a drug, having the same effect as raising its cost.
Animal producers, finding the cost of compliance excessive, will ignore
the regulation unless enforcement is very strict (the probability of
detection is close to one).

For the enforcer, the implication is that costs of compliance will
affect the level of compliance, an important consideration when resources
are being allocated. Where compliance is cheap, the enforcers need not
expend resources to raise the probability, and hence expected costs of
detection, as high as in cases where standards have been set at levels

which make compliance expensive.

2.3.4.2 Ethics

Underpinning all social behavior are ethical or moral constraints
which encourage or discourage, legitimize or condemn certain categories
of actions. In developing a theory about what effective levels and
methods of enforcement are, all of the factors which motivate the regu-
lated group are of interest. Belief in the legitimacy of laws control-
ling behavior, or what may be described as the level of "law-abidingness"
present in the social environment, is an example of an ethical code

which may affect enforcement costs.

41

The potential benefits from such informal codes can be said to con-
stitute "public goods," from which society derives a benefit of costless
enforcement. McKean (1979) described the public good characteristics of
ethics, notably underproduction due to the free rider problem. Members
of society are "free riders" of moral codes to the extent that all de-
rive equal benefits from the obedience of society as a whole to these
rules regardless of one's own level of obedience. Individuals cannot
capture all of the benefits of obeying informal ethical codes and there-
fore have little economic motivation to incur the costs of doing so.
Therefore qualities such as law-obedience in the absence of enforcement
will tend to be underproduced.

Often government's role is partially perceived of as the provision
of such underproduced public goods. By definition, however, government
cannot legislate law obedience. Provision of this good requires the use
of alternative institutions--religious, educational, or propagandistic
in nature. The level of willingness of firms and individuals to comply
with ethics is dependent upon the general level of societal compulsion
to obey these rules, the visibility of infractions, the willingness of
others to recognize, point out, and condemn the noncompliant behavior,
the costs of doing so, and the payoff for social approbation (a function
of the individual's attitude toward and position in society.)

The amount of costless benefits which accrues to society is also
linked to the perception of the lawmaking process, and the laws themselves,
as legitimate by those under regulation. Those laws perceived of as
"fair" have a better chance of being obeyed with fewer (costly) enforce-
ment activities on the agency's part. Furthermore, disobedience builds

upon itself, or increases at an increasing rate. The fewer people who

42

obey a particular ethic, the fewer there are to perform the "sidewalk
surveillance" which supports the ethic, and the more difficult it is for
those who do to get results such as social disapproval of or action

against the violator.

2.3.5 The Information Problem and "Optimal" Enforcement

An evaluation of the effectiveness of agency programs which will be
useful to the agency in guiding its future decisions about allocation of
resources must take into account conditions which it faces in its day
to day operations. Members of the regulated group are not perfectly in-
formed about the true expected costs of being detected in violation, and
the enforcement agency does not know with certainty what types of changes
in compliance behavior it will generate by changing either probabilities
or costs of detection. Such information is difficult to obtain and there-
fore represents a cost of enforcement which has not been considered by
theories which stress the normative goals implicit in optimal enforcement
policies.

We now turn to the problem of obtaining the infbrmation necessary
to make economically efficient allocation decisions, and the implications
of the costs of information on the achievement of maximum levels of goals.
This comprises an inquiry into what effective deterrent strategies will
be in the face of the actual, imperfect information conditions faced by

enforcers.

2.3.5.1 The Effects of Uncertainty on Agency Decisions
The lack of information required to make decisions about the levels
of probability and cost of detection which should be imposed upon the

regulated group in order to create the maximum deterrent for the

43

enforcement dollar affects the agency in several ways. Becker and Stigler
suggest that in order to change or set these probabilities and costs in
such a way as to deter, some knowledge about the responsiveness of the
regulated group to these two elements of the expected costs of noncompli-
ance is required. The agency is faced with a resource allocation choice:
it may devote time, money and available expertise to creating a high,
intermediate, or low probability. It may alternatively devote resources
to other programs not intended to affect the probability of detecting
violators, for example on efforts to educate potential violators about
the consequences of their actions. If the regulatory task of setting

the level of punishments is within the agency's jurisdiction, it may de-
cide to raise the costs associated with detection and conviction in vio-
lation.

When the precise nature of the reaction of violators to changes in
Becker's p and f is unknown, the agency or agencies responsible for de-
termining and implementing the levels of these variables will have diffi-
culty in computing the effect of such changes. Furthermore, as previously
mentioned many elements detennining the costs of violation are outside of
the power of the agency to control. In the meat residue case, for
example, it has been pointed out that FSIS controls only certain aspects
of the probability of detection as discussed by Becker et a1. It may
set the probability of an individual carcass being sampled and tested
for residues, but the prosecution of violators is the job of the FDA.

The level of punishment or costs imposed may be affected in some ways by
FSIS, but is primarily set by the courts following prosecution.
Therefore we can say that any individual agency is limited by both

informational and structural/jurisdictional considerations in its ability

44

to implement what Becker and Stigler define as an “optimal" enforcement
program. Even if the primary goal of decision makers within the agency
is economic efficiency defined as the maximum deterrence level per en-
forcement dollar, we would not expect to observe this condition in an
examination of the enforcement structure created by any individual agency,
When information is costly and extrajurisdictional factors intervene.

2.3.5.2 The Effects of Uncertainty/Information

Costs on Compliance Decisions

When describing decision processes and determining the relative
strength of decision variables, the presence of uncertainty in the en-
vironment has important implications for the compliance decision and
hence for the effectiveness of enforcement strategies. Becker et a1.
resolve the problem of the producer's uncertainty over whether the en-
forcing agency will or will not inspect, detect, and prosecute by assum-
ing that she is cognizant of the actual probabilities of these events.
According to the model, she then acts according to these known probabili-
ties in a fashion which maximizes her level of utility, given her per-
sonal attitude toward risk.

The producer making a decision about use of chemical substances
cannot know the probabilities that her herd will be sampled for chemical
analysis, that the analysis will correctly indicate a Violative level of
the substance, that if sampled and detected in violation, she will be
prosecuted, or what the results (fines, imprisonment, acquittal) of such
a prosecution might be. All of these depend upon prevailing agency
policies and intensity of enforcement efforts. The producer, when con-
sidering the possibility of adverse enforcement consequences, operates

under "perceived" probabilities rather than actual known probabilities.

45

What are the determinants of a perceived probability of inspection,
detection, or prosecution? Bartlett (1973) indicates that these factors
may differ significantly from the determinants of the actual probabilities,
since the existence of uncertainty highlights a crucial role for informa-
tion. Uncertainty is, essentially, a lack of information about the out-
comes of one's actions.

The acquisition, processing, and use of such information is a costly
process. This creates the potential for the development of influence by
the enforcers over compliance decisions through provision of subsidized
information of a type which will induce individuals to act as desired by
the agency. All information received must be sorted into various cate-
gories according to urgency, relevance, and reliability before it can be
used. Information from sources considered unreliable or "interested" may
be discounted to some extent. However, as long as the discount rate on
this information remains less than 100%, the agency providing it may still
derive some benefit through the influence upon compliance decisions which
is thereby created.

Optimizing theories neglect the economic ramifications of the psycho-
logical deterrent effect created by perceptions of agencies' ability and
willingness to act, in concentrating only on actual probabilities and
costs of such actions from the agency's and society's point of view.

For example, the uncertainty of the producer about the actual prob-
abilities gives the agency a tool for control and influence over producer
behavior which we may refer to as "threat." Threat is the opportunity
that the agency has to magnify the perceived probabilities--without in-
creasing actual probabilities--through the provision of subsidized infor-

mation (via public relations and the media). The effectiveness of this

46

tool will be limited by the agency's credibility among producers--i.e.,

by the extent to which the subsidized information is discounted. The
size of the discount applied is related to the actual probabilities of
inspection, detection, and prosecution since the past history of the
agency's perceived willingness to follow through on its threats will af-
fect producers' standard operating procedures for handling agency-provided
information. The difference between the probabilities created by agency
action (actual) and those acted upon by producers (perceived), although
presumed to be nonexistent in the Becker model, in fact create a poten-
tially effective and low cost enforcement tool.

The agency should realize that there may be a difference between the
level of probability of detection which it creates through its programs
and the perception of that level of probability within the regulated
group which is determining compliance decisions. Even if it knew the
actual levels which should be implemented, a gap or delay between these
and the perceived levels could mean a lower deterrence effect than planned.

2.3.5.2 Implications of the Information
Problem for Program Evaluation

The preceding section offers some potential explanations as to why
an enfbrcement program, however well conceived and implemented, may fall
short of the maximum level of deterrence per enforcement dollar spent.
Gathering and processing infbrmation represent a cost to the agency of
creating deterrence; one which, if it could be measured accurately, might
explain the use of standard operating procedures even by goal maximizers.

Although the maximum level of deterrence may be as difficult to
measure as it is to achieve, this does not invalidate the contribution

of maximizing theories to program evaluation. Lack of perfect information

47

on risk preferences, responsiveness to changes in probabilities, combined
with limitations in authority and jurisdiction, mitigate the ability of
the enforcement agency to choose and implement Becker's "optimal p's and
f's." We may still employ the considerations raised by such theories to
assess the economic efficiency of an agency's programs, however. Exigen-
cies of the system and environment within which it works must be taken
into account, but the results of the Becker and Stigler models still
point out important general directions which a program should take when
efficiency of resource use is the measurement criterion of success.

For example, although the value of social harm created by a potential
violation may be difficult to measure, we are still interested in knowing
whether the agency uses some internally consistent standard of potential
damage in prioritizing the problems it addresses. Or, as Stigler claims,
is FSIS primarily motivated to justify its programs purely through the
detection of large numbers of violations with no consideration of their
value? Does marginal deterrence exist--that is, are the expected costs
of noncompliance, and hence the agency's expenditures to create these ex-
pected costs, in proportion to the potential damage of different types
of violations? Does the agency pay attention to the responsiveness of
the regulated group to its various programs, assess this responsiveness
in any systematic way, and adjust the programs accordingly? Are the
standard operating procedures for decisions within the agency designed
with these goals in mind? And finally, to what extent do agency prac-
tices, as simplified into standard operating procedures, achieve the
goal of creating the highest level of deterrence possible for the amount

of resources used? All of these equations are motivated by the choice

48

of assessment criterion made in the normative, "optimal" or economic

efficiency approach to evaluation discussed in this chapter.

2.3.6 Implications of a Normative Evaluation Criterion

The choice of a standard by which success of a program will be
judged sets the tone of any investigation. It implies that the evalua-
tor has knowledge of what the objective function of the agency being
examined should be. It also points out what the results of a program
with the specified objectives should look like. In so doing, the choice
of a measurement criterion narrows the focus of the evaluator, and limits
the agency outputs which are of interest.

What would be the result of choosing as an agency's proper objective
a very different goal than agency decision makers have? Becker's theory
does not maintain that it describes accurately the goals of agencies.
Instead it starts from the premise that maximizing output per unit of
input should be the primary goal of bureaucrats spending scarce public
resources. Yet these bureaucrats operate within social and political
systems as well as economic--who is to say that their only objective,
or even the most important objectives are economic or "resource saving"?

Agency behavior which is motivated by forces other than the drive
to economize may appear quite different than what the economist expects
to observe. The problem is that in narrowing her perspective to the
economic as Becker defines it, the analyst may fail to look for the other
incentives or signals which might explain agency behavior. When the
evaluator merely makes assumptions about the objective function of the
agency and judges agency behavior as if it attempts to achieve these ob-

jectives, failure may be observed. A potential explanation for the

49

agency's "failure" is that it is not trying to do what the analyst has
decided it should be doing. The analyst reports failure but cannot ex-
plain why because the structure of incentives facing the individuals
within the agency and the goals of the agency have not been researched.
Specifically, this issue faces the economist analyzing a political
institution. The political and organizational framework which the agency
must function within offers rewards and punishments for actions of the
agency. Due to this incentive structure, some agencies are viewed as
serving particular clientele to the exclusion of others, or being the
captive of special interests. USDA has been accused of responding to
the needs of agricultural producers at the expense of consumers, for

example (see Sowing the Wind). This becomes a problem when USDA is asked

 

to regulate producers in the interest of consumers, especially when the
alleged danger facing consumers is as controversial as chemical residues.

One way to avoid the problems inherent in any choice of a normative
assessment criterion is to analyze the entire structure within which an
agency works to determine which incentives affect it the most. Intra-
agency dynamics would also have to be examined in order to explain the
relationship of the particular division (in this case, the Residue Evalua-
tion division within the FSIS) to the agency itself, and the outside
world.

Such an approach, useful as it might be in explaining agency enforce-
ment behavior, is beyond the scope of this thesis. Therefore we shall
take the approach of studying the results of this incentive system, rather
than the entire system itself. These results are the actual objectives
of the Residue Division. Rather than simply assuming that the objective

function of our enforcement agency includes the elements that Becker and

50

Stigler would call "rational," Chapter Three will examine what these
objectives actually are, as revealed by SOP's and interviews with offi-
cials. Then a comparison can be made between the actual objectives and

those proposed in the theory.

2.4 Conclusions

The objective of this chapter has been to review several theories
which contribute to an explanation of the decision processes within
agencies which result in the enforcement structure facing regulated pro-
ducers. The first section briefly discusses the existence of standard
operating procedures as a tool used by organizations such as enforcement
agencies to distribute information processing and decision making respon-
sibilities within the organization. It indicates that in order to under-
stand what an enforcement agency does--how it makes the decisions which
create or contribute to the enforcement structure which faces the regu-
lated group-~we must look fbr these standard operating procedures. The
implications of these SOP's must be examined in order to determine how
and at what levels decisions are made which distribute costs of maintain-
ing a quality such as "safety of the meat supply" among different pro-
ducers, processors, the agency itself, consumers, and taxpayers.

The second section described theories of what an "optimal" enforce-
ment policy should look like. These theories maintain that the criterion
of economic efficiency--creating the highest level of deterrence possible
with each dollar spent on enforcement programs--is the normative standard
by which success or failure of a program should be judged. Qualifications
to this type of theory which have been pointed out include the difficulty

and cost of obtaining the information required for truly maximizing

51

decisions, and the legislated structure which limits the ability of an
individual agency to completely determine all elements of the expected
costs of noncompliance. The conclusion is that although infbrmation
problems make an accurate assessment of whether an agency is performing
at peak efficiency very difficult, these theories nonetheless offer use-
ful general directions for the program evaluator when the goal is econo-
mic efficiency.

We therefore turn to a more specific application of the theory dis-
cussed above to the meat safety case. The purpose is to draw out of the
theory conclusions about what characteristics of an enforcement program
should be, when agency decision makers attempt to economize in the face

of uncertainty, to which FSIS programs can be compared.

2.4.1 Factors in Violative Behavior

The following conclusions about Violative behavior in the meat safety
case can be drawn from the economic theories presented in this chapter:

a. A high level of violation can be expected, other things being
equal, when the perceived benefits of violation are high. This may occur
when meat producers believe that subtherapeutic application of a particu-
lar drug is extremely effective against certain costly health problems,
and no adequate substitutes are available.

b. Where expected benefits of noncompliance are low, compliance may
occur regardless of very low probabilities of detection in violation.
This may be the case when equally effective, low cost substitutes are
available. (Substitutes may be other substances or a change in manufac-
turing practices which can produce the same effect on animal health as

the residue-creating substance.)

52

c. Where uncertainty exists about the technical relationship between
use of a substance and its effects on animal health and growth, the pro-
vision of information will play a vital role in the extent and way in
which the substance is used. Major sources of infbrmation on animal drug
use are labelling instructions and advertising. It can therefore be pre—
dicted that clarity and precision of labelled instructions--required and
controlled by law, and provided by drug producers--will be an important
factor in use decisions of producers, and therefore in the level of vio-
lations. Provision of subsidized information required by the government
can influence the producer to comply with manufacturing practices which
prevent residues, when information from other sources is scarce and costly,
or heavily discounted.

d. Where probabilities of detection are low in combination with
low fines and punishment, optimal theories would predict a very high rate
of violation. Intervening factors might include a perceived probability
of detection larger than the actual, fear of market consequences above
and beyond the enforcement consequences, or strong identification with
the interests of the producer group as a whole (as opposed to purely per-
sonal interests), and a high degree of ethical compulsion to obey laws

regarding chemical use.

2.4.2 "Efficient" Enforcement

In the face of the above listed predictable responses to enforcement
programs on the part of the regulated group, the following are some of
the reactions we would expect on the agency's part where efficiency of

resource use is at least a general guideline.

53

a. The agency should allocate resources to preventing the crimes
which cause the greatest social harm, when the goal is economic efficien-
cy. Therefore FSIS should be devoting more attention to problems which
it perceives of as having the greatest potential for creating social harm.
We might observe more surveillance or educational programs directed toward
particularly toxic chemicals in widespread use. (The difficulty is in
defining and quantifying "social harm." The agency may feel that the
most toxic of chemicals cause the most harm to society, since human
deaths may result from exposure. Alternatively, they may be concerned
most about carcinogenic residues, since these have the potential of af-
fecting a large number of consumers with a particularly ”costly" disease
over a lifetime of exposure. 0r evidence of use of a substance not ap-
proved for food animal production may indicate a high level of social
harm to the agency. Deciding among these candidates for greatest danger
to society is outside the scope of this study. In explaining agency
priorities, we are primarily concerned with the agency's SOP's developed
in response to its perception of the danger.)

b. The agency should allocate its resources to problems where it
knows it can be effective, when it has a cost effectiveness goal. That
is, it must have the capability of deterring improper substance use.

One factor affecting this is the existence of substitutes for the sub-
stance. Another is the perception of the industry that continued viola-
tions threaten consumer confidence, along with industry willingness to
cooperate with FSIS. A third is the availability of testing facilities
which are reliable and low in cost. A fourth is the ability of the agency

to trace and identify violators in order to impose costs. All of these

54

factors would contribute to the ability of FSIS to deter violations with

a relatively low expenditure of resources.

2.4.3 Effective Enforcement

When information is costly and unevenly distributed, the agency has
the opportunity to affect producer decisions through provision of subsi-
dized information and by using threats to magnify the perceived probabil-
ity of detection. Effective enforcement must take into account the role
of information in compliance decisions in these two ways.

If violations are observed at very low probabilities of detection,
the theories examined here offer two explanations. Either the regulated
producers are risk preferring, or the low probability has placed the
problem of detection in violation outside of the producer's bounds of
concern. Effective enforcement policy would make good use of data on
past correlation of variations in these probabilities with violation
rate changes. This data should indicate which levels or changes of ex-
pected costs have proven effective in the past, provide evidence on pre-
vailing risk attitudes, and guide the agency in future enforcement design.

Due to the potential for strong intra-group relations to affect in-
centives of individual members, industry structure is a factor which must
be taken into account by enforcers. A well organized, integrated indus-
try may take to self policing to emphasize to its members that when the
product is undifferentiated, the violation of one may affect consumer
confidence which is ultimately dangerous to the industry as a whole. A
highly fragmented group of producers, on the other hand, requires more
frequent monitoring by the agency itself since group interests may not be

most effective if it takes into account these variations within the

55

regulated group--in Stigler's words, subdivides the "market" for offenses--

when it allocates resources to different enforcement programs.

CHAPTER THREE
RESULTS, PART I: THE ENFORCEMENT STRUCTURE

3.1 Introduction and Methods

This chapter describes enforcement of chemical residue tolerances
in meat. The description centers on the two agencies primarily involved,
FSIS and FDA. Foremost is a consideration of the way in which resources
are allocated by these agencies, in implementation of the legal mandates
they have been charged with by Congress.

Included in the description is a discussion of where and how deci-
sions are made about changing the sampling rate, and hence the probabil-
ity of detection of a violation of the tolerances. This decision pro-
cess is broken down into several levels of responsibility within the
FSIS: at the slaughter plant, at the regional office, and at the national
level. The decision hierarchy funnels information about the more serious
problems to the higher echelons, while screening out smaller problems--
via certain standard operating procedures and pre-set decision criteria--
for solution at the lower levels.

Theory has emphasized that we must understand the decision processes
of the agency in order to discover how it allocates its resources (both
time and financial). Once we have achieved this objective, we can pro-
ceed to answer the question: are these resources allocated properly?
That is, does the enforcement structure created by FSIS and FDA decisions
effectively deter Violative behavior among producers? The latter ques-
tion will be the subject of Chapter Four.

56

57

A further objective is to describethe channels which transmit in-
formation between the various levels of FSIS and from FSIS to FDA. The
result will be a picture of the types of decisions about enforcement
which are made at each of the strata within FSIS, what the criteria are
for making these decisions as described by the decision makers themselves,
and what the implications of these decisions are for producers of live-
stock.

In order to describe the structure and process of decision making
within FSIS for Residue Programs, several techniques were used. Documents
were obtained from the agency which described its programs and goals.
These, along with critiques of agency procedures from government and pri-
vate sources, were reviewed to establish a basic understanding of the
options which have been chosen by the agency in the past.

A more detailed understanding of the decision processes which lead
to implementation of enforcement programs at the various levels of the
FSIS was obtained through open-ended interviewing of agency officials.
Since the document review indicated that different types of enforcement
programs could be initiated at several different stages within the FSIS
hierarchy, individuals working at each of these stages were contacted.
These included an Inspection Supervisor and the Veterinary Medical Offi-
cer (VMO) and in-plant inspectors in his circuit, a regional Staff Assis-
tant in charge of residue programs for the Northeast Region, and national
level FSIS staff concerned in various ways with residue problems. The
national level staff interviewed included the Director of the Residue
Evaluation and Planning Division, members of his staff, members of the
Mathematics and Statistics Division staff, and a Compliance Division

officer.

58

In addition, information was required about the involvement of the
Food and Drug Administration (FDA) in enforcement of tolerances in meat.
Therefore members of the following divisions within that agency were
interviewed: Human Food Safety, Pre-approval Section for Animal Drugs,
Budgeting, and Compliance. Representatives of several producer groups
which have offices in Washington, D.C. comprised the final category of
individuals contacted. These included the American Feed Manufacturers
Association, the National Milk Producers Foundation, The American Meat
Institute, and the Animal Health Institute. All of these interviews took
place during September 1983 while discussions with FSIS officials occurred
at various times between March and November of 1983. Specific dates,
position titles, and names of interviewees are provided in the List of
Interviews following the bibliography.

Topical Outline: The next section of this chapter describes the
general inspection procedures used at slaughtering plants in order to
put the particular type of inspection in which we are interested-—residue
sampling--in perspective. Specific procedures used for collection of
samples are then described, including an explanation of the factors
which would cause an in-plant inspector to take either a monitoring or a
surveillance sample. Of particular interest are the indications which
might lead an inspector to initiate an animal-or producer-specific sur-
veillance program. Following this description, some of the implications
of these inspection procedures for the producer and the slaughterer will
be pointed out.

The following section shifts the focus to the regional level. It
offers a description of the types of responses to information from the

monitoring system which are possible at this level. These include

59

routine follow-up, initiation of the Contamination Response System, and
other surveillance responses. The role of the regional office in imple-
menting special programs initiated at the national level is also included.

A discussion of the program design process, which takes place at the
national level at FSIS offices in Washington, D.C., is provided in Sec—
tion 3.2.3. The method and criteria for setting sample sizes for the
monitoring program, the types of residue problems which are dealt with at
the national level, surveillance responses, and factors in their initia-
tion are the primary topics covered in this section. A further discussion
of the implications of these enforcement decisions and policies follows
the description. Finally, the section includes a brief mention of educa-
tional programs about residues which are run by the agency.

The fourth and final general topic is the responsibilities of the
Food and Drug Administration in following up information on violations
provided by FSIS monitoring. The criteria and procedures for on-farm
investigations are presented.

On the following page is an organizational chart of the FSIS. The
Residue Evaluation Division, which is within the Science Program, and the
five Regional Offices under the Meat and Poultry Inspection program are
the major organizational components examined in this study. Extending
from the Regional Office box should be a number of Area Supervisors--the
number differs for each region--and within each area, a number of in-
plant veterinary and non-veterinary inspectors. Area Supervisors are in
charge of a circuit of slaughter and processing plants which they visit
periodically in order to maintain communication with inspectors, ensure
that inspection is proceeding smoothly, and assist inspectors in the

solution of any problems they may be encountering.

60

usaupsuwsm< co acmEpsmgmo .m.: .mow>som cowpomamcH new preom toad

—.m weaned

.ommp .eorpowamcH xsppsoa use “we: ”musaom

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

61

3.2 FSIS Enforcement Programs

3.2.1 Meat Inspection: Slaughter/Processing Level

The purpose of this section is to describe the inspection process
which takes place in slaughtering plants in order to provide a context in
which to examine inspection for chemical residues specifically. In under-
standing the implementation of residue programs by the Food Safety and
Inspection Service it is important to note the historical and programmatic
environment in which they developed. This environment has conditioned the
way in which residue inspection has been conceived and carried out.

Federal law provides for inspection of all meat and poultry intended
for wholesale or retail sales at the point of slaughter, with certain
exemptions.1 Each animal slaughtered is subject to both ante and post

mortem "organoleptic" inspection.2

State governments may provide their
own inspection systems, if they employ standards at least equal to federal
standards. The federal government will provide 50% of the necessary fund-
ing to support these state systems. If state inspection systems are found
to be inadequate during periodic federal review, they can be designated

for federal takeover.3

 

1Custom slaughter and slaughter for home use only (not for resale)
is exempt. Retail operations, defined as those which sell less than 25%
of total volume to nonhousehold purchasers, which must not exceed in value
$25,000, are also exempt. Small, local poultry operations which raise and
slaughter fewer than 20,000 birds annually for intrastate commerce are
exempt.

2Organoleptic inspection is that which can be accomplished by the
senses of touch, sight, and smell of the inspector.

3Twenty-three states run their own inspection programs for meat and
poultry, and four for poultry alone, according to the Government Account-
ing Office. U.S. Congress, General Accounting Office: USDA's Oversight
of State Meat and Poultry Inspection Programs Could be Strengthened, GAO/
RCED-84-23, Oct. 21 41983. p. 36.

 

62

Ante mortem inspection consists of observing liyg_animals for signs
of illness, overmedication, or abnormalities prior to slaughter. This is
performed by the in-plant inspector, who may designate animals as "U.S.
Suspect" if they display these symptoms. Such animals are held in a
"suspect pen" for final disposition by the circuit veterinarian and, if
condemned, are not permitted to be brought onto the slaughter floor.

The fate of such animals is a source of some concern to inspectors.
supervisors, and circuit veterinarians. They must be removed from the
slaughter plant; to do so involves a cost to the owner of the animal which
may yield no return if the animal dies subsequent to transporting it back

to a farm.1

According to one regional supervisor of meat inspectors, a
condemned, live animal is no longer within the jurisdiction of the FSIS
since its concern is exclusively with animals at slaughter or processing
stages. State authorities are informed, but may or may not take any ac-

tion.2

A local veterinarian informed me that this situation has provided
an economic incentive for unscrupulous individuals who purchase the animals
for salvage and attempt to have them slaughtered without the benefit of
inspection, slipping tainted meat into marketing channels.3
Post mortem inspection occurs on the production line after the ani-
mals have been slaughtered, eviscerated, and skinned. Heads, viscera, and

carcasses are examined for lumps, bruises, tumours, and other abnormalities

 

1Owner may be an animal producer, the slaughterer, or a dealer, de-
pending upon marketing arrangement.

2Interview with Dr. Harold Connaughton, D.V.M., Regional Supervisor
of Meat Inspectors, Southern Michigan area, on March 233 1983.

1 3Interview with Dr. James Lloyd, a Michigan veterinarian, February
984.

63

which may indicate meat borne diseases such as brucellosis, tuberculosis
or cholera. Again, suspect meat is held aside for confirmation by the
circuit or in-plant veterinarian. If condemned, the carcass is removed
to the "inedibles" room where it is dyed with a chemical to prevent its
being used for human food. Such meat may be salvaged for rendering (con-
verting into tallow) or ground into animal feed or fertilizer.

This procedure was designed to solve a particular set of regulatory
enforcement problems which came to public attention during the last de-
cade of thel9th and first of the 20th centuries. Major public health
hazards at that time included unsanitary conditions at slaughter plants
which fostered the spread and transmission of disease, as well as the
presence of rodents and other pests. Gradually, other health concerns
have become predominant; although inspectors are still required to en-
sure sanitation standards and prevent the spread of disease, less visible
conditions have also been discovered which present a hazard to human
health. The slaughterhouse has traditionally been the point at which in-
spection first occurred since it represented not only the point of entry
of animals into human food markets, but also the point at which negligence
was most likely to cause health problems.1

As brought out in the introductory chapter, however, chemical resi-
dues involve a set of enforcement problems which are created at the farm
level, and of which the slaughterer is more likely to be the economic vic-
tim than the perpetrator. Yet FSIS has an inspection system set up at the
slaughter and processing level, and efforts to detect and prevent residues

have been grafted onto this system.

 

1On-farm inspection is the responsibility of the FDA and will be
discussed later.

64

In-plant inspectors participate in residue enforcement in several
ways. Some symptoms which indicate the presence of residues can be ob-
served organoleptically. These include live animals which stagger and
cannot rise or walk straight. Such behavior can be caused by exposure of
animals to toxic pesticides or intentional drug overdosing which is used
to keep them standing long enough to be slaughtered. The symptoms may
be observed during ante mortem inspection while animals are being un-
loaded or moved between pens. They are most often observed in culled
dairy cows which are usually sent to slaughter when they cease producing.
Frequently the failure of a dairy animal to produce is disease related,
and represents an economic loss to the owner. The incentive is great
for the farmer to do whatever possible to salvage at least some value
from the animal; to do so she may overdose it with antibiotics which are
readily available without prescriptions in an effort to keep it standing,
while ignoring required withdrawal periods.1

Post mortem inspection may also yield indications of potential resi-
due problems. Bruise marks on the flesh of a carcass, observable when the
skin has been removed, can indicate a recent injection site. In this
case, some inspectors are equipped with facilities enabling them to per-
form a Swab Test on Premesis (STOP) which will indicate the presence of
antibiotics and sulfas within 24 hours. The sample is swabbed and cul—
tured overnight while the carcass is held aside, and prevented from leav-

ing the plant. If the result of the test is positive, the carcass is held

 

1This information was derived from interviews with Dr. Spaulding,
Residue Division, FSIS, and Dr. Harold Connaughton, Regional Supervisor,
MPIO, for the Michigan area. Furthermore it is supported by FSIS moni-
toring data.

65

while a sample of the tissue is sent to one of the FSIS laboratories for
more sophisticated testing in order to confirm the results. An indica-
tion of residues above tolerance level, when communicated to the plant,
must be followed by condemnation and destruction of the carcass.1

The same procedure can be followed for other indications of residues
apparent during organoleptic inspection. Recently, inspectors in the
Northeastern Region noticed the presence of undigested sulfa boluses
(large pills) in slaughtered veal calves due to the brilliant green color
produced in the viscera by a dye used in the medication. Veal calves are
subject to high morbidity and mortality due to their young age, the stress
they undergo during marketing and transport, and the failure of farmers
to feed them the cow's first milk containing the colestrum. Therefore

the temptation to overmedicate is great.2

Organoleptically detectable
indications of improper sulfa use in veal calves such as "green gut" can
result in a STOP procedure as described above.

The presence of residues in otherwise healthy animals, however, is
not easily detected by the human senses. Therefore, FSIS has developed
a random monitoring program which collects and analyzes samples of animal

tissue and organs. The in-plant inspector collects the samples based on

instructions of the residue division headquartered in Washington, D.C.

 

1The infbrmation on in-plant inspectors' duties with regard to resi-
dues provided throughout this section was gathered from interviews with
Dr. Harold Connaughton, area supervisor for Southern Michigan, and veteri-
narians and inspectors to whom Dr. Connaughton introduced us during an
inspection tour on March 23, 1983, as well as a course outline for in-
struction of inspectors about residue sampling.

2Interview with Mr. John Adams, National Milk Producers' Foundation,
Washington, D.C., September 8_ 1983.

66

Each month, the inspector-in-charge at a slaughter plant receives sampling
information printed out of FSIS computers on FSIS 6000-2 forms, which con-
tain the following information: sample number, species, state and number
of the slaughter establishment, tissues to be collected, date of collec-
tion, address of the FSIS laboratory where the sample is to be sent, and
a code indicating which substance or substances are to be tested for (see
next page for sample form). Various tissues are collected for different
substances or groups of substances because of variations in the paths
traced by these substances through the body of the animal and differences
in testing methods between substances.1 Certain chemicals tend to collect
in organs such as the liver or kidney while others permanently change
organs in such a way as to indicate whether a substance was ever used on
an animal. Others are indicated in chemical analysis by the presence of
"metabolites," which are compounds created in the animal's body during
the process of metabolizing the drug or substance of interest. One
pound of muscle, fat, kidney, or liver is collected according to the in-
structions on the 6000-2 form, except when the substance being tested for
is diethystilbestrol (DES), which requires four pounds of liver tissue
(see page 68 for list of substances with code numbers).

Several aspects of this procedure have important enforcement impli-
cations. First, sampled carcasses are not retained pending results of

the laboratory analysis. They continue down the production line and are

 

1One testing procedure, in some cases, can indicate the presence
of any one of a group of substances. This is true for antibiotics and
chlorinated pesticides. Therefore, the testing of one sample may yield
more than one violation if several substances from the group are present.
Similarly, a negative result can indicate that the sample is free of
several substances.

67

For- GOOD-2

3.; oceanuc-v 00 contusfvlc
' 0... ‘0“ all. Owns." "In“
“II“. on...“

MITORING IBIDU‘
'ROGRAM

  

' 905.9-0h. "IOU.

.. «2002:36-

'Oavt 030-! V. “OCQAVOIQ

      

 

JSSA
1?. O. 3:: 2423
San Francisco. CA 94125

Biological Residue
Attach-ant OR 1

an" an. coo-cu on our-c- clans-one 8. Cocoa

 
 
    

     
  
 

-FSCS

 

  

 
 

   
 

-Scicnce

 

    

'0. MIDIA?$.V .1" ONLY

moan.“ c-Iu- nausea. rm... coca

53:33:.-

YI”UC C...

'flubd cone I'lufll I.“

 

“News ' ounce
can ‘W' cone

5' ‘353717??fifio

 

coo-owe I”. " uni-v.

coon can. i was. ' .‘v

 

m

 

' I
C

 

 

=" “Afi'u. a.“ 6‘:

 

 

l l I
' L j

_
room-vane .0 cans!”

A.“

®

 

Bl! ICC-ll COCO-ll

CW“. ‘0 I!" "I’M. U-OC. I. .u.“'.

'vlnua

(:cnev Liver Muscle
m “V'- m ICOthG' filial

O“. I”.

    

 
  

.0". .‘h? 7. HOOOATOCV

6-5-34

“M” I. I)" ":9“. CU“. 0. emu“

Source:

     

—————-—
-~“. .' 0....‘n.

 

 

_- -- .—

 

_
."0.‘.. .'

MPH”

lad:

'-
l

FIN-Hutu.“

U.S. Department of Agriculture, Food Safety and

Inspection Service, Meat and Poultry Inspection Program, "Training
Modules for Meat and Poultry Inspection Officers," 1981.

Figure 3.2
Form 6000-2

I?! DIAICTIVI 917.1
loo. 2
3831311 6

100 Chlorinated pesticides
101 Aldrin

102 Ientene‘heaath1orlde
203 Chlordane

10‘ Dieldrtn

103 00? and Metabolites
306 ladrin

107 Beptachlor and Metabolites
103 tindane

109 asthoxychlor

110 Toxaphese

111 rca's

112 Beaachlorobensene

113 Hire:

11‘ Strobane

130 Unidentified rel. rot.
1‘0 Unidentitted rel. out.
131 333

200 Antibiotics

201 Penicillza

202 Streptouyttn

203 Chloranpbendcol

206 Tetracycline

203 Tylostn

203 lrythrsaycin

207 Isonycdn

203 Oxytetracytldne

209 Chlortetrscyclane

210 UnidentilLed litrobtel
inhibitor

211 Gentaaactn su1£ats

212 Ldnconytdn

300 Organic phosphorus pesticides
301 Counaphos

302 01thlorvos venous

303 Dtasiaon

306 tthion

303 Halatnaou

306 Parathaon

307 flannel

303 luelene

309 trichlorton

310 Methyl parathion

311 Dtoxathion

312 01su1£oton

313 Tenetrothdon

314 Gardona

330 Unidenttzded rel. rot.
3‘0 Unidenttlzed rel. set.

‘00 frets eleeents
‘01 Arsenic
‘02 Mercury

‘03 Copper

Source: Ibid.

303
303
‘06
307
303
309
310
$11

68

Esme: Inna-Ingmar

Lead
Itnt
Cadmium
Anticany
Selenium
Aluainun
Titaniu-
Iron

300 Eornones

301

Diethyisttlbestrol

302‘Dienestro1 dtacetats

303
30‘
303
306

307 Tgstosterone prooionsts

tetradiol benaoete
Helenzeatrol acetate
trogescrone
Testosterone

320123 CODE

01 Iorse

11 3u11s

12 Steers

13 Con

16 letter

20 Calves

30 Sheep

60 Gears

30 Swine

61 Young chickens
33 Mature chickens
71 Fryer Roaster
72 Young turkeys
73 Mature turkeys
31 Ducks '

303 throxyorogcsterone scetats82 Geese

309
310
311
312
313

300
301

Chlorusdtnone acetate
Zesralanol. (zeranol)

latrsdiol nonooehnttate

lexestrol
Zeeralenons

Carbenetss
Carbsryl

too Herbicides '

701
702

300
301
302

.303

304
303
306

.307

303
309
310
311
312
330
360

300
301
302
303
306
303
306
307
303

231-0
2.3.3-3

Sella
Sultaethonypyridaznne
Sullachlorvyrtdaaane
Sulradtaethoaine
Su13an1tran
Sultanetnaatne
Sulfecnloropyraaine
Sullanetnoxypyrtdaaine
Sultaneraains
Sultetniaaole
Sulraquinoxalzne
3u1£abrc=d=etheatne
Sultanethaasole
Unidentified rel. ret.
coldestitted rel. sat.

Drugs, general
Clootdol
furarolxdone
Nttroturaaone.
Decoquanats
Honensxn

Iarc Idarole
Cash den
3obensdtne

Figure 3.3

3 1 Rabbits

mnnrtonnnrnunaos

01 7st

02 Ldvsr

03 Muscle

06 xidney

.03 Injectton lesion
06 Other

07 Lung

03 Lynph node

09 Heart

10 Skin

11 Spleen

1! Drain

13 Eye or lye Lesion
1A Peritoneun

13 larva

16 lures {abrtcus
17 Adrenal gland

List of Substances Tested for by FSIS, With Codes

69

marketed along with all other carcasses slaughtered that day, except
under special circumstances such as those described above for STOPs, or
those under surveillance, which will be described in section 3.2.1.1.
Secondly, the inspector is instructed that the date and time of sampling
are to be kept secret from plant management, to prevent any manipulation
of the animals presented for inspection which might bias the sampling.
Further, samples are collected completely at random among all animals
slaughtered on the specified day, so that each animal coming down the
line has an equal chance of being chosen for sampling. This is to be
done regardless of the ability of the inspector to trace the producer of
that particular animal. All available identification material pertinent
to the animal such as back or ear tags, brands, plant lot or feedlot
numbers, as well as the establishment where the animal was grown are to
be included with the sample, however. This is to assist in tracing the
producer should the results of the testing indicate one or more Violative
residues. It should be noted that the proportion of inspectors' time
spent collecting samples is very small compared with other duties. As
an illustration, consider the data from 1982 presented in Table 3.1.

The highest probability that an animal will be sampled is among
calves - only a quarter of one percent. This indicates that an inspector
is asked to collect samples very infrequently, unless at a very large
plant.

Inspection for disease and wholesomeness must be done on every car-
cass regardless of the line speed. Although inspectors have the power to

slow down the line, in high volume plants a great deal of pressure can be

70

TABLE 3.1 Percentage of Meat Animals Sampled at Slaughter1

 

 

Cattle Calves Sheep Swine
Total inspected
no. of carcasses 33,260,932 2,647,362 5,971,542 80,593,850
Total samples
taken for residues 8’6'4 6:715 473 3.922
Percent of animals
subJECt t° ”'5' .026% .25% .0079% .049%

inspection which
are sampled

 

exerted to prevent them from doing so.2 The point is that other activities
can compete with residue sampling for the inspectors' time, especially
when the meat packer operates under narrow profit margins which depend
on high volume production. Although an inspector is strongly motivated
to pursue his sampling duties as instructed by FSIS, she may be less in-
clined to initiate sampling on her own when under time pressure at the
plant.

Form 6000-2 contains two parts. The bottom section is in triplicate;
one part is retained at the plant pending laboratory analysis, one part
is sent to the area inspection supervisor to confirm that the sample was
taken, and the third part, along with the top section of the form, is sent
to the laboratory with the sample. The sample is frozen overnight and

sent to the laboratory the following working day.

 

1U.S. Department of Agriculture, Food Safety and Inspection Service,
Statistical Summary, Federal Meat and Poultry Inspection for Fiscal Year
1982, March 1983. pp. 1 and 6.

2Harrison Wellford, Sowing the Wind, New York: Bantam Press, 1972.
Chapter 3.

71

The inspector's responsibilities continue when laboratory results
have been received. When results indicate no violation (the majority of
cases), no further action is necessary. However, if a violation has been
indicated, and the producer can be traced through the joint efforts of
the regional office and the in-plant inspector and Veterinary supervisor,
the inspector must carry out surveillance sampling.

3.2.1.1 Surveillance Sampling Procedure:
Responsibilities at the Plant

As described in the introductory chapter, a second phase of the re-
sidue sampling and testing program is the surveillance phase. There are
two directions from which the initiative to begin surveillance can come:
from the bottom up (i.e., inspector initiated) and from the top down
(i.e., through regulatory channels commencing at the national, regional,
or area level). In either case, however, it is the in-plant inspector
who carries out the sampling procedure, which closely resembles the sam-
ple collection for the monitoring phase. This section will discuss only
the conditions under which an inspector may initiate surveillance, leaving
the procedure by which surveillance instructions come from the top down
for a later section.

Any organoleptically observed evidence of residues can be followed
Up by surveillance. The performance of a STOP and subsequent confirma-
tion testing, for example, comprise an inspector-initiated act of sur-
veillance. The distinguishing characteristic of a surveillance program
is that it is performed when some reason exists to believe that a residue
problem is present. That is, this type of sampling is not done on a
nationwide, random basis, but is specific to certain animals. These

animals may have exhibited signs of containing residues, such as the

72

staggering or bruise marks mentioned causing the inspector herself to
initiate surveillance. If they are the product of a producer with a
history of residue problems, the inspector should initiate sampling when
the next lot arrives at slaughter (see section 3.2.2.1 for a description
of this procedure). They may have been produced in an area where con-
taminated groundwater or other point sources of contamination such as
animal feeds are known to have existed. If this has happened, and such

a source has been picked up by random monitoring, a surveillance program
will have been designed at the national level (this procedure will also
be discussed in a later section). Inspectors in these cases will receive
specific sampling instructions from the Residue Division in Washington via
the regional office. The same is true for animals produced in an area
which has a history of a particular residue problem, such as consistently
high levels of sulfa.

The sampling procedure differs in several respects. First, a form
6000-1 is used rather than a 6000-2 (see next page for a sample form).
Second, the animal is chosen not randomly from among all animals being
slaughtered at the plant, but because it displays some evidence of con-
tamination with residues or originated at a grower or in a region with
a history of residue problems. Third, in many cases the carcass must be
retained at the slaughter plant and may not be sold until results of the
testing indicate no residue above tolerance. When the in-plant inspector
initiates surveillance the area supervisor must be contacted for a de-
cision on which FSIS laboratory it should be sent to. When surveillance
has been initiated at another level, instructions on where to send the
sample are provided with the request for the sample. If testing indicates

Violative residues, the carcass or carcasses must be condemned or destroyed.

 

.M- $3. 3“” Aa.ue'.33
3 nlflwdo

73

 

usoa uses 3613~C3°.OGRAH
LABORATORY REPORT

Mood mmvruens on m eaters commune)

teams. no.

383352

 

0. 0.93337 NAMI OI CA8! NO.

‘ I J ' J J '

:- RIUYIO SIRIAL NOS.

I A

3. RITA"! TAG NO.

 

D s IMW mosses

 

s
3ssse¢ees

Jason- !OI eauocmc is. seen ano soonest OF 'ROOUCII. valid. on ILOCK Owe-ll 0. costs
hotel I .
‘ lessee Iss- Cites

 

l I I .s

7. ISTAILISRMINT no

 

 

 

 

 

 

 

 

D s 9...... mm... vecees
I I I 1 1 l _ 1 1 I 1 I L- '
‘ eeooucinsncms “sense s. eat (mu to. sex (Chose) “a. Der! COLLICTIO 'He. can manta
as. see. w. m 2' use. Dev w. use. Dev Y'
I 1 4 I D g ' A; 1 I 1 I I 1 I __L_
O “AMI? A. COUNTRY NAM .COO‘ I. FORCION 33?. NO. 6. C9370” ENTRY NO. 0. 04' loans 410 NO.
”00¢? s I
A 1
ea. I160!" 000 lace] may. JPICWIC Auiput
0m!“ lees Fee. Seseiss Code
0 'DW 0 MICNOIOM ”ammo" 0 Room.” Hearings-0G L 1 l l

 

se. sncwtn (Choral
0‘ lot 02 Live!

 
  
 

Doom. 0 seem Dame-«8m

ID as cram”)

u, neurso "nonhuman (Inclusion Maseru. Mrs-dorm. ”Thu-Morten Remotes are Name o! harem-red Latent”!

 

 

 

 

 

 

l.

 

 

flit um! or mantras (Type «mu

 

33. TILI'NONI NUMIIR ”Add!“ A!!! C640}

 

 

 

st. eon “Balaton? US! ONLY

 

 
   

3.. DIAGNOSIS

3. “3. COO!

 

22- OONOITION ON RICIIPT

20. eon CHIMIS?IY U33 ONLY

 

 

D 10»:th a In Violation

I 23. anecvms are?"
.lTaarlO (flaDevJ'IJ I. com-.1?" (Mace.

 

 

 

irfinnraaT—l

JUL? I.”

on. r s.~o. s/ss. sienna“ or seams-rial

IIDLACII so! 3311:1771 semen II OIOOLITI

on e . usoearoev (sane to masts W!

a... a . “mu taunt-«arias (Io-esteemed -
m mm)

s... g . meescros We. ”alum!

Figure 3.4a

Form 6000-1, Side One

 

3
1 I
:27. auaL £303.30 1 ' 2 I I

1
'sn 1 - ”.AIOIA‘I’O-V lasso :o lesions: Office!
(War: Insert ms and seems on .m or so
or memos an self-scorers“! envorooeo

. mt‘rlauia. ICVIIWID llmr.

 

Pen s - casoeareev my

74

REVERSE (F305 FORM 3000-1)
“STRUCTIONS
I?!” 2 - - RELATED 33IIAL N08.
Include the serial number of samples collected from the same herd or flock (or the serial number ofsomples sent

to e aileron: laboratory). Use item is it more space is needed.

1‘ ITEM 3 - - PROOUCTISPECIIS
For residue analysis. use the following:

mess—s , edeBesciss Macias
1 0 Nora-s 8 O Len-Is 71 - Fryer lessor Turkeys

"Oldie 3000“. fl-Yemsfurusvs
903000 31 OMS!!!“ 73-min. Turkey-s
600m 320.com ss-o-uu
“and. 330” '02.“
30'Caloes 31 IYeuneChHIoI- suma-
fl-flsuusm ”One-enemas“-

W 13-- 030038? FOR

I? you check (41 Diagnostic Microbiology. or Food Microbiology. and/or Food Poisoning submit microbiology
moles (as per instructions in Port 231.5qu 23-3). Meet and Poultry Inspection Manuel (MP!- 7).

It the "dues: is for Residue only, Enter the residue code for the analvsis of the class of compounds
Then enter the specific residue in that series: two examples follow:

(I! the case represents e ”dieldn’n followup sample") Entcr- 100 Dieldrin

III the case represents an “arsenic followup mote") Enter . 400 Arsenic

333151331;
i

mm

was is - - “Land INFORMATION (Ins-ham; remark: on airports. dare-Morten. and Posrxorrem and Name ”Recognised
Momma)
3sriel mmoers of some: motes from 0’10 same test lot (Continued tom Item 2).
Serial mmbers of other samples from the same sound:

For Residues . Refer to Meet and Poultry tnsoecsion Manual, Subosrt 20.10 sodMPl Directive 917.1. Heinz.
For Pathology - Include a complete description of me ante-mortem and post mortem lesions and refer
Sudden 230 of me Most and Poultry Insoecrion Manual for remote couscuon. preparation. and smomens
instructions.
new 23 .. - no. H3023 T1803 RICIWID FOR PATHOLOGY

true as - - NO. SUD” PRI’ARIO

WMuMfir- - - ResidueSemplesrpl‘r with Recognised Mercury

For. 1. 2 - lateness W tum cer-one mutt
m Panto - “CW (wanes—resonance!

M 3 - heels Confirmed-e

Due. - We Pile (recorder-mm:

Figure 3.4b

Form 6000-1, Reverse

75

Another responsibility of the in-plant inspector and the area super-
visor is to keep a list of producers with a history of violations. When
a monitoring or surveillance sample indicates an illegal residue, the
inspector at the plant where the sample originated is informed. as well
as the area supervisor. From these test results, a list is generated.
When the producer can be traced, she is informed that five animals out
of the next lot sent to slaughter must be sampled and tested for residues
at the slaughter plant. This procedure provides the producer with both
opportunity and incentive to bias FSIS sampling. She can do so by merely
sending as her next lot only animals she is certain are residue free. It
is the responsibility of this producer to inform the inspector at the
time these animals are brought to slaughter that they are subject to
FSIS surveillance. If testing of these five animals indicates no resi-
due, the producer's name is removed from the list. Should a producer's
name remain on the area supervisor's list for a long period of time,
the supervisor investigates to find out what has been happening to that
producer's livestock. This may indicate that the producer is evading
surveillance by taking animals to another slaughter plant or selling
them at auction without informing the purchasers of the status of the

animals.

3.2.1.2 Implications
Several implications of the structure of inspection for residues can
be pointed out before proceeding to a description of the duties of re-
gional offices. The first aspect to note is that the monitoring system
is not designed directly as an enforcement tool. Its primary objective

is to provide the residue division of FSIS with a statistically accurate

76

picture of the location and magnitude of residue problems. Probably for
this reason, certain difficulties are encountered when FSIS attempts to
wring some enforcement functions out of it. For example, as our examina-
tion of the regions' responsibilities will point out. much additional
work often must be done in order to trace the producers of volative
animals if any enforcement follow-up work is to be accomplished. Even
the task of informing the producer of her violation is difficult. and
going further to impose any fines or penalties (FDA's job) is almost un-
heard of. Therefore one is tempted to conclude that the monitoring sys-
tem fails to perform these enforcement functions, which it is not de-
signed to perform. Furthermore, it fails to satisfy certain criteria

of a successful information system, which it is designed to be. As we
move from the bottom to the top of the organizational chart of FSIS and
then outside of the agency to FDA, more evidence of the ad hoc nature of
adjustments which have been required by various intermediate bodies to
make up for such failures will accumulate.

The second issue is the incentive structure created by FSIS choices
about how to enforce chemical residue tolerances. The monitoring system.
for the most part, imposes costs on the government alone; these include
inspectors' and veterinary and regional supervisors' time, mailing costs
and laboratory time and facilities required to carry out the testing, and
the time of regional officers and inspectors spent in tracing producers
from (private) slaughterhouse records. Almost the only cost imposed on
producers who are traced is any mental anguish which may be caused by the
knowledge that they have contributed to the contamination of the consumer's
meat supply. A conscientious livestock raiser may incur expenses in in-

vestigating the cause and solution of her residue problem, but these are

77

self-imposed and voluntary. If she were sure that her next lot of animals
would be subjected to surveillance, a much greater incentive would exist
for such investigation and action on her part. Under the present system,
however, she may evade follow-up surveillance fairly easily be merely
selling subsequent lots of livestock to a different slaughterer or at
auction.

This option may be less viable to larger producers. As of 1978,
83.7% of steers and heifers (the majority of high quality beef cattle)
were sold from producers to packers directly, with only a total of 16.3%
going to terminal and auction markets:l Under such a thin market, sudden
large inflows of fed cattle would be remarkable and somewhat suspect.

One would predict that the smaller producer would have more success in
slipping a few less desirable animals by inspection in this way. In
fact, the following information serves to support the thesis that more
animals sold at auction are culled cows and bulls: when cows and bulls
are included, the percentage of all cattle marketed directly drops to
73.4%, while terminal and auction sales rise to 10.6% and 16.0% of the

2 Culled cows tend to be the product of farmers for

market, respectively.
whom beef production is a by-product of some other primary activity such
as dairy farming--as far as meat production is concerned, they would be
considered "small" producers.

Under the surveillance programs. costs may be imposed on the private

sector through enforcement actions, but these often are imposed on the

 

1U.S., Department of Agriculture, Packers' and Stockyards Adminis-
tration, Packers and Stockyards Resume, Vol. XVII No. 3. Dec. 28 1979.

2

 

Ibid.

78

owner of the animal carcass after it has been slaughtered rather than on
its producer. That is, if a carcass must be held it not only preempts
productive space in the slaughterer's cooler, but also represents a com-
plete economic loss to that slaughterer if it is eventually condemned
and destroyed (with the exception of any salvage value which may be ob-
tained through rendering or converting into fertilizer). The imposition
of any other costs by FSIS within the limits of these programs, on a
regular basis, is unlikely. Extraordinary enforcement actions such as
seizure of product, product recall, or condemnation of large numbers of
animals on the farm may be taken in extreme cases. The fact is, however,
that a simple, one time residue offense is not likely to cost the pro-
ducer anything except a minor inconvenience, from FSIS's perspective.

FSIS instead relies upon the action of the market to apply indirect
pressure to the producer. The slaughterer who regularly purchases animals
from an identifiable producer may be motivated to offer such a producer
payment on a "yield and grade" basis, whereby the animal grower receives
a certain amount per pound after slaughter,inspection, and grading, rather
than a set price per animal before slaughter. Financial loss associated
with the risk that carcasses will be lost to condemnation is thus shifted

from the slaughterer back to the producer.1

This possibility would seem
to strengthen the incentive of the producer to sell subsequent lots to
different slaughterers or at auction where such risks are unlikely to be

discovered.

 

1Interview with Dr. John Spaulding, July 29,1983.

79

3.2.2 Regional Level1

The five regional offices of Meat and Poultry Inspection Operations
coordinate and supervise operations at the area and slaughter/processing
plant levels, and transmit information and instructions from National
Headquarters to these lower levels. In addition, Veterinary Staff Assis-
tants for slaughter operations at Regional Offices have certain responsi-
bilities in the initiation and follow-up of residue cases within the
region.

Reports from FSIS laboratories indicating Violative samples are
transmitted to both the Washington office and the relevant regional office.
Each violation is followed up by the regional office, with follow-up pro-
cedures varying according to the seriousness of the problem. Monitoring
results are divided into two broad categories: Contamination Response
System (CRS) cases, and a group which will be referred to as "routine"

C3585.

3.2.2.1 Routine Follow-up
Much of the work involved in a routine case can be accomplished by
secretarial or other non—veterinary staff. The major task tends to be
identification of the producer from whatever information can be gleaned
from the 6000-2 form transmitted from the laboratory or from slaughter
establishment records. This identification procedure is accomplished by
telephone from the regional office. and often fails for the following

reasons:

 

1All of the information in this section, except where otherwise
specified, was obtained from an interview with Dr. F. R. Rellosa. USDA
Northeastern Regional Office, on September 12 1983.

80

1. If the animal came from an auction, it may have been identified
only by a stick-on cardboard tag, which is frequently lost in
transport or at the slaughterhouse and therefbre is not recorded
on the 6000-2 form. This is especially a problem in veal calves.

2. Slaughterers keep only whatever records they see fit to keep.
There are no special requirements that their records be kept in
a form useful to the in-plant inspector or FSIS. They may record
only the number and species class received from a particular
source on a particular day; if animals from many sources were
slaughtered on the same day and no individual identification
number was recorded, there may be no way of determining the
source of a violation.

3. Even if the source can be identified, it may be an auction house.
This may not be the source of the residue for reasons explained
in Chapter One. Here, auction records must be examined and the
same difficulties are encountered as with slaughterhouse records.

For these reasons, the Northeast Regional Staff Officer in charge of resi-
due cases was able to identify only 395 producers of Violative veal calves
(sulfa residues) in l983, while 746 condemnations of veal for sulfa took
place. In 1982, 1032 violations were detected for the same species/sub-
stance pair and only 621 identified, and in 1981 417 violations occurred
while 213 could be identified. These figures represent an average of ap-
proximately 55% identification of sulfa/veal violations.1
In addition to the difficulties encountered in tracing previous owners,

regional offices have a limited amount of staff time to devote to following

 

1Pennsylvania State University, College of Agriculture, Cooperative
Extension Service, "RAP: Sulfa Residues in Calves," Unpublished, no date.

81

up residue cases. In Philadelphia, one clerical and one veterinary
staff member's time were allocated to this activity. Some 850 of these
cases had been received in the office during the previous year, or an
average of 16 per week. Thus a limited amount of time may be devoted to
any one violation, precluding any complicated investigations on such
routine cases.

Once the producer is identified, a case file is established. Each
is given a number and filed by producer's name in order to facilitate
a quick check for repeat violators. Subsequent procedures include send-
ing a warning level or violation level letter to the producer, depending
upon whether the residue level was near (warning) or over (violation)
the tolerance level. Letters are also sent to the regional office of
FDA and to the State Department of Agriculture informing them of the vio-
lation. (See following pages for examples of letters similar to those
sent to producers at warning and violation levels, as well as a sample
letter to a State Department of Agriculture.) The producer's letter in-
forms her that animals from the next lot sent to slaughter must be sampled
and carcasses retained pending results in order to assure that the problem
is cleared up. That is, the producer is now subject to follow-up sur-
veillance. Follow-up surveillance usually means that 5 animals will be
randomly chosen from the next lot sent to slaughter for sampling and re-
sidue analysis. The entire lot may be held pending the results. Alter-
natively, the producer may send in the required number of animals in ad-
vance of the rest of the lot. These will be tested and if no residue
violations occur, the rest of the lot (still at the producer's establish-

ment) is cleared for slaughter with no further surveillance. (The

82

April 1, 1910

Example: Producer Letter warning Level

Dear‘Mr. Case no.

 

This letter is to inform you of tests performed on tissues collected
from (species, class) identified as originating from your promise.
The samples were collected as part of Meat and Poultry Inspection's
Policy of sampling livestock and poultry at slaughter. The test
results are listed on the enclosure along with the required

followup sample plan.

The Food and Drug Administration and State Department of Agriculture
have been furnished this information for whatever action they consider
appropriate.

we appreciate your cooperation in this program to minimise residues.

This will protect both the producer from financial loss and the con-
sumer from the intake of excessive residues.

Source: FSIS regional office. Philadelphia, Pennsylvania.

Figure 3.5

Letter to Producer Informing Her that Testing
Revealed a Below Tolerance Level Residue

83

april 1. euro

lasmple: Producer Letter Violation New Case

Certified Mail - leturn Receipt Requested

DearIHr. Case No.

This letter is to inform you of tests performed on tissues collected
from (speciesL class) identified as originating from your premise. The
samples were collected as part of Heat and Poultry Inspection's Policy
of sampling livestock and poultry at slaughter. The test results are
listed on the enclosure along with the required followup sample plan.

The Pond and Drug Administration and State Department of Agriculture
have been furnished this information for whatever action they consider
appropriate.

It*will be necessary to retain the (carcass and other edible tissues)_
from all (species,ic1ass) sent to slaughter pending the analysis of
representative samples from the lot. As an alternative the lot can be
protested. Pretesting means that the required number of individuals
for sampling can be sent to slaughter first. They will be retained
and samples submitted to the laboratory. If the results are acceptable
the remainder of the same production lot can be sent to slaughter
without retention, although additional samples may be collected.

It is recommended that you select a private laboratory recognized by

our program to analyze the samples at your expense. Split samples will
he sent to the Meat and Poultry Inspection'Laboratory (HPI). As an
alternative the samples can be run in the HPI Laboratory without special
”tbritye 7.. “‘3"'6 A K‘CD‘flltéO LAJI‘Afa‘Kf

It attempted to contact you by telephone while this letter was being
prepared. If our efforts were successful you.are already aware of our
residue followup requirements. Please feel free to contact this office
for additional assistance. In addition notify us where find when you

plan to slaughter so that the required samples can be collected. Our
telephone number is area code 202 4&7-2807. It is also important for you
to assure that the {animalsI birds) are identified to the inspector at
the slaughter plant.

we appreciate your cooperation in this program tO'minimise residues.
This will protect both the producer from financial loss and the con-
sumer from the intake of excessive residues.

Source: Ibid.

Figure 3.6

Letter to Producer Informing Her that Testing Revealed a
Violative (Above Tolerance) Level of Residues

84

Ionorable

Bastian s

_This letter is to inform you of the laboratory analytical results-of tests
performed on tissue collected from produced in your State. The
tissue samples were collected as part of the Meat and Poultry Inspection's
residue program, Animal and Plant Health Inspection Service. The results
of the analyses and the individual producers are listed on the enclosure.

The enclosure contains the basic information on the available history and
current findings on this residue problem. The exact status of each producer
is indicated in the column headed "Compliance Status." If the residue
finding is'below the stated regulatory action level, the producer is in
compliance with present regulatory requirements and the column is generally
soaked "In." However, if the residue finding is just below the regulatory
-action level, we believe the producer should be advised he may have a prob-
lem and, therefore, we mark the column "Warning." -Naturally, any residue
finding above the regulatory action level is not in compliance and the
column is marked "Out."

we have already furnished the Food and Drug Administration with the same
information for whatever action they consider appropriate. We would
appreciate your cooperation in this investigation to eliminate and mini-
mise residues found in . . This‘will protect‘both the producer from
financial losses and the consumer against the intake of excessive residues.

,Sincerely,

Enclosure
Case No.

Source: Ibid.

Figure 3.7

Letter Informing State Department of
Agriculture of a Violation

85

April 1. 1976

Example: Producer Letter Continued Violations

Certified Mail - Peturn Receipt Requested

Dear Hr. Case No.

Uh are adding the enclosed information to our file on this case. It
now appears that the measures you have taken to prevent excessive
residues have been unsuccessful to date. As a result the enclosed
sampling plan will be utilized on future lots sent to slaughter. Re-
tention of ( ) pending test results will be continued
until the problem is resolved.

Uh suggest that you consult with (your veterinarian, county agent, state
department of agriculture, university, FDA) as appropriate to help with
this serious problem. If we can be of further assistance feel free to
contact us. Our telephone number is area code 202 647-2807. In
addition continue to provide advance notification of time and place of
slaughter.

Source: Ibid.

Figure 3.8

Letter Informing a Producer Who is a Previous
Violator that Results of Follow-up Surveillance
Demonstrate the Continuing Presence of a Residue Problem

86

potential for abuse of this practice was discussed in section 3.2.1.1 of
this chapter.)
3.2.2.2 Contamination Response System (CRS)1

At the regional level, Meat and Poultry Inspection Operations (MPIO)
Staff are responsible for identifying monitoring program results which
indicate an early warning condition requiring a CRS. Such conditions
include findings of various levels of residues of particularly dangerous
chemicals. These include most of the chlorinated hydrocarbon and organic
phosphate pesticides (refer to list on page 68) and the trace elements
mercury, lead, and cadmium. The levels which represent a warning that a
CRS is necessary range from "any detectable level" in PCB, PBB, or Kepone,
up to 7 parts per million in Toxaphene.

Field Operations respond to a warning condition at the regional
level in much the same way as to a routine case. The producer or owner
must be informed by telephone, and information pertaining to the cause of
the residue acquired if available. Any related product still at the
slaughter plant must be retained. The same type of case file must be
prepared as described in the previous section and regional FDA offices
must be informed. MPIO regional staff may also request information from
the regional FDA office on other incidents which may have occurred in the
area involving the same contaminant.

However, the Residue Evaluation Division in Washington must also be
informed. National level staff must make the determination, in coopera-

tion with the contamination response team (comprised of various USDA

 

‘Information in this section taken from U.S., Department of Agricul-
ture, Food Safety and Quality Service. Report on the PCB Incident in the
Western United States, January 1980. pp. 89-l06.

87

staff members in Washington), about whether further sampling or a sur-
veillance program is needed. If so, the region will be informed of the
size and nature of such a program and will implement it.

More serious indications of a contamination--such as a clustering
of Violative samples in a particular area--are likely to be discovered
at the national level. This is one of the uses of monitoring system infor-
mation at the national level. When a CRS is initiated from the top down,
the region is informed about any special sampling requirements which it
is expected to implement. Regions may also be asked to assist in the
coordination of any product recall or seizures necessitated by a residue
break, which can be carried out by State Departments of Agriculture or

FDA.

3.2.2.3 Other Surveillance Programs
Aside from CRS programs, many other situations call for the initia-
tion of surveillance programs. These include clustering of antibiotic,
sulfa, or estrogenic compound residues in a particular area. In such
cases, the regional office may receive a request from the Residue Division
in Washington for a number of samples to be taken randomly within a par-

ticular state or over the entire region.1

The particular conditions which
cause such a reaction at the national level will be discussed in the next
section. Regional offices are instructed in the taking of such a random
sample over whatever sampling frame Washington requires within the region.
In these cases, the regional office is merely implementing an initiative

from the national level.

 

1Interview with Dr. John Spaulding, July29. l983.

88

To summarize, we have observed that surveillance sampling programs
in general increase the probability of detecting residue violators. They
do not necessarily increase the probability that any enforcement actions
will be taken against such violators, since this depends also on the abil-
ity of FSIS to trace the violation to its source and its willingness to
prosecute, or FDA's willingness to investigate and prosecute. In any case,
surveillance may be initiated at the slaughter establishment level, by
the inspector who observes a symptom of residue violation in a particular
animal or lot of animals, or at the regional level by the Staff Assistant
for Residue, who generates follow-up surveillance in the case of a trace-
able violation picked up by the monitoring system. Let us now turn to
the generation of both monitoring and surveillance programs by the Residue
Evaluation and Planning Division at the U.S. Department of Agriculture

headquarters in Washington, D.C.

3.2.3 National Level

The overall program design for both monitoring and surveillance is
accomplished at the national level. The division having primary respon-
sibility for determining sampling rates and frames is the Residue Evalua-
tion and Planning Division within the Science program of the FSIS. Sam-
pling plans are designed with the assistance of the Mathematics and Sta-
tistics Division, also within the Science program (see FSIS organizational
chart at beginning of chapter).

The program of testing for chemical residues in meat had its incep-
tion in 1967, and has been operating in its present form only since l972.
Hence it is relatively "new" when compared to the seventy-eight year his-

tory of inspection of meat for the more readily detectable sanitation and

89

disease problems. The size and scale of the program have grown consider-
ably over the past 12 years, emphasis has shifted and broadened, and
testing methods increased rapidly in sophistication. Therefore, there
are many aspects of the program which could still be described as experi-
mental. Seen in this historical context, it is not surprising that de-
sign of surveillance and monitoring programs is an ongoing process, with
small adjustments and innovations being made continuously. A primary
characteristic of Residue Division's programs is their adaptability to
new situations such as market developments, introduction of new sub-
stances, and changing producer practices.

This section will describe the continual process of interaction and
redesign of the sampling programs by the Residue Division at the national
level. In doing so, some of the important factors in decisions about
changing the sampling rate or probability of detection of violations

will be pointed out and discussed.

3.2.3.l The Monitoring Phase
In some of the literature describing the monitoring program, it is
stated that the program is designed in such a way as to provide a 95%
level of confidence that a l% or greater annual incidence of violation

of a given tolerance among a species of animals at slaughter will be dis-

1

covered. To do so, a sample of 300 from each "population" is required

2

in order to determine the annual violation rate for a tolerance. Many

 

1U.S. Department of Agriculture, Food Safety and Quality Service,
"Improving the F505 Residue Program," July l980. p. 9.

2Determined statistically using the binomial tables. Interview
with Dr. W. Burke, Mathematics and Statistics Division, Séptember l3 1983.

90

different tolerances exist and therefore many types of violations are
possible. The potential for a different violation is created each time
a particular species of animal comes into contact with a drug, pesticide
or toxic substance whose presence in meat is regulated (i.e., for which
a tolerance or action level exists). Therefore each population, con—
sisting of a species/substance pair, must be sampled 300 times within a
year to provide the level of statistical accuracy promised by FSIS.

However, the data on the results of monitoring sample testing indi-
cates that the number sampled for a given species/substance pair ranges
from zero to several thousand for any particular year (see Table 3.2).
What explains this apparent contradiction?

The size of monitoring samples for a given month is set by Residue
Division staff members in a monthly meeting. A list of species/substance
pairs and the number to be sampled from each is supplied to the Mathema-
tics and Statistics Division which determines the plants from which such
samples are to be collected with a computer program. The program chooses
randomly from among all slaughter plants subject to federal inspection
and prints out 6000-2 forms which are then sent out to the in-plant in-
spectors. The slaughtering plants are weighted according to the volume
slaughtered at that plant in the past; a large plant therefore has a
greater probability of being chosen for sampling than a small plant.

This is done in order to give each animal passing through the point of
slaughter an equal chance of being selected for testing for chemical
residues.

Several factors other than statistical validity enter into Residue
Division's decisions on the size of the monthly monitoring sample for each

species/substance pair. For example, some species such as mature turkeys,

91

TABLE 3.2 Number of Samples Collected Under National Residue Monitoring
Program l973-l982, Selected Species, by Calendar Years1
(number sampled/number found Violative)

 

 

mmancsz'3' 5

1973 1974 1975 1976 1977 1978 1979 1980 1981 1982
m1- -— - 15/0 13/0 99/0 179/4 84/1 93/1 195/0 11.x.
Dunn - 20/0 123/0 98/1 216/1 224/0 94/0 99/0 335/0 14.».
cou- 1595/42 1599/16 322/6 346/6 1207/21 1172/43 935/32 915/13 473/3 14.11.
was 2/0 18/0 96/0 89/0 217/0 227/1 89/0 103/0 320/1 8.11.
mm 1889/149 3118/97 2237/156 1377/77 1120/46 1408/94 1485/91 1382/76 1094/30 11.21.
cusp 23/0 300/1 57/2 69/3 149/2 128/5 185/0 332/1 505/1 14.14.
m 840/15 307/8 150/4 247/2 449/6 1402/80 1618/28 1613/3 999/8 11.11.
m
m 683/4 313/2 179/5 100/1 248/0 319/7 310/0 309/1 327/1 11.8.
Ibture
cum:- — — — 60/0 118/0 152/2 193/0 208/0 300/0 ILA.
v.8. 43/0 145/1 119/2 107/0 140/0 125/5 110/0 94/0 172/0 11.71.
W
1:::? 132/1 85/1 351/15 116/o 261/3 248/7 210/1 290/1 325/6 8.x.

m"5

88am 15/1 115/0 89/0 35/0 29/0 124/2 76/0 151/1 69 '0
cou- 344/2 211/2 261/0 219/0 120/4 178/2 426/6 183/2 313/1 241/2
1811626 - 6/0 121/0 119/1 20/0 31/0 115/0 72/0 156/1 67. '0
061m 306/3 297/5 336/5 331/4 166/5 214/6. 411/11 263/5 444/13 1380 '32
M -- -- 91/0 81/0 7/0 25/0 58/0 6/0 11.8. 12 0
Mn. 1 70/1 343/33 383/35 1512/145 9410/1242 6682/647 5991/387 4622/204 3488200 1052 "51
m
cm 1 181/0 311/8 327/0 170/0 1/0 101/1 210/0 11/0 316/3 288 0
lhtum __ _ -
m 10/0 210/0 175/1 18/0 57/0 8/0 7010 182. 0
2.0.

M11417. 15/0 57/2 300/11 265/ 2 161/1 14 5/13 138/1 133/2 1W0 65/0

m
mm: 23/0 161/0 421/20 319/4 265/3 272/7 294/13 422/20 534/10 390 '5

 

Imus. rot-u. nature mm. dun. geese. ml rabbits excluiad.

2n): 118: of nab-tame: included 171 each category. on. page 11.

JI'SIS data did not specify “the: violatiau or purely m We level of W as
W for this category

‘Dnu for sulfmmlde Implmg (run 1977 1:0 1m includes 178910101 minutes u all 88
mimrlng unples

sullmpluinuesemwflmmtoudtorglmmmgm.

m: —-1b8qlumllactdz o-mooncuamtmmmu; 8.5.nwtmnahlo.

92

TABLE 3.2 (Cont.)

 

mm mums”?

1973 1974 1975 1976 1977 1978 1979 1980 1981 1982
7/0 41/2 113/9 110/7 32/0 30/1 69/0 126/1 114/1 136/1
118/1 227/0 519/8 231/0 91/0 79/0 114/0 254/1 277/0 269/0
528/15 603m 771m 124442 656/7 781/9 962/4 1111/11 321/1 341/1
57/1 126/o 330/3 200/3 101/0 86/0 113/0 243/0 275/1 267/0
84/1 284/5 269/7 327/8 124/6 207/2 575/4 406/4 276/0 305/0
249/3 267/3 292/2 206/1 75/0 81/0 171/2 241/0 377/0 246/0

giggiiggigé

232/0 329/5 324/4 442/6 . 215/0 4157/2 1305/3 1818/8 673/6 779/9
135/1 470/5 362/0 301/1 05/7 ZN/ 0 235/0 596/0 949/0 434/0

 

 

 

395/0 568/8 415/2 546/0 289/1 164/o 247/1 550/3 698/0 348/3
2.8.
W 114/0 249/0 209/0 116/0 , 60/0 50/3 73/0 127/0 136/0 64/0
”:3” 360/10 477/6 250/0 227/0 156/0 132/2 270/0 327/0 423/0 317/0
ammo moms m5“

8811- 4/0 -— - -- - - - -- - -
866-:- 155/0 71/0 1/0 - - 24/0 — 23/0 234/0 54/0
on. 21/0 5/0 - 1/0 — 38/0 - - - 9 '0
18mm 132/0 73/0 — 1/0 - 22/0 . — 23/0 244/0 530
film 0/0 1/0 - - - - — 9/0 77/0 18 0
amp 133/0 122/0 - — -- 22/0 -- 15/0 - --
881m 297/0 147/0 - - - ‘ 86/0 — 52/0 262/0 64 '0
m — .- — .- — on. '
cum 126/0 109/0 27/0 10. 0
14am
unclean 270/0 36/0 3/0 - — -- — - 50/0 24/0
v.2.
Minn 38/0 22/0 - -- — -- — -- - --
M
W 130/0 87/0 1/0 - -— - -- —- — 9/0

5

All-mpluinthemmmmmngMtMm.

7

Nanchmr 18 included (run 1977 «and. Stratum 18 occluded tor 1973. It 18 interesting
83 note that in 01108138158. Dieldrin, 001‘. Btu-in, Heptachlor and 11:8. altlnagh an nunhnr of
violatlau 18 low. 1:13 amber of 33818.1». hat m—violatlvu 18811“ 18 high.

0
00:81:71 838 tested for in 1902 only. Gamma. unethical-1, hlchlorfm, and Dimlfabm were
not tutor! for in 1%1-1982.

DE: --lb8n1u(b118cudx o-wumllcbdhatmvlomfaud

TABLE 3. 2 (Cont. )

93

 

 

 

 

nus. m6
1973 1974 1975 1976 1977 1978 1979 1980 1981 1982
8.1118 - — - -- - -- - -- — -
88-6:- - - 1/0 248/0 147/0 98/0 23/0 52/0 148/0 122/0
a». .. _ .... _ _ — _. _ .. ..
8811c. - - — 142/0 81/0 85/0 17/0 43/0 135/0 125/0
0811768 -- - — — -- -- - - - ..
818-0 — - - 21/0 86/0 29/0 13/0 - 62/0 20/0
8171» 29/0 828/3 661/2 703/5 503/0 212/0 391/2 64/0 283/0 85/0
m - 151/0 1314/5 874/4 382/2 480/0 38/0 160/0 327/0 190/0
144m
Chisinau - - -— - — - - - - -
8.11.
man-yo - 2/0 123/2 119/0 3/0 6010 - - -— —
M
W - 6/0 340/15 188/0 56/0 211/0 16/0 -— — -
1113843036"

81111- - -- 80/0 22/0 — -- -- -- - -
m- 2516A6 1803/6 1580/22 1215/12 1112/19 1111/9 919/0 505/0 52/0 116/0
onu- - -— 90/0 21/0 - - ' - -- - --
mreu 831/1 851/3 745/7 561/1 329/2 533/0 295/0 78/0 337/0 303/0
041m -- 1/0 - -- -- - - -- - --
M 474/0 192/o 225/0 120/0 164/0 114/0 81/0 17/0 — --
Mm - - - - — - - - - -
M
cum- -- -- - -- - - —- -- -- -
141mm
emu-u - — -— - — -— — -— - —
8.11.
W - - - — - - — — — -
M
m: — - - — -- — — — - —

6

Mmhlnthecmmqmusmfm‘

Wmudlunmaggregawmallm'

9
1183118180125“an

- - 1b 5.19188 museum

0-Sllple8mllccudhatmvio18uautm

94

rabbits, and geese are so infrequently slaughtered that requests to in-
spectors in plants slaughtering these animals must go unanswered. More
importantly, logistical factors constrain the numbers of samples which
can be handled by a particular laboratory. With a given amount of labor-
atory time and technicians to do testing for residues, the Residue Di-
vision faces an allocational problem. When emergency situations such as
the allegations of impurities in groud beef produced for school lunches
by Cattle King arise, analytical facilities are strained and the number
of monitoring samples for some species/substance pairs must be reduced.1
Similarly when large scale surveillance programs are underway, monitoring
sampling may have to be cut back to compensate. If monitoring for EDB/
species pairs is increasing, monitoring for penicillin/species pairs may
have to decrease because the total number of samples (testing capacity)
of the laboratories is fixed within a given time period. Although an
annual plan for the monitoring program is drawn up by the Residue Division,
it must be abandoned when surveillance requirements arise. Thus the two
programs, monitoring and surveillance, are interrelated in the implemen-
tation stage although the objectives and uses of the data generated by
each differ.

Laboratory resources are required not only to test for residues,
but to accomplish many of the other functions of USDA as well. In 1979,
laboratory upkeep and expenses totalled only 4.8% of the Meat and Poultry
Inspection Program's budget. 0f the capacity provided by this expenditure

80% was employed in checking for economic adulteration of meat products

 

1"Agriculture Department to Investigate a Meat Plant in Denver,"
New York Times, September 20 1983.

 

95

(checking that allowable percentages of water and other ingredients were
not exceeded). Out of the remaining 20%, half was used to confirm veteri-
nary disease diagnoses. The remaining l0% of the total expenditure on
laboratory facilities was allocated to residue analysis (Roberts, l983).
This emphasizes the fact that constraints on resource allocation are also
imposed from the outside. Allocational decisions were being made in which
residues apparently played a very small part in l979. Even had the Resi-
due Division wished to divert funds from other uses in order to drastical-
ly step up sampling and increase the probability of detection, it could
not have obtained laboratory space in the short run which was allotted to
other programs. (Many programmatic alternatives suggested by the regula-
tory enforcement theorists are outside the realm of the feasible for the
policymaker.)

At this point, we need to examine further the situational character-
istics which lead FSIS to initiate surveillance sampling. These programs
are related to monitoring not only because they compete for laboratory
space, but also because they comprise enforcement of tolerances inasmuch

as they raise the perceived probability of detection in violation.

3.2.3.2 The Surveillance Phase
Surveillance programs initiated at the national level differ to some
extent from those which are initiated and implemented at the plant or re-
gional level. Since programs initiated at the lower levels are in the
nature of standard operating procedures they comprise a programmed re-
sponse to a specific environmental signal. For example, when the inspec-
tor sees signs of residue contamination in a carcass, she initiates a

"surveillance" program on that one carcass. When a previous violator is

96

identified bringing a subsequent lot to slaughter, five animals are sam-
pled in a surveillance "follow-up." When certain specified results of
monitoring are received at the regional office, CRS procedures are begun.

Problems which must be dealt with at the national level are, by a
process of hierarchical elimination, those too complex or potentially
dangerous to be responded to in any standardized fashion. Each has as-
pects which require individualized attention, since scope, location, com-
pound, species, and gravity of the problem differ from case to case.

Since each case has special requirements, a systematic delineation
of the factors in agency decisions about how to change the sampling rate
or frame is difficult. It has, however, been possible through conversa-
tions with the director of the Residue Division, members of his staff in
Washington, D.C., and members of MPIO Regional staff in Philadelphia,
Pennsylvania, to broadly classify six factors and their general effect
on agency decisions (these are summarized in Table 3.3).1 The aggregate
effect on FSIS decisions of these factors within a given time period is
subject to the constraint of fixed laboratory capacity. The outcome of
agency decisions in terms of s0rveillance program design in any particular
case could be affected by any combination of these factors, which are ac-
tually nothing more complex than a selection of situational characteris-
tics to which FSIS has responded in the past. They are as follows:

a. Indications of an industry wide or regional problem:2 Such in-

dications are one type of information that the monitoring system was

 

1Please note: these categories are the researcher's interpretations
of statements made by agency personnel, and not direct quotations.

. 2Information on these six factors in FSIS decisions was gathered dur-
ing an interview with Dr. Spaulding on November 23 1983.

TABLE 3.3 Factors in FSIS Resource Allocation
Decisions: Initiation of Surveillance

 

d)

e)

1")

Indications of industry wide or
regional problem

Acute problem
i. Extremely toxic
ii. Banned substances

Systemic/spreading potential

FSIS "effectiveness"
i. Testing methods
ii. Availability of substitutes

Consumer concern/media attention

Costs to producers

Examples

Sulfa in swine
Antibiotics in veal

Dieldrin, Aldrin
DES

Feed contamination:
PCB incident in
western U.S.

Antibiotic STOP
Sulfas

£08 in Florida,
telarche in Puerto
Rico

General concern in
all cases

 

98

designed to provide. For example, high violation rates were observed in
monitoring for sulfa drugs in swine during the mid l970's. As a result,
sampling was stepped up in regions where hog production is concentrated.
The way in which this is done varies, but often takes the form of a

random sample of l00 or more animals chosen within a state such as Georgia.
That is, the population of hog slaughterhouses from which tissue samples
can be taken, which under normal monitoring conditions includes all such
establishments subject to federal inspection throughout the country, is
narrowed down to those within a particular state or region. The sample

is then randomly chosen from this smaller population.

As to the choice of sample size for a state or regionwide surveillance
program, it approximates the following: a small number, say 30, will be
chosen during the first month of implementation. If no reaction is ob-
served by Residue Division the following month, the sample size may be
increased to l00 per month. A "reaction" can mean several things; one
reaction the agency seeks is an observable lowering of the violation rate.
A second type of reaction is communicated through more informal channels
and might be characterized as the creation of a "fuss" within the indus—
try. Producers upset by being detected and traced in violation communi-
cate their distress either directly to the Director of the Residue Divi-
sion or to their industry's representative in Washington, or alternatively
to their Congressional representative who may take the matter up with Dr.
Spaulding or his superiors.

According to Dr. Spaulding, an intermediate goal of raising the
sampling rate is to create a "critical mass" of violators within a parti-
cular industry. He estimates that approximately lOO violations detected

by a surveillance program directed at a region or industry within a short

99

period (several months) is necessary. Eventually it is hoped that this
significant proportion will communicate to the industry as a whole that

a problem exists and needs to be paid attention to by all of its members.
This may be done through the producer group or informally through word of
mouth. The existence of a strong association between members of the in-
dustry is thus a key to successful enforcement when these methods are
being used by the agency.

b. An "acute" problem: A second factor which affects the reaction
of the Residue Division is its perception of the level of danger caused
by the existence of particular reSidues. This means that the reaction
to some residues is quicker and more severe than the reaction to others.
Indicators of acuteness fall into two categories, to either of which the
Residue Division may react after a single incident of violation.

i) Extremely toxic substances: carcinogens and chlorinated
pesticides fall into this category. A typical surveillance reaction to
the discovery of a high level of Dieldrin during the monitoring phase,
for example, would be designed to discover the source of the contamination.
A recent incident of this kind in Mississippi caused FSIS to initiate
surveillance throughout the state until the source could be pinpointed
and removed. Chlorinated hydrocarbons like Dieldrin are considered parti-
cularly dangerous since residues from these substances dissipate very
slowly. That is, they are persistent both in the environment and the fat
of animals. Therefore the discovery of a single residue incident under
monitoring may indicate that the animal has been contaminated for many
months; during all that time the same source may have been affecting many
other animals. Contaminated groundwater used for watering beef cattle is

an example of such a potential source. The statistical design of the

lOO

monitoring phase is such that early detection is not guaranteed. Sur-
veillance under these circumstances is an attempt to catch up with other
animals originating from the production system within which a source of
contamination has been revealed by monitoring.

ii) Banned substances: Such substances have not been approved
for use in animal production, or have had a previously granted approval
withdrawn when new toxicity or carcinogenicity data was presented. Die-
thylstylbestrol (DES), an estrogenic compound, is an offender in this
category. Allegations by the FDA in early 1983 that DES was being used
at four New York state veal producing establishments resulted in initia-
tion of a major program by the Residue Division. Not only were resources
expended to step up sampling at slaughter for DES in the area, but USDA
entered into judicial procedures along with FDA to enjoin veal producers'
use of the substance, and participated in seizure of product.-I

c. A systemic problem with a potential of spreading: Feed con-
tamination incidents are the most common in this category. An incident
of this type in 1979 resulted in the design of the Contamination Response
System by FSIS. PCB's from an unused transformer had leaked into cattle
inedibles which were subsequently converted into poultry feed. Detection
occurred only when a chicken fed with the contaminated feed was randomly
selected for monitoring sampling. A great deal of detective work was
required to trace the source of the PCB within the bird and the source of
the feed from which the PCB had been absorbed. Furthermore, the contami-

nated feed had been sold throughout several states to many different

 

1Interview with Ann Petersen, USDA, General Counsel's Office, Sep-
tember lS l983.

10]

producers. Once these were traced, poultry products originating from
these producers had also to be traced forward through marketing channels.

This incident illustrates that a combination of the factors causing
FSIS to enter the surveillance phase can strengthen its reaction. PCBs
(Polychlorinated Biphenyls) are considered particularly dangerous or an
"acute" problem by the agency. The PCB problem originating in Billings,
Montana which is described above also demonstrated a potential for affect-
ing many other poultry flocks and hence a large number of consumers were
considered to be at risk from a very hazardous chemical. Therefore, the
agency expended a large amount of financial and technical resources on
the solution of this problem, including initiating several producer spe-
cific surveillance programs and participating in the seizure of product.

d. A problem which FSIS can be effective in solving: These fall
into several categories.

i) Testing methods: One requirement is that effective testing
methods are available; without these, little can be accomplished by the
agency. The existence of reliable testing methods with a very quick
turnaround time, and which can be accomplished at the slaughter plant, is
a positive factor in causing FSIS to commence a surveillance program.
Such a technique is available for sulfa and antibiotics, so although resi-
dues of these substances are not usually considered life threatening, in-
plant programs involving "Swab Tests on Premises" (STOPS) are often ini-
tiated. The prevalence (indication of an industry wide problem) of sulfa
in veal calves and swine and antibiotics in veal calves is another con-
tributing factor to FSIS's allocation of resources to STOP testing for

these substances.

102

STOPs provide a way of relieving the laboratory resource constraint
to increasing the probability of detection. Without STOP, sampling an

extra lOO animals would cost the agency $5000 to $7500.1

Additionally,
the agency must consider the opportunity cost of doing lOO sulfa or anti-
biotic tests against the foregone possibility of doing tests for £08 with
those laboratory facilities. With STOP, the only cost is the additional
inspector time required to perform and analyze the swab, the cost of
laboratory tests to confirm positive results, and the cost of the testing
materials. Since FSIS pays inspectors on the basis of a normal 40 hour
work week in any case (overtime must be paid by the slaughtering plant)
this extra duty will not cost the agency any wages. As for confirming
positive in-plant results, these comprised only 6% and 9% of the total
in-plant STOPS performed in 1982 and the first eight months of l983. Of
the positive in-plant results 8l% and 87%, respectively, were confirmed
at the lab.2 This means a total of l065 violators were caught during the
period out of only l275 tests performed in the laboratory. In the ab-
sence of STOP, l7,6l2 laboratory analyses would have had to be carried

out to achieve the same result, at an expense of some $880,600 to $1 ,320,900

(using Dr. Spaulding's estimates of laboratory costs). This is a savings

 

1This range is derived from an estimate of the cost of analyzing a
sample for a given substance or group of similar substances of $50 to
$75 made by Dr. Spaulding during an interview November 23 l983.

2STOP data supplied to the researcher by Dr. Spaulding.

103

of $816,850 to $1,225,275 in laboratory costs which could be put to use
in other monitoring or surveillance programs:l
ii) Availability of substitutes: Another category of factors
which enable FSIS to enforce tolerances effectively are those that affect
the ease of producers' switching to alternative practices. The avail—
ability of substitutes which perform the same function as the one causing
the illegal residues without creating the health hazard means that pro-
ducers, once informed of the problem, can change their behavior without
great loss. The cost of such substitutes is a critical factor also in
an industry with narrow profit margins. An example is the use of powdered
sulfa in medicated feed for swine. Electrostatic clinging in feeding
and storage equipment created a carryover problem which contaminated
finishing feed (which must be free of sulfa in observance of the with-
drawal period). New regulations were proposed in l98l in an agency
attempt to reduce persistently high violation rates: these would have
reduced the concentration of sulfa allowable in on-farm mixing (as op-
posed to registered medicated feed mill mixing). Options open to pro-
ducers who did not wish to pay the higher costs associated with the new
requirements included "substitution of Tylan-Sulfa, an alternate feed
additive, using waterers as the route of sulfa administration or substi-

tuting a combination of growth promotants, rhinitis vaccinations, and

 

1This is merely a rough estimate to give the reader an idea of the
magnitude of the savings possible through the use of inexpensive in-plant
procedures. One would have to subtract from these figures all of the
program initiation and maintenance costs for STOP to get an accurate pic-
ture.

104

selective therapeutic use of sulfas or other drugs in place of traditional
reliance upon ASP-250" (Kramer, 1982, p. 387).1

It is interesting to note that one substitute for feed medication is
the improvement of sanitation and other facilities for animals, or im-
provement of manufacturing practices in general, in combination with
special treatment for the rhinitis which sulfas are effective in prevent-
ing. That is, it is possible to achieve reduced morbidity and mortality
and reduce sulfa applications simultaneously. This may also be true with
the generalized use of antibiotics in animal feed, which some claim is
not only indiscriminate but uneconomical (Kramer, 1982, p. 251). The
animal producers' perceived benefit from subtherapeutic feed applications
of drugs, in other words, is greater than the actual benefit derived.
Bringing these two things into line might reduce improper usage and thus
observed illegal residues; in this case no further enforcement action
(or expenditure) need be undertaken.

e. A problem which has raised a "valid" consumer concern: Several
"barometers" of the feelings of the general public about particular resi-
due problems are considered by FSIS when decisions are made about whether
to initiate special surveillance or other enforcement programs. One of
these is political pressure which may be brought to bear on the USDA by
Congress as a whole or by individual Congressmen and women as representa-
tives of their constituents. A second "barometer," which is extremely
important in both reporting and shaping public opinion, is the media.

The Residue Division must have answers to questions the press will ask

 

1ASP-250 is the trade name of a commonly used form of sulfa for
medicating swine feed.

105

before they are asked; this means that it must be able to predict issues
which will create public concern before they are known to the public.
This provides it with the necessary advance notice to carry out programs
which can provide information on the magnitude or gravity of a problem.
Thus the Northeast regional office was instructed to carry out random
surveillance sampling for Ethylene Dibromide (EDB) in September 1983,
even though the alarm was not raised about this potential toxin in the
press until several weeks later, and even then only with regard to cit-
rus and wheat production in Florida.1
The question of what constitutes a "valid" concern from the point of
view of the Residue Division is a complex one. It is partially revealed
through observation of the way in which the agency has elected to spend
its resources in the past, but a perusal of the types of surveillance
programs which have been implemented might easily lead to erroneous con-
clusions for several reasons. Firstly, one might be confounded by the
intervention of factor d described above. Programs might have been im-
plemented primarily because FSIS was able (had the technology) to do so
in an effective manner. Secondly, the Residue Division may consider the
concern valid but decide that a different approach than surveillance is
required. Dr. Spaulding has often emphasized that a consultative, edu-
cational approach can be more effective in changing producer behavior
than simply raising the sampling rate. In these cases, a valid concern
might not be responded to via surveillance, but instead by a "RAP" type

program (described in section 3.2.4).

 

1Observed during September 1983 visit to that office by the research-

er.

106

Simply having been the focus of media attention is not enough to
ensure that FSIS will allocate resources to a response to any particular
issue. For example, Dr. Spaulding expressed doubts about studies claim-
ing that use of certain antibiotics in animal feed is contributing to the
creation of a pool of bacteria with genetic resistance to treatment with
these drugs when used on humans. He was unconvinced about the scientific
validity of these studies and and therefore did not support efforts to
ban subtherapeutic applications of antibiotics in animal feeds. This
demonstrates one way in which personal assessments within the agency about
the validity of public opinion formed by exposure to the media's inter-
pretation of controversial experimental results can be translated into
action (or inaction) by the enforcers.

f. A consideration of the potential costs to livestock producers:
The threat of FSIS action on a residue problem may involve an investiga—
tion by FDA, collection of samples throughout an entire management system,
condemnation of large herds of market-ready animals, closure of contami-
nated facilities, and other costly inconveniences. For example, a single
detection of a Violative residue in a hog originating from a large beef
and swine raising operation in Iowa in February, 1980 resulted in the
initiation of a producer specific surveillance program under which 20 of
the 30 hogs tested from his next lot sent to slaughter were discovered
to contain Dieldrin. On adjacent premises owned by the same individual
35,000 hogs and 3,400 head of cattle were affected. The source of con-
tamination was determined to be corn purchased from a local grain supplier.
By the time the animals which could be salvaged were decontaminated,
thousands of samples had been analyzed, the value of the production oper-

ation had declined to virtually nothing, the grain supplier had agreed

107

to pay a $6 million settlement to cover damages to the producer and his
creditors, and 6,500 animals had been slaughtered and rendered for meat
scraps with the fat discarded (organochlorines such as dieldrin collect
in fat) (National Association of Federal Veterinarians, 1983).

This is not to imply that many of the violations discovered by FSIS
in its monitoring program are as dangerous or costly as the incident
described above. Yet the potential is there, and the Residue Division
has indicated that it views animal producers as well as the meat-eating
consumer as a part of its constituency. After all, it is part of an
agency whose overarching responsibility is facilitation of the marketing
of agricultural products, which may occasionally conflict with its own
responsibility to protect the consumer. Furthermore, imposing costs
on producers which can be passed on to the consumer in the form of higher
prices serves no one unless the costs are incurred in preventing a clear
danger to human health. From an economic perspective, all costs to
society of enforcement and their distribution should be considered, as
well as benefits.

3.2.3.3 Objectives of Surveillance:
Information and Enforcement

As the factors mentioned above illustrate, surveillance programs
have several objectives. Investigation and information gathering is a
primary goal. Stepped up sampling gives a more detailed description of
a specific population than does monitoring. Specifically, the Residue
Division seeks an indication of the source or cause of residue problems
which display a potential of spreading (or having already spread) to
other producers. Contaminated feeds, when sold over broad geographical

areas, have been a particular culprit. Since the collection of samples

108

on farms involves cooperation with the FDA, only "acute" problems can be
given a full investigation, and hence the concentration on chlorinated
pesticides and industrial chemicals with a tendency to endure in the
environment and in the fat of animals.

Surveillance contributes to enforcement in several ways, both direct
and indirect. It provides some of the information necessary to detect
and trace a violation to its source. As such, it at least creates the
potential for catching a violator after the fact and prosecuting or other-
wise imposing costs (the extent to which this has been done will be dis-
cussed below in section 3.3). Indirectly, increased sampling creates an
atmosphere wherein a larger absolute number of violations is being dis-
covered and traced to its sources. In Or. Spaulding's opinion, this
seems to start a type of dynamic within the group of producers whereby
members become more aware of potential causes of the problem and work
together voluntarily to reduce violations. This is his explanation for
his observation that an increase in the sampling rate is associated with
an eventual fall in the violation rate.

The precise nature of the relationship between the probability of
detection and the violation rate, or surveillance and deterrence, has
been the topic of some of the theoretical discussion in Chapter Two and
will be examined empirically (subject, of course, to the usual data con-
straints) in Chapter Four. Suffice it to say here that FSIS personnel
operate under the assumption that such a relationship exists and struc-
ture enforcement programs accordingly. This is so in spite of what theory
tells us to expect about the estimation of expected costs of violation.

The probability of imposing costs on violators does not increase even

109

if the probability of detecting violations does, due to mitigating factors
such as inability to trace or unwillingness to prosecute.
3.2.3.4 Enforcement and Information:
The Importance of Labelling

Several different sources of residue violations exist. Some are
primarily due to accidental occurrences, and others to deliberate prac-
tices on the part of livestock producers. Among the latter category, en-
forcement of tolerances means enforcement of labelled instructions and
good manufacturing practices which the agencies maintain will prevent the
above-tolerance residues. The existence of an excessive residue is as-
sumed by the agenoy to indicate an improper or extra-label use or a vio-
lation of the withdrawal period. It is also assumed that labelling is
clear and unambiguous, when violation of tolerances is linked to a bad or
unethical (with full knowledge of the law and the consequences of one's
actions) practice on the part of the producer. Yet this assumption may
be unfounded in some cases, given the incentive structure facing drug
manufacturers with regard to labelling which is created by FDA in its
regulation of that activity. It is clearly in the interest of the drug
manufacturer, who must invest 8-10 million dollars and 8-10 years of re-
search before a drug is granted approval for use in animal production by
FDA, to cover as broad a range of uses as possible on the label.1 This
practice assures as wide a market as can be obtained; in addition over
the counter availability of most animal drugs facilitates unsupervised

use by farmers themselves.

 

1Interviews with Mr. Steven Kimbel, Animal Health Institute, Septem-
ber 13 1983, and Dr. Richard Carnevale, FDA Scientific Evaluation Div.
September 14 1983.

110

Animal producers thus face an information problem which may encourage
improper use of medications. Unlike veterinarians, they have little train-
ing on the consequences of substance use either on the metabolism of the
animal or the human who consumes meat products from that animal. They
have free access to the substances, however, and their major source of
information on potential effects is advertising by the drug manufacturer
(which concentrates on potential benefits rather than dangers) and the
label. When growth promotants containing estrogenic compounds are labelled
for use in "beef calves," what are the implications for the veal grower?
Our observation of the presence of zeranol in veal calves may be a result
of this information problem.1 When information is costly to obtain, the
provision of subsidized information can be profitable for those seeking
to promote use of certain products. This is the reason for truth in ad-
vertising legislation and labelling requirements for many products. Al-
though a detailed discussion of FDA's procedures for establishing the
contents of labels is not intended here, it is important to point out
the potential links between activities regulated by FDA and the enforce-
ment problem faced by FSIS. We can seek explanations for Violative be-
havior faced by one enforcement agency in the regulatory structure created
by another when each has an impact upon different aspects of the same

production process.

3.2.4 Producer Education Programs
When the goal is changing producer behavior which creates residues,

theory has indicated that several alternative approaches exist. Optimal

 

1Interview with Leslie Rubin, USDA, Residue Division, Nov. 23 1983.

111

enforcement theories have focused attention on the probabilities of de-
tection and being forced to pay costs or fines as the primary deterrents
available to the enforcement agency. Several qualifications to this
theory are worth noting, however.

An important underlying assumption when considering the "criminal"
behavior which government agencies are trying to prevent is the motiva-
tion which causes the individual to commit the crime. Change the motiva-
tion, and the resulting behavior will also change. Optimal theories pre-
sume that the most important motive for a firm is profit, and that viola-
tions are committed in a rational attempt to maximize achievement of that
goal. In making the compliance decision, Becker's and Stigler's "straw
people" are rational, profit (or efficiency) oriented, and perfectly
informed about the probabilities of detection and consequences of their
actions (see Chapter Two for a detailed description of these theories).
The decision maker, in other words, does not face an information problem.

When information is costly or difficult to obtain, we may observe
non-compliant behavior which is motivated not by profit, but by mere
ignorance. When dealing with issues on the forefront of scientific and
technical knowledge such as absorption and metabolization of drugs by
animals, ignorance affects agency and producer decisions alike. There-
fore, agency resources are expended on programs to provide animal pro-
ducers with information which, it is hoped, will fill in the information
gap which contributes to non-compliance.1 One such program is described

below.

 

1This point has been emphasized by Dr. Spaulding during interviews
on July 29, September 13-16, and November 23, 1983. It was also made by
Mr. Steve Kimbel of Animal Health Institute on September 15 and Mr. Howard
Frederick of American Feed Manufacturers Association of September 9 1983.

112

The Residue Avoidance Program (RAP): The goals of RAP are as fol-
lowsz'l
1. To examine livestock and poultry systems to determine the
critical points of control f0r residue avoidance.
2. To develop management recommendations to eliminate factors
contributing to residue violations.
3. To educate producers and others in the livestock and poultry
industries to avoid residues and improve the safety of the
meat supply.
To accomplish these goals, a cooperative agreement was made on February
24, 1982 between the Extension Service (ES) and FSIS to design a program
initially funded with $1 million in pass-through funds from FSIS to ES.
Additional funding in the amount of $1,450,000 has been provided by these
agencies since the initial agreement. The program was conceived follow-
ing the reduction in violation rates fer sulfa in swine from 14% to 4.4%
after an education program had begun by the Extension Service in 1978.
Most of the funding has been parceled out to Land Grant Universities
at the state level. During fiscal years 1982 and 1983, 49 projects were
funded in 33 states. Many of these projects are designed to provide in-
formation on the effects of various substances on food animals to county
extension workers. Others contribute to the development of FARAD, (Food
Animal Residue Avoidance Database), a computer based system to be used in
conjunction with the National Agricultural Library. Still others research

the economic aspects of residue problems. All, however, are designed to

 

1U.S. Department of Agriculture, Extension Service, Agricultural Pro-
grams, Livestock and Veterinary Sciences, "Residue Avoidance Program:
Summary" September 30 1983. All information in this section on RAP was
taken from this document.

113

contribute to the information available to the agency and the producer
of livestock about the consequences of use of chemical substances on the
meat supply. It is hoped that this will reduce the number of violations
due to ignorance, at a low cost to the producer, through the provision

of subsidized information.

3.3 FDA Responsibilities

The Food and Drug Administration, although it is a division of the
U.S. Department of Health and Human Services (HHS), receives its funding
from the USDA. It has responsibility for setting tolerances and approv-
ing new drugs, their uses, and labelling. This section concentrates on
its enforcement activities in cases where residues have been discovered
in meat.

The Compliance Division within the Bureau of Veterinary Medicine
(BVM) of FDA is the branch of that agency primarily responsible for en-
forcing residue standards in meat. While USDA is concerned with the
maintenance of a safe meat supply, FDA regulates the use of animal drugs.
This distinction explains many of the differences between the agencies
in terms of policy and attitude, and is the basis for the boundary between
their jurisdictions. Because the use of animal drugs is an on-farm activ-
ity, FDA may investigate on the farm in pursuance of its mandate. USDA
may not normally investigate on farms (except in cooperation for sample
collection with FDA) because its legislated mandate is over "meat"-—a
term which refers to food animals only after slaughter.

FDA's Compliance Division receives all monitoring data on violations
from FSIS regional offices when producers have been identified. All such

producers have been sent a Notice of Adverse Findings by FSIS Regional

114

office. As of this writing, however, first violators have not been sup-
plied with the information that they may be subject to prosecution under
the Food, Drug, and Cosmetic Act. At the time of my conversation with a
member of FDA compliance staff (September, 1983), however, FDA was in the
process of requesting that FSIS include this information in its first
letter to violators, for reasons which will become clear below.

Listings of violators are cross checked by computer in order to com-
pile a shorter list of those who have histories of previous violations of
residue tolerances. These second and third time violators become FDA's
primary target. Although some 1000—2000 violations are detected annually
under monitoring, a much smaller number are detected a second time. A
rough estimate made by a compliance staff member of the ratio of first to

1 FDA, within the limits

secondvioiations was approximately ten to one.
of its budget, makes an attempt to investigate each second violator.

The investigation is comprised of both letters requesting explana-
tion of the occurrence of the residue and on-farm investigations by FDA
field officers. Evidence of the manufacturing practices which may have
led to the residue is sought. Medications in or near the animal enclosures
are examined, and samples of soil or animal fecal material may be taken
for analysis. Subsequent to the investigation, the farmer is sent a reg-
ulatory letter explaining the violations and relevant laws, and the pos-
sible consequences.

About one in ten second violators will be found to go on to commit

yet a third infraction of tolerances. Having established procedures

 

1Interview with Homer Ransdell, FDA Compliance Division, September
16 1983.

115

whereby a producer by this point has been given adequate warning about
the seriousness and consequences of her offense, FDA can take legal ac-
tion to request an injunction against the persistent violator. The court
usually requires that FDA officials and the producer come to terms and
prepare a consent decree to be signed in the presence of the court. This
decree outlines the conditions under which the producer may continue to
operate, including procedures she must follow with regard to drug (or
other residue creating substance) use in order to ensure that the problem
is resolved. The agreement, once signed, has the force of an injunction
and if violated is punishable as a violation of an injunction by the
courts.1

Other legal options open to both FDA and FSIS include seizure of
product which the agency has reason to believe is tainted, product recalls,
and similar actions to prevent meat containing dangerous residues from
moving in commerce. It should be noted that these actions, along with
the seeking of injunctions against repeat violators described above, are
extraordinary and taken infrequently in only the most severe of contami-
nation incidents or against recalcitrant violators. The day to day en-
forcement structure with regard to residues which is most likely to af-
fect the average producer involves surveillance and possible condemnation
of product by FSIS, and/or an investigatory visit and a regulatory letter
from FDA. As yet, no producer has had a suit brought against her by
FDA for endangering consumers through residues in her product, although

FDA Compliance Division staff indicate that this possibility is being

 

1Interview with Ann Petersen, USDA, General Counsel's Office, Septem-
ber 15 1983.

116

considered for the near future.1 In requesting that USDA include the
additional information in its initial letters to violating producers
that they may be subject to prosecution, FDA is attempting to establish
earlier in a violator's history the fact that she has been warned and is
cognizant of the consequences of her actions. This idea originated dur-
ing the legal procedures surrounding an incident in New York state early
in 1983, when four veal growers were suspected of illegally using DES.
The agency feels that this could redress some of the legal shortcomings
of the agency's position when attempting to prosecute.

FDA enforces not only animal drug residue tolerances, but also
tolerance levels for environmental contaminants which affect the food
supply. Although tolerance levels for these substances are set by the
Environmental Protection Agency, FDA is responsible for follow-up inves-

tigation in the same way it is for animal drugs.

3.4 Conclusions

The purpose of this chapter has been to provide a description of
the standard operating procedures and decision rules which govern FSIS
implementation of enforcement programs. In so doing, certain agency goals
have been pointed out along with factors which constrain their achieve-
ment. This section discusses the relationship between the actual agency

objectives as observed by the researcher and demonstrated by its SOPs,

 

1Many complex legal questions intervene in an attempt to prosecute
on the basis of the agency's stand that residues above tolerance, in an
individual case, present a threat to human health. Tolerances include
safety factors and assumptions about aggregate exposure throughout a life-
time. Testing methods are sometimes based not on discovery of the sub-
stance itself, but a metabolite or "tag" substance which indicates that
the substance did exist in the animal. FDA staff members indicated that
these are very difficult to defend before a jury.

117

and those objectives which would be expected of Becker's “rational" en-
forcers. The next chapter will examine whether agency resource alloca-
tion shown by monitoring and other enforcement program expenditures con-
form to the economists' guidelines.

Examining agency decision making is a way of asking the question,
what do agency SOPs reveal about its objectives? This approach has been
taken in order to test the validity of the assumptions made by the econo-
mic theories about agency decision makers' objectives. In simply assuming
that what Becker proposes shoulg_be the primary goals of enfbrcers actual-
1y are their goals, an evaluation would miss important clues to agency
behavior, if the assumptions were incorrect. Therefore it has been ne-
cessary to take this empirical approach, analyzing information about the
agency's reactions to various enforcement problems to discover its impli-
cit and explicit priorities. These priorities can then be compared to
those which arise from an application of Becker's theory to the meat

safety case, as provided in the conclusion of Chapter Two.

3.4.1 The Probability of Detection

A fundamental element of "rational" enforcement is the setting of an
"optimal" probability of detection so that enforcement costs and the
value of crimes deterred are equal at the margin. What has this chapter
revealed about the probabilities of detecting meat residue violations?

Firstly, it is clear that under random monitoring, the probability
that any individual carcass will be sampled is extremely low. It is low
because the volume of meat slaughtered under inspection is so high that
to increase the proportion sampled significantly would be prohibitively

costly. A reason must exist for doubting the purity of a carcass in order

118

for this probability to be raised, such as physical evidence on the car-
cass or a history of violation on its producer's part.

If physical evidence of contamination exists, there is a good chance
that the in-plant inspector will discover the violation by initiating
sampling. If, as is more likely, no such evidence exists, it is up to
random sampling to detect a violation. Even in cases where the producer
has a history of violation, it has been illustrated that difficulties in
tracing and the abundant possibilities for evading follow-up surveillance
make it easy for the small producer to escape the sampling consequences
of violation.

The goal of monitoring is not to detect and punish individual viola-
tors or prevent all tainted meat from reaching consumers. It is to pro-
vide a statistically valid estimate of the proportion of carcasses con-
taining Violative residue levels. If every case of above-tolerance resi-
due levels in and of itself threatened human health, this approach would
probably be inadequate to protect consumers. In fact, the health danger
created by most of the residues FSIS is concerned with is a matter of
long term, dose related exposure. Therefore, a system such as that de-
scribed in this chapter which creates a very low probability of detection
performs the information function well at a relatively low cost.

However, when deterrence of violations is the goal, does such a low
probability provide the desired deterrent? Furthermore, does it provide
the information needed to impose costs on violators? The empirical issue
of whether or not violation rates are high and whether they respond to
increases in the probability of detection will be the subject of the next
chapter. As for the adequacy of the information provided, several barriers

to FSIS imposition of costs have been pointed out in this chapter. One is

119

the inability of FSIS to identify previous owners of livestock upon com-
pletion of laboratory tests. Another is the limitation of its legal man-
date. FDA's Compliance Division is responsible for prosecution and im-
position of civil or criminal penalties. Its own internal priorities and
legal problems defending tolerances seem to have limited its participa-
tion in creating the "expected costs" of violation which Becker maintains
are the main deterrent to violations.

One might be tempted to conclude, in regarding the probability of
detection and costs of violation to be the sole indicators of deterrence,
that the goal of the Residue Evaluation is not deterrence since the ex-
pected costs of tolerance violations are so low. However, this ignores
the possibility that other contributing factors to deterrence might in
fact keep the violation rate low enough to assure the safety of consumers.
The mechanism of the market and consumer confidence, the level of purely
"ethical" obedience to laws, and the use of subsidized information by
the agency to magnify perceptions of the consequences of violation are
three candidates which have been pointed out. In order to draw the con-
clusion that deterrence was not important to the agency, we would have
to observe that deterrence does not exist, i.e. that violation rates are
high, and the agency is making no effort to design new enforcement pro-
grams to bring them down. The following chapter will address these issues.
The stated SOPs of the agency as revealed here do attempt to respond to
continued high violations by initiating special surveillance and educa-

tional programs.

3.4.2 The Criteria for Surveillance
The criteria used by the Residue Division to decide when to initiate

surveillance and other special programs indicate the way in which it

120

prioritizes health and enforcement problems for allocation of staff time,
laboratory capacity, and other resources.

Criteria a through c (as listed in Table 3.3) indicate that the
Residue Division does attempt to assess the social harm of residue prob-
lems. These are the spreading potential, level of toxicity, and widespread
residue-creating practices throughout an industry. Each of these factors
contributes to the amount of residues which reach the consumer, either by
affecting a large number of animals or by affecting them with a particu-
larly health threatening substance. (Pursuing users of banned substances
performs the same function, since substances are usually not allowed for
use because they are carcinogenic or toxic.) When one or more of these
conditions arise, the SOP of the agency is to allocate the resources
necessary to address the problem, including raising the probability of
detecting violators by taking extra samples.

No systematic effort seems to be made by the Residue Division to
weight the factors in a residue incident which threaten human health.

It is not clear what would happen, for example, if FSIS were faced with

a case of chloramphenicol contamination and a DES incident, and had the
resources to resolve only one. The criteria mentioned seem to serve as
ways of deciding that FSIS will act, rather than as ways of measuring the
social harm which would be prevented (the agency's output) in terms of
agency resources required to create deterrence (its inputs). Such a cal-
culation is necessary if the agency is to make "optimal" decisions about
how to use its resources to obtain the most deterrence. However, any
effort to make this calculation would suffer from the difficulty of cross
case comparisons between "units" of output. How, in the example hypothe-

tically posed above, could endangering the lives of a very small number

121

of chloramphenicol-sensitive individuals be compared in terms of "social
harm" to the presence of a carcinogen such as DES which threatens all
consumers with a smaller risk of cancer?

In conclusion it can be said that the Residue Division is clearly
concerned about the amount of social harm created by residues, but does
not attempt to calculate its value. Rather, certain situations are per-
ceived of as "harmful enough" to merit the Residue Division's attention.
Programs are then designed with the characteristics of use of the sub-
stance in mind. In the near term, resource constraints must be observed,
but it should be noted that resources are not completely fixed. Total
allocations may change from year to year, and special funding and cooper-
ation from other branches of the USDA is occasionally available, as with
the RAP program.

Certain of the remaining criteria demonstrate other elements of
"rationality" as defined by Becker. The ability of the agency to be
effective at low cost is a primary concern of the Residue Division, as
demonstrated by criterion d (FSIS effectiveness). The availability of
a low cost testing method for antibiotics and sulfa residues has certainly
been a factor--bolstered by their prevalence throughout certain indus-
tries--in the Residue Division's continuing commitment to their eradica-
tion.

Yet, again there seems to be no effort to weigh inputs in terms of
the outputs they will create. Perhaps this is because, as the sole body
enforcing tolerances in meat, the Residue Division is obligated to deal
with all "serious enough” residue problems in some way. It cannot pick
the most "profitable" (i.e., most deterrent creating per dollar spent)

investment and exploit it to the exclusion of the more difficult

122

enforcement problems. It is legally mandated to address all meat resi-
due issues to some extent. Between this important consideration, and
the problem of weighting the "social harm" of different residue problems
mentioned above, attempting to determine "optimal" enforcement strategies
is not an SOP of the Residue Division. However, concern over the amount
of social harm deterred by diverse strategies and their relative costs--
to producers as well as to FSIS--do provide general guidelines fer re-

source allocations which are incorporated into the agency's SOPs.

CHAPTER FOUR

RESULTS, PART II: FSIS RESOURCE ALLOCATION
AND ITS EFFECTIVENESS

4.1 Introduction and Methods

The objective of this chapter is twofold. First the question is
asked: what have the characteristics of Residue Division's enforcement
program been? This is the topic of section 4.2. Secondly, it is neces-
sary to assess the effectiveness of these programs by asking whether
residue enforcement programs succeed in reducing the number of violations
of tolerances in meat. This is done in section 4.3.

In order to discuss what the enforcement programs for residue toler-
ances in meat have looked like over the period since 1973, when sampling
for these residues began on a large scale, data on the number of samples
taken for different substances and substance groups will be examined.
This type of information is used because the number of samples taken re-
presents the allocation of agency resources between substances and species.
The variations in sample sizes over a ten year period should be explicable
in terms of the decision criteria presented in Chapter Three, which ac-
cording to agency officials reflect the diversity of public health prob-
lems which residues create. Primarily, the assessment of the sampling
data will take the form of an examination of the trends in resource allo-
cation (represented by the size of the sample) as they have varied between

substances and industries over time.

123

124

Another subject which will be examined is the implications of the
size of the samples set by the Residue Division for the precision of its
estimate of the true proportion of the population (the meat supply) which
is contaminated by residues. This will be done by examining the claims
of the Mathematics and Statistics Division about what is accomplished by
the random sampling programs in the light of statistical theory on sample
size.

The objective of section 4.3 is to determine the effectiveness of
agency allocation of resources. To do so, the term "effectiveness" and
what comprises "allocation of resources" must be defined. The legislated
objective of any enforcement agency is to enforce the law by deterring
activity on the part of the regulated group which violates that law.

In this case, the "law" is residue tolerances in meat, and the activity
which the FSIS Residue Division tries to prevent is producer practices
which lead to violations of tolerance levels for these residues in their
livestock. Therefore, the change in the violation rate associated with
changes in agency resource allocation will be used as a measurement of
"effectiveness."

Resource allocation by the Residue Division takes several forms.
Analysis of samples is costly and the resources available to do so are
fixed within budgetary periods. Changing the number of samples which
will be collected for an individual substance/species pair represents
a primary way in which the level of resources allocated to that particu-
lar residue problem can be changed. Therefore, changes in the sampling
rate will be used as an indicator of "resource allocation" by the agency.

Furthermore, the theory of deterrence has indicated that changes in

the probability of detection in violation is one of the two elements of

125

the potential violator's calculus of the expected costs of detection.

As such, it deserves attention as an explainer of the level of violative
behavior observed. The "probability of detection" referred to by the
“rational" enforcement theories is partially determined, in the meat
residue case, by the probability of being sampled. Mitigating factors
also exist which weaken the relationship between the discovery of a vio-
lation through sampling and the agency's ability to impose punishment on
the violator, which have been touched upon previously and will be further

examined here.

In the models discussed in section 4.3, the relationship between
sampling (probability of detection) and violations is examined. Yet,
Becker has directed our attention to two aspects of expected costs of
violating laws: the probability that such costs will be imposed, and
the costs themselves. It has been pointed out that the violation rate
does not equal the number of violators who are informed of their viola-
tions, but only the number of samples found to be over tolerance by FSIS
laboratories. The number of violators caught (traced) must be some
fraction of the total number detected by the laboratory. In using the
latter as a proxy for the former, the assumption is made that this frac-
tion is constant over time and between states.

A further qualification is that actual costs of violation have been
left out. This can be justified on two grounds. Firstly, it has been
demonstrated that these tend to be zero, inasmuch as they are imposed by
the agencies. Secondly, no differences have been observed between time
periods or states in the application of such costs. That is, costs of
violation are assumed to be constant and zero by the models used to test

the relationship between probability of detection and violative behavior.

126

If costs in fact do differ between states, we may find significant dif-
ferences in the reaction of violators to the sampling rate, when each
state is run as a separate model. The same may be true if the proportion
of violators traced differs markedly between states.

A linear regression is used to test the relationship between sampling
rates and violation rates. Since the type of monitoring/surveillance
program which raises both the probability of detection and allocation of
resources has been used only fer sulfa in swine, that is the case which
is studied here.

This chapter will include an assessment of the extent to which pro-
grams match the stated priorities of the Residue Division presented in
Chapter Three, summarized in Table 3.3. An estimation of the effect of
changes in the sampling rate on violations and a discussion of the limi-

tations of the models provides the conclusion of section 4.3.

4.2 Enforcement Program Resource Allocation

In this section, programs representing resource allocation between
substance groups are described. For each substance group, programs are
divided into two categories. First, the random monitoring allocations
for a particular substance group are examined. Secondly, all other pro-
grams for that substance group are discussed, including special sampling
(over and above the national random monitoring) such as spotlighting,
non-random surveillance, on—site testing, certification, etc. The sub-
stance groups covered include hormones, sulfa, antibiotics, chlorinated
hydrocarbons, organophosphates, and trace elements. Data for each group
is aggregated over seventeen species and subspecies (bulls, cows, heifers,

steers, calves, horses, sheep, goats, swine, young chickens, mature

127

chickens, young turkeys, fryer-roaster turkeys, mature turkeys, ducks,

geese, and rabbits).

4.2.1 The Monitoring Data

Table 4.1 on the following pages was constructed from a ten year
summary of the annual random sample sizes for each substance, by species.
The number of samples taken for all species for each substance group has
been aggregated fer each year. A total number of random samples was com-
puted by adding together all of the samples taken for each substance
group. The proportion of the total random monitoring samples devoted to
each substance group was then derived by dividing the substance group
sample size by the total sample size. In some cases such as arsenic,
chloramphenicol, hormones and general drugs, separatetotals have been
provided for substances within a group. This is done where the sampling
program differs significantly between substances within a group.

The percent of total monitoring sampling devoted to selected groups
has been graphed in Figure 4.1 to show changing allocational priorities

over the period 1973-1983.]

Inthis section, the trends in the major sub-
stance groups will be explained according to the criteria listed in the
previous chapter as contributing to the gravity of a residue "problem."
Also provided in Table 4.1 are two additional pieces of information
which are necessary to provide an understanding of changes in monitoring

program has been computed, using 1973 as a base year. It demonstrates

 

1Please note that this does not represent total allocations to each
substance, as special programs, "spotlighting," on site testing, sur-
veillance, and method development are not included. Such programs are
classified here as "extra-monitoring" enforcement programs, and discussed
separately, for each substance group.

128

 

 

 

 

8.. 8.. 8.. 8.~ 8.8 8.. 8.8 8.8 8.. 8.8 .8. 0860 :0.»8.0.>

88 88 8.. .8. 888 .8 88. 88. 88. ..~ 880.88.0.>

8n n... 8.8. 8.8. 8.88 8... 8.8. 8.88 8.88 8.88 .880» 80 8 84

8888 8888 88.8 «8.8 8888 «.88 8888 8.88 8888 8.88 8.88. 8.4

8. m. 11 ii 11 in 11 ii 11 11 .8. 0880 :0.ue.o.>

8 8 11 11 11 ii 11 8 8 ii n:0.uo.0.>

8.8. 8.8 11 ii 11 11 ii 8.. 8.. 8 .880» 80 8 84

.888 «88 11 ii 11 11 ii 88. 888 11 .88.. .ou.z88828mo.=u

8.. ... 8.. 8.8 8.8 8.. 8.8 8.8 8.8 8.8 .8. 0.80 80.88.0.>

no no 8.. .8. 88.. N8 88. 88. cu. ..~ 8.3.38.0...

....~ 8.8a 8.3 8.8~ 8.8.. 8.." 8.8. 8.8. 8......” 8.3 .880» no a 84
.88. 88.88.68o.

8888 8888 88.8 88.8 8888 «.88 8888 8.88 .888 8.88 .8.88~. .88.80.8.ez<

8. 8. m. 8. 8. 8.. 8.. 8.8 8.. 8.. .8. 0.80 :0.un.o.>

.8 8. 88 8. 8. 8. m8 .8 88 88 880.8o.o.>

8... 8.88 N... 8..~ 8... 8... 8.88 N... 8... 8.8. .8808 80 8 84

8888 88mm 8.88 8888 8888 8.88 8888 .888 8888 .888 .n.88.. one

«88. .88. 888. 888. 888. 888. 888. 888. 888. 888.

.mmapimuo. .888. .8 8888888888 8 88 888
88888 mucmumasm >8 mmN.m 8.858m 88.888.882 888.888 ..8 m.m<.

129

 

 

 

 

 

.. u- u- u- u- 88 u- u- 888. 88. 88888.88888
.88 888 88 888 8.8 u- u- u- u- u- 88:
88. 88 888 888. 888. 888. 888. 8888 8888 8888 .888. 888
.8.888. 8820288:
..8 8.8 8.8 8.8 8.8 8.8 8.8. 8.8. 8... 8..8 .8808 no 8 88
88.. 888. 88.. 888 8888 888 8888 8888 8888 8888 8.888 888
8. 8. 8. 8. 8 8 8. 8. ..8 8.8 .8. 8888 80.88.o8>
8 8 8 8 8 8 8 8. .8 8.. 888.88.o.>
8.8 8.8 8.8 8.8 8.. 8.8 8.8. 8... 8.8. 8..8 .8888 88 8 88
888 888. 888. 888 888 888 .88. .888 888. 8888 ..88. 8828888
8.. u- 8. u- 8.8 8. 8.8 8.8 8.8 8. .8808 88 8 88
..88 88888.888.
888 In 88. u: 888 88 888. 88.. 8.8 88 8.8.888. 88282888 88888
8 8 8 8 8 8 8 8 8 8 880.88.o.>.
8.. 8.8 8. .. ... u- 8. -u 8.8 8.8 .8808 88 8 88
888 888 .8. u- 888 u- 88 8 888 888. .8.888. 8888888888uoz8888
888. .88. 888. 888. 888. 888. 888. 888. 8888 888.
..8888. ..8 8.888

130

uu nn nu uu Nva oaa can .nam a nu oaouooacuuma

 

 

nu nn un un mu can nu un un nu oaOvaauvaao
uu un nu ea an uu v nu uu uu vacuaeu>ma
uu nu nu nu ma ham aqa uu uu uu uaONmucmnmasa
uu uu nu nu mnw aha aan mmc uu uu acacmcmnom
no nma cm can «am oaa mum ago mum m- xocmnumu
«me an» mmw nn mam «cm can oNa ama uu camcmcoz
nu nu un un nn un om mmm nn un oumcaavoown
uu nu nu oma oa moa nu «ma uu un aocamoao
.a.oom. onusmo aammzuo
n.~ a.m m.m ~.m m.” c.~a ~.q o.~ a.c m. .a. «you coaumaoa>
«a acm «ma mac she omma mma ma om o uncaumaoa>
~.a~ m.m~ a.m~ p.5m a.mn o.m¢ n.o~ “.ma m.n ~.m auuou no » m4
anov naaw amcm mowm chap hmvoa mama pc~n hmqa mmm .m.oom. «papa
o o o o o. n.a n. a.a m. a. .a. ouau coaumaoa>
o o o o a a~ ma aw a pa mucoaumaoa>
n.~ m.a m.~ ¢.m v.a 5.5 ~.oa a.ma m.ma >.a~ auuou mo » m<
mac mom com mama mmpa mama amma owpu amen anon u.com and

 

~o¢a Mama coma Grad chad shad chad mama «baa nhmd

A.ucoov _.¢ m4m<h

131

.o who: mumnuo aa<

a~.n umpaa macsauacnuma

«a. ”mnma :amnmco:

aw. ”mama «a. "nhma «a. “mama an. «mama av. “chaa xoumnumo
ac.a «mnaa am.a “mama onavmcwnom

"cm>uonno mums uwunu coauuaoa> mcasoaaow 03H .uom vmumou can voamemm maucwv

ncmmmcca ma muommuuo was» :a mama sosu mocau .vmusmeoo ma wumu coauMa0a> mammmummw oz
.wmocmumnam
uwnuo Ham cm>ummno mumz occauMa0a> o: macaw .aaco mun ham vmusaeoo ma mama ceauwa0a>m

.mca

nuouacoe amauom can» Hmnumu owccmam unmmmumou .mwusmau mama mean an aam3 mm. amasmau mama
.m>auavvm ovum maauoeoum :u3oum m no can: ma noan3 .oacmmuc ma coauamoxm mamzam «sea
.mcauusuoo maamusumc was noan3 .mmocmumnsm mums» new umm mun umocmuoaou oz ”mamuos momma
.cmxmu mmamamm amuou can

nmamemm ca on a: mo mama ca uuaxm.nmao:ummuonao .aooacmsmEMHOanu wvzaona uo: on mamuoa
.m.m ousmam o» ummou .msoum nomm saw

mmozmumnsm mo umaa Mom .GOama>an unaccmam can acaumsam>m wscamom ”mama..<amn "mousowa

w

mNNa «ZOHH€H>HD nm¢oz<9m

oo~.ou "usam mam=<m zcmz

 

moa maa maa ona maa maa voa vaa Boa ooa .ooa u Mbmav
mwamz<m UZHmOBHzoz
A4809 ho xNDZH

 

mnmma mmmam mmoca hmhau momow mmmcu hmmma aath omvma wbaoa~ mmamz<m Dszoaazoz Adaoa

 

m.N q.v m.a m.N m.m m.w n.~a o.Na a.m m.a amuo» No v 04
new mmm man «on noaa amma mmmu mmvN ham mNN u.ocm and

 

mama amma coma mhma ohma hhma whma mhma vhma nhma

A.pcouv a.w m4m<H

132

that although some changes in size of the monitoring program occurred,
the variation around the mean of 20,200 samples was not large until 1983.
(The standard deviation is approximately 6% of the mean for the period
1973-l983.) Therefore, a certain proportion in one year would indicate
approximately the same sample size for a particular substance as it would
in another year. This makes interpretation of Figure 4.1 more straight-
fbrward than if total sample size varied dramatically.

Secondly, numbers of violations (above tolerance residue levels)
discovered in the chemical analysis of the meat samples are reported in
Table 4.l. Since the absolute number of violations is not comparable
between substance groups when the sample sizes differ, "violation rates"
are also reported. These are simply the percentage of samples analyzed
which were found violative. This information is important because it
provides an indication of the amount of exposure the meat-consuming
public has had to residues of various kinds.1 As such, it is one of the
factors to which the Residue Division says it responds in initiating
stepped-up sampling.

A final point to be made before explaining and interpreting these
trends is that the level of aggregation of Table 4.l prevents certain
interpretations. Since differences in sampling patterns between species
are not shown, important trends will be missed. For example, the sulfa
problem in swine is more pronounced than indicated by Table 4.1 and
Figures 4.1-4.2, which are aggregated over all species. Lower violation

rates for this substance among some poultry species pull the average down.

 

1Qualifications about the statistical validity of this statement
are discussed in section 4.2.3 of this chapter.

133

maaoao mocmamaam go moccaamaam umaumamm com uwammh mmaasmm meagoaaco: aaaoa mo ucmugma

a.¢ wasmad

Noma omma whma mhma whoa

 

 

now

     

-“-'-'-"-“

  

n _u¢
822:8: nuu ow $323
2.8 lul octoficoz
weapoummhzc .uuun :58.
93.5 In D6 23ng

 

 

om

134

Furthermore, different trends among species represent problems in differ-
ent industries which the Residue Division must respond to appropriately.

These more specific trends will be pointed out where relevant, and addi-

tional data supplied. Table 3.l in Chapter Three points out some of

these inter-species differences and will be referred to where relevant.

4.2.2 Trends Among Substances and Substance Groups

A glance at Figure 4.l illustrates that the proportion of monitoring
samples devoted to various classes or groups of substances has varied
significantly over time. The stated criteria of the Residue Division for
changing sampling intensity include indications that misuse of a substance
is industry wide, the toxicity and legal status of substance use, its
spreading potential, the ability of the agency to effectively control its
use, consumer concern, and producer costs. Research into the historical
conditions leading to the observable trends in agency resource allocation,
detailed below, indicates that these criteria have guided program imple-

mentation. Several other important factors, however, have also emerged.

4.2.2.l Hormone Monitoring

Perhaps the easiest to explain is the steady, downward trend in sam-
pling for hormones. This is done primarily to the decline in the number
of samples taken for Diethystylbestrol (DES). Two factors combined to
reduce the agency's concern with this growth promoting hormone. One of
these is the downward trend in the violation rate since the initiation
of sampling in 1973 (see Figure 4.2). A decline in the violation rate
indicates to FSIS officials that improper use is declining among producers,

thus reducing the need to use sampling as an enforcement tool. (Sampling

135

masocu mucmamaam umaomamm go» mwamm coaamaoa>

~.¢ mazaau

 

 

1"
‘ l ’
'Il ‘ ‘ ' -."l""-_ O’II’ \\\|"|'l
III ' \I
I ------| ' l- \\ III, \\
I- ’ I, ‘
I, I.
l I \\\
’I ‘ cl “-
ll \ II I‘-‘ fit!
I s I s
’l’ \ ” ~ ’I
I, \\ I, ‘ ’
II 5 I s I
I I s
I Q II \ II
II N I s I
I § I \ ’l
I \ ll \ .
II \ I s
I \ II ~
I s I s
'\ C

wmzozmo: nnn
uzo lul
mouhoamahzc uuuuu
cmqam.lll

 

 

ca

Na Aacmucmav

wamm
8.529;

Va

136

is nonetheless still required to perform the information function--that
is, to determine what violation rates are.)

The second important factor, however, which explains the decline in
violations since l979 and the resulting decline in sampling is an exoge-
nously created decline in use of DES by livestock producers. This is
an example of factors outside the control of the FSIS which may affect
the violation rate which it uses as a measure of its success or failure.
In this case, the FDA, after several years of public controversy over
the carcinogenicity of DES when used in producing human food, finally
withdrew approval for use in food animals. The fear of the industry that
this would eventually occur may have provoked the switch to another hor-
mone, Melangestrol Acetate (MGA), in some species. In any case, no DES
violations have been observed since l979. Monitoring is still done in
order to assure that illegal usage is not occurring.

4.2.2.2 Extra-Monitoring Enforcement
Programs for Hormones

An incident described in Chapter Three regarding the alleged use of
DES by veal producers in New York provoked coordinated action by the
FDA, the New York State Department of Agriculture, and FSIS requiring
surveillance sampling by the Residue Division. As mentioned in that
chapter, such allocation of resources indicates FSIS concern over illegal

substance uses which indicate clear contravention of the law.

4.2.2.3 Sulfa Monitoring
The trend in the proportion of monitoring samples devoted to sulfa
drugs, and the observed violation rate for this drug (as portrayed in

Figures 4.l and 4.2) are also similar. Mounting violations from l975

137

through 1977 caused the agency to devote nearly 50% of all monitoring
sampling to sulfa. This proportion has fallen steadily since that year
for several reasons.

One reason for the decline in sampling was the decline in the viola-
tion rate for sulfa overall, which appears dramatic in Figure 4.2. It is
important to note, however, that the overall trend misses important dif-
ferences between livestock species. In swine, for example, the decline
in the sulfa violation rate was not as marked. From a high of l3.2% in
l977, it fell to 9.7% in l978 and 6.5% in l979. Subsequent rates have

1 In

hovered around 5%--consistently higher than most other residues.
veal calves violation rates in excess of 2% have been observed nearly
every year. The problem is particularly acute in bob (very young) veal.
Although a decline has been observed, sulfa residues have remained fre-
quent enough to be perceived of as a problem by FSIS.

A second reason for the decline in monitoring for sulfa since l977
is that new programs have been designed which attempt to address the
shortcomings of monitoring as an enforcement tool. Data on the precise
costs of these programs to the agency is unavailable, but it is clear
that they represent a significant expenditure over and above the costs
of sampling. A description of these follows in the next section.

A final point to be made about the marked increase in monitoring

for sulfa between l976 and 1977 is that it has been accompanied by a de-

cline in the proportions allocated to other substances. since the total

 

1FSIS data--l0 year summary, Monitoring Program.

1 Izlnterview with Dr. Rellosa, USDA, Northeast regional office, March
5 984.

138

amount of monitoring samples has remained approximately constant. There-
fore while allocation to sulfa rose in l977, the number of samples taken
for all other substance groups except antibiotics fell in that year.
It illustrates an extremely important factor which was not explicitly
mentioned by agency officials in their explanations of how resources are
allocated between different residue problems. That is, when the labora-
tory capacity is fixed, sampling more for one substance means sampling
less for others. Thus the sample size for a particular substance is af-
fected not only by its own violation rates, but the violations observed
for other substances.
4.2.2.4 Extra-Monitoring Enforcement
Programs for Sulfa

In terms of the amount of the Residue Division's time and resources
devoted to the development and use of enforcement tools over and above
monitoring, sulfa has received the most attention. The following is a
description of these programs and an explanation of why each was developed.

a. Producer education: In l978, FSIS cooperated with the Coopera-
tive Extension Service to promote better dissemination of infbrmation
about sulfa use in swine.1 The decision to use this approach resulted
from a determination that despite large increases in sampling of swine,
violation rates remained at unacceptably high levels. The assumption was
that many livestock producers were unaware of withdrawal times and the
consequences of subtherapeutic feed applications. A change from use of
powdered to granulated form reduced the problem of electrostatic clinging

in feeding equipment which had contaminated withdrawal feeds (Kramer, 1982).

 

1Department of Agriculture, Extension Service: "Residue Avoidance
Program Summary," September 30 1983.

139

b. Surveillance: Nhile producer education programs continue, the
Residue Division has also persisted in its efforts to catch and inform
violators. In young calves, visible evidence in the form of "green gut"
has indicated the presence of undigested sulfa during postmortem inspec-
tion. Inspectors have been instructed to initiate sampling when such
indications are observed. In the past, this has meant sending samples
in to laboratories while holding the carcass at the slaughter plant. An
average of 69.l% of the nearly 2000 such samples over the period October
l982 through January l984 were confirmed as violative.1 Using the esti-
mate of $50-75 per sample of laboratory cost given by Dr. Spaulding, an
estimated $l00,000—$l50,000 were spend on sulfa surveillance in calves
alone.

Stepped-up monitoring or "spotlighting" has been initiated in several
states for swine sulfa residues as another form of surveillance. Some
l363 of these samples were collected in 1980, l844 in 198l, 28l in 1982
and 397 in l983, for a total outlay of $l94,250-$291,375 (using the cost
estimate range mentioned above).2 It should be noted, however, that all
of this extra sampling is not translated directly into informed violating

producers, since the animal identification problem persists. For example,

 

1FSIS data: Surveillance Samples for Sulfa Residues in Calves for
FY 83 and FY 84. Samples taken under the surveillance phase are not in-
cluded in Table 4.1, which includes only national random monitoring
sampling.

2The Residue Division has admitted to some confusion over the moni-
toring data for sulfa in swine for the period 1977-1983. It is not al-
together clear whether these spotlighting samples were mixed in with the
regular monitoring figures for these years, or should be considered as
expenditures over and above monitoring.

140

only 57% of sulfa/calf violations in the Northeast region have been
traceable to their producers between 1980-1983.1

c. Test development: A new procedure has been developed for swab
testing on premisis (STOP) which can detect the presence of sulfa in the
organs of recently slaughtered animals. Development of the testing
method for sulfa STOP (SST) was costly to FSIS yet the potential returns
to this type of testing are great because of its low cost in use. As
demonstrated in Chapter Three, use of on-site testing methods allows FSIS
to alleviate its laboratory capacity constraint. Another benefit of the
quick test is that it does not impose holding costs on the slaughterer.
If used systematically, SST creates an incentive for the slaughterer to
assist in the identification of animal producers.

d. Certification: A new proposal made by the Residue Division re—
presents a direct attempt to address the tracing/producer identification
issue which has mitigated the success of agency enforcement programs
which attempt to increase the probability that a violator of tolerances
will suffer adverse consequences. The program would initiate a certifi-
cation procedure for veal calves, whereby owners of a calf prior to
slaughter would be identified and required to record all uses of drugs
on that animal.

FSIS does not have the power to legally require such a certificate.
but has attempted to create incentives for its use through the market

mechanism. The Residue Division has stated that all calves coming from

 

1Pennsylvania State University, College of Agriculture, Cooperative
Extension Service, "RAP: Sulfa Residues in Calves." Unpublished, no
ate.

141

lots which are not certified will be tested using SST at the slaughter
point. Since the probability that a violative animal will be caught
and condemned is great under this system, the incentive for the slaughterer
is to purchase certified animals (which will be tested only randomly as
under the old system).1

In summary, although all of the costs to the agency have not been
estimated here, it has been demonstrated that FSIS has devoted a signifi-
cant amount of attention to control of this sUbstance. Since the health
hazards of sulfa are created by long term, dose related exposure, the
objective of the agency has been to control the level of exposure (as
measured by the violation rate) rather than to react strongly to single
incidents. While FSIS attitudes toward antibiotic residues (with the
exception of chloramphenicol) are similar, programs to control chlorinated
hydrocarbons (pesticides and industrial chemicals), which are more toxic

in small doses, differ significantly in objective and design.

4.2.2.5 Antibiotic Monitoring
As with sulfa, the overall level of antibiotic residues present in
the meat supply is of more concern than specific incidents. With the
exception of chloramphenicol, the health risk to the human population
is due to long term, dose related exposure and the possible creation of
resistant bacteria due to commonly followed poor management practices.
The high overall level of sampling for all antibiotics including

chloramphenicol shown in Figure 4.l does not seem explicable in terms of

 

1Information on this program was gathered from a telephone conversa-
tion with Dr. Rellosa, Regional Office on March 15 l984. The program
was scheduled to begin in April 1984, but has been delayed.

142

the trend in overall violation rates, which have declined every year
since l978. However, when separate species trends are examined and
chloramphenicol sampling removed, the sampling and violation patterns
are no longer contradictory.

Figure 4.3 illustrates these trends for two of the species which
account for the majority of antibiotic sampling. Both violation rates
and sample sizes move in a similar fashion in most years. Violation
rates for calves previous to l980 were erratic but consistently high.
Antibiotics are considered a "major problem" in the fancy veal industry
and as a result llOO-l400 samples have been taken each year since 1976.1
While the number of samples taken from cows was less than that fbr calves,
the violation rate was also lower.

Chloramphenicol is the one antibiotic which is not tested for in
the same way as the rest of the 200 class. For the majority of antibio-
tics, one screening test is used on each sample to indicate the presence
of any microbial inhibitor. If the result is positive, further work
must be done to determine which of the group is present. However, this
process is not used for chloramphenicol. Separate samples are collected,
which have been added to those for the remainder of the antibiotic class
in Figure 4.l. A small number of samples was taken for this substance
in 1974 and l975. Larger numbers of samples were collected in 198l and
1982, which partially account for the observable peak in antibiotic sam-
pling during 1982.

Large numbers of samples are taken for chloramphenicol for two rea-

sons. First, it is not labelled for use in food animals in the U.S.

 

Ibid.

143

 

3600
— anves

:uftntner """ COWS

Samples 3000
2500
2000
1600
1000

600

 

 

 

 

1974 1978 1979 1990 1902

 

Violation 14 a
Rate
(Percent)

12-

10-

 

 

 

 

1974 1978 1979 1990 1992

Figure 4.3

Antibiotics in Cows and Calves: Sample Size and
Violation Rate (Excluding Chloramphenicol)

144

(although this use is permitted in Canada). Secondly, certain individuals
are extremely sensitive to any dose of this substance, developing a con-
dition called aplastic anemia which can be fatal. Although it is illegal,
FSIS and FDA officials are aware of its widespread use among veterinar-

ians.1

4.2.2.6 Other Enforcement Programs for Antibiotics

As for sulfa, a swab test on premises has been developed for anti-
biotics. Furthermore, tests have been developed whereby live animals on
the farm can be tested for the presence of antibiotics. This test is
available to farmers.

The use of STOP tests represents an expansion of FSIS capacity to
test for antibiotic residues at a low cost (compared to the cost which
would be incurred if all testing had to be performed in the lab).2 In
addition, it can be done overnight while the carcass is held, so that if
the test result is positive it can be condemned and prevented from being
marketed. As this causes an immediate loss to the slaughterer (unlike
normal monitoring) it creates an incentive fbr slaughterers to keep better
records of purchases, purchase only from producers they know, or purchase
conditionally (hold payment until residue status is determined). These
arrangements shift risk of loss from the slaughterer to the producer
(who creates the residues), via the market. Better vertical integration
in meat markets is viewed by the Residue Division as a positive step which

will ease its job of preventing producer practices which create residues.

 

Illnterview with Homer Ransdell, FDA Compliance Division, September
l6 983.

2An estimate of these savings is provided in Chapter Three.

145

The agency's use of STOP and the certification programs for sulfa and
antibiotics represent an effort to encourage the type of integration
between producer and slaughterer in the beef industry which has already

occurred in the poultry industry.1

4.2.2.7 Chlorinated Hydrocarbon (CHC) Monitoring

This group of substances includes pesticides such as Dieldrin as
well as industrial chemicals such as polychlorinated biphenyls (PCBs).
The chemical structure of this group of substances makes it extremely
persistent in the fat of affected animals. As for antibiotics, one test
is used to screen samples fbr the presence of any CHC compound.

Detection of CHC compounds at both positive (any detectable amount)
and violative (amounts detected above established tolerances) levels has
declined steadily since the early 1970's. Figure 4.l demonstrated the
aggregated trend over all species for violations. Figure 4.4, a graph
of Table 4.2, shows that the percentage of positive results for selected
CHC's detected in cows has declined drastically.

What accounts for this decline? Factors outside of agency control
have assisted it in its efforts to reduce residues in some cases, just
as they have interfered in others. Most CHC pesticides as well as indus-
trial chemicals such as PCBs were banned by EPA in the mid l970's. The
slow pace of the decline in observed levels is due to the tenacity of
these substances in the environment.

The causes of violative levels of CHC's, therefore, are not usually

related to production practices, but to accidents. Such accidents may

 

1Interview with Dr. Spaulding, U.S.D.A., R.E.P.D., July 27 l983.

l46

Saw—55m ..8» ca .33 octoaacoz

mama soc» amasnsoua

 

 

m.ma m.a~ ¢.m~ _.~N a.am ¢.mo o.~o m.aa _.mm m.mm ago
a.oa a.aa a.m m.oa m.mm a.am m._o m._o m.~o m.om 01m
m.¢ m.a m.o a.a m.ma a.m~ m.m~ o.a~ m.mm a.- Loacuaagaz
m.m a.m a.m m.oa a.a~ m.ma a.am o._w m.om m.mm cacaaaao
Nam. amma omma maaa mama aama mama mama aama mama nameoaaaaoa>

a

aozaua>aaamoa a

3:53. 3.530; vmzogm 5.23 9019 umuumamm Low 03mm; 3353 we acousma Né WEE

147

 

 

 

 

 

 

1400
Number of
Samples
for G“: 1200 -n
Group

1000 a

800 -i

600 .u

400 a

200 a n,

1974 1978 nova 1980 1982

Percent 100 —
Positive. """ 8110
Selected -" gllJELDRIN
Cuts in Cows "" HEPTRCHLOR

 

 

 

 

I n n n n
1074 1976 1979 1900 1982

Figure'4.4

Percent Positive vs. Sample Size, CHC in Cows

148

create large scale, quickly spreading contamination of the food supply
with poisonous chemicals. This type of incident is typified by the PCB
contamination of poultry feed which occurred in Billings, Montana in
l979. The potential for contamination spread over 20 states and two
foreign countries via the feed itself and livestock which received the
feed, before the incident was cleaned up.1
Other than through accidents, the primary culprit is the longevity
of CHC's; for example, PCB's used on silos before the manufacture and
use of the substance were restricted in l979 are still the source of si-
lage contamination and lawsuits today.2 Substances used to finish the
interiors of coolers where meat is hung are another potential source.3
The trend in sampling does not correspond closely to the trend in
discovered violations. That is, although violations have declined, the
percentage of monitoring devoted to CHC's has remained high. Several
factors contribute to this phenomenon. Extra "special monitoring" sam-
ples were taken for PCB's during 1979 and 1980 in FSIS's effort to deter-
mine the Western U.S. incident's scope. (Since all CHC's are tested for
in a single test, increased sampling for one substance leads to increased
testing for all of those in the group.) A further contribution to the
high proportion of monitoring resources going to CHC's has been the re-

quirements of method development. Extra samples were required in l983

for use in testing new methods of detecting these substances.

 

1U.S., Department of Agriculture, Food Safety and Quality Service,
Report on the PCB in the Western U.S. l980. p. 43.

2National Association of Federal Veterinarians, "Proceedings," l983.

2 3Interview with Dr. W. Leese, USDA, Compliance Division, November
5, l983.

149

Another possible reason for taking so many CHC samples is that the
nature of the health risk involved is quite different from that posed by
animal drugs. A single exposure at high enough levels to a CHC may cause
a human health hazard, since they are highly toxic and persist in both
animal and human tissue. Tests of PCB exposure on laboratory animals
indicate the possibility of reproductive disorders, tumors. gastric dis-
orders, and skin lesions. Studies on human exposure have shown skin dis-
orders, digestive disturbances, jaundice, respiratory tract irritations
and severe headaches.1

A single incident of violative CHC's, therefore, is considered much
more dangerous than a single antibiotic or sulfa violation. However, it
should be clear by now that the random monitoring system does not guaran-
tee detection of single incidents. Its statistical claim is only 95%
confidence of detecting a 1% violation rate or higher. Many single vio-
lations may have gone unnoticed under the present sampling system. Un-
less a single incident affects a very large proportion of the livestock
in a region, monitoring does not have a good chance of picking it up due
to its low intensity.

The toxicity of CHC's makes a somewhat different approach to detec-
tion necessary. One objective of monitoring is to give a valid picture
of the level of residues present in the meat supply. Yet, monitoring
has also been used frequently as an enforcement tool, as demonstrated
by the large variations over time in the sample sizes for some substance

groups. When danger to health is clearly a function of the percentage of

 

1National Association of Federal Veterinarians, "Proceedings," 1983,
"Avoiding Residues in Food Animals From Pesticides and Industrial Contami-
nants" by William Buck et al. p. 4.

 

 

150

meat consumed containing a particular residue rather than of single in-
cidents of the residue occurring in meat, a system which satisfies the
information objective may be adequate as a first step toward enforcement.
However, merely giving FSIS an idea of the violation rate would not seem
adequate to reduce or prevent the health risks created by CHC incidents.
This realization may explain the continuing high proportion of monitoring
devoted to this group of substances in spite of the decline in violations.
Another problem which differentiates this group from the animal
drug groups is that it is not administered nor intentionally related to
production practices. Accidents are more difficult to predict than in-
tentional overdoses because their causes are more varied. Therefore
FSIS's approach has been to improve its capacity to react to CHC viola-

tions rather than attempting to change behavior which creates them.

4.2.2.8 Extra-Monitoring Enforcement Program for CHC's

The PCB incident described in the previous section gave rise to the
Contamination Response System (CRS) developed by FSIS to improve coordi-
nation and systematize responses to incidents involving toxic substances.
Initiation conditions and details of implementation have been discussed
in Chapter Three. The key word is "response"--CRS reacts to information
from monitoring.

Such responses do include special sampling. Over 2,500 samples were
taken after the western U.S. incident, for extra monitoring. producer
specific surveillance, and certification purposes.1 Several thousand

samples were taken and analyzed in a contamination incident involving

 

1U.S. Department of Agriculture, Food Safety and Quality Service,
Report on the PCB Incident. pp. 53-56.

 

151

Dieldrin in feed used on a hog and cattle production operation in Iowa

in February 1980. All of this extra sample taking is necessary to deter-
mine the extent of the spread, via sales of contaminated feed and residue
containing animals to farms farther and farther away. The later in the
food producing and processing chain the incident is discovered, the more
sampling must be done to determine the source. The wider the spread, the
greater the market-imposed economic damage to animal producers who have
unwittingly purchased the contaminated feed. The investigation of the
Iowa incident, for example, disrupted the marketing of feeder cattle in
Oklahoma which had been given the feed. It was subsequently discovered

that the 3500 Oklahoma animals did not even contain violative residues.1

4.2.2.9 Organophosphate Pesticide (OP) Monitoring

This group of pesticides presents yet another set of characteristics
which change the nature of the enforcement problem. These substances
are more acutely toxic to animals, yet dissipate rapidly from the tissue--
usually within two weeks of exposure. When animals are contaminated
with these substances, they become ill and exhibit symptoms such as pro-
fuse salivation, stomach pains, diarrhea, twitching of facial muscles,
a stiff gait, weakness and paralysis of skeletal muscles.2

Because no violations have ever been detected, monitoring for the

0P group has been low and erratic. In some years, no OP samples have

been taken at all. Sampling continues occasionally in order to satisfy

 

1National Association of Federal Veterinarians, "Proceedings," 1983,
"Market Quarantine of Beef Cattle and Swine in Iowa." p. 2.

2Ibid., "Avoiding Residues in Food Animals from Pesticides and
Industrial Contaminants," by William Buck. p. 4-5.

152

the requirement that all potential contaminants are tested for by FSIS.1

However, it is clearly not a priority because violations have never been

observed.

4.2.2.l0 Trace Element Monitoring

With the exception of arsenic, no tolerances have been set for trace
elements because they are naturally occurring substances such as copper,
iron, and zinc. FSIS collects and analyses samples for these substances
in order to study the relationships between the elements and ratios in
which they might appear with other substances. Samples are therefore not
an urgent matter and are done on a space available basis, or to make the
most efficient use of laboratory capacity when it is not strained by other
requirements.

The single exception is arsenic, which is approved as a feed additive
for hogs, chickens and turkeys. Arsenic is often used with other drugs
for which no testing method is available in a combination feed additive.
Therefore it can be used as a "tag" substance--where testing indicates
that the arsenic level is acceptable, it is assumed that levels of the
other substances used in combination with the arsenic are also acceptable.2

Increased sampling was used as a tactic to force the poultry indus-
try to adopt a withdrawal feed in early l983. The failure of the indus-

try to use withdrawal feeds free of arsenic was viewed by the Residue

Division as a problem of production system management which could be

 

IInterview with Dr. Spaulding, U.S.D.A., R.E.P.D., March 12, 1984.
2Interview with Dr. Spaulding, U.S.D.A., R.E.P.D., April ll, l984.

153

solved by the poultry industry if it were threatened with economic loss.
Due to the integration of poultry operations, the identification problem
is not.as severe as that in large animal industries and producers can be

traced:I

4.2.2.ll Conclusions

This substance by substance examination of sampling trends illustrates
that many additional factors unrelated to those mentioned by officials
discussed in Chapter Three affect sampling. A summary of these will be
provided here.

a. Factors exogenous to FSIS: The regulatory behavior of other
agencies such as FDA and EPA affects the level of some toxic substances
in the environment. When use of DES and CHC pesticides declined due to
changes in the laws regulating use, observed violations showed a declin-
ing trend. This resulted in a reduction in sampling for DES. A strong
factor in the continuing high levels of sampling for CHC's is the con-
tinuing occurrence of large scale, expensive, dangerous "incidents"
which contaminate parts of the food supply. Increasing sampling has
been the policy response despite the small likelihood that individual
incidents will be detected by random sampling.

b. Test development: Extra samples are taken in some instances
for reasons unconnected to violative behavior among producers. Occasion-
ally, the justification for sample taking is "efficient" use of laboratory

capacity.2 That is, laboratory capacity requested in annual sampling

 

1Dr. Spaulding reports that arsenic free poultry feed came into use
by mid 1983.

2Memorandum from Residue Division to Science Program Administration
re: National Residue Monitoring Program, dated November 4, 1983. p. B.

154

plans must be utilized to prevent waste (or possibly reductions in budget
allocations in the future).

c. Violations among other substances in the same group: Since
several substances can be detected with some testing procedures, FSIS
sampling response to PCB (for example) affects the sampling for Dieldrin
and other CHCs.

d. Violations among substances in different groups: Limitations in
laboratory capacity mean that resources diverted toward one problem are
diverted away from another. Therefore sampling increases in one area
help to explain decreases in others. In a similar vein, the relationship
of one substance to another in use may affect sampling, as for arsenic.

To summarize, it has been illustrated that practical and resource
limitations affect sampling by the agency. These limitations are unre-
lated to any considerations of the social harm created by violations.
Rather, they are ad hoc adjustments to the administrative necessities of
operating within a government agency setting, the characteristics of com-
pounds, and other technological factors. It is difficult to offer a clear
judgment about which of these forces--administrative adjustments or con-
sideration of social harm--is more powerful in guiding the agency's re-
source allocation decisions. All that can be unambiguously concluded is

that both play a role in determining agency enforcement policy.

4.2.3 Monitoring Sample Sizes and Statistical Validity

A concern was raised in Chapter Three over the variations in sample
size between substances, species and years. and the effect that this may
have on the validity of the violation rate as a measure of human exposure

to toxic substances in meat. What follows is a more detailed examination

155

of the implications of the size of a sample on the reliability of statis-
tics extrapolated from the sample to the entire population.

A clear relationship exists between the amount of resources expended
on collecting and testing samples, and the accuracy of estimates made of
proportions (p) of the population which have certain characteristics.
That is, the closer the sample size (n) is to the size of the population,
the more accurate will be any estimate of p (in this case, the proportion
of meat slaughtered containing violative residues).

This relationship is illustrated by the formula for determination

of sample size:

2
2 (pg)
n=_GLZ_2_
e

where n = sample size, 2 = the standard normal probability for a desired
level of confidence 1-a. p = the true population proportion. q = l-p,
and e = the allowable error in the estimate. “n" and e are inversely
related (Bhattacharyya and Johnson, l977, p. 274). Furthermore. the larger
p is, with a given n, the larger will e be. In other words, if a sample
of 300 is taken when p is .l, the estimate of p will be less accurate
than if p is actually only .Ol.

Let us return for a moment to the agency's statistical claims: a
95% confidence level is assured that a l% violation rate will be detected
within a given year's sample of 300. In this case, n = 300, a = .05 and
p is hypothesized to be about .Ol, while e is unknown. Using the formula

for sample size, however, e is determined to be:

156

2 W
2 (pg)
- a/2
e :V n
2 l
_ 1.96 (.01) (.99) _
e V 300 - .011

That is. the estimate may over or underestimate the true population pro-

 

 

 

portion by 1.1%.
Yet we have observed estimated violation rates as high as l4%. and
sample sizes frequently under l00. What would the potential error in an

estimate be when the true population proportion was .l0 and the sample

2 I
e =V1.96 16(1)) (.9) g 5.9%

The estimate of the violative proportion of meat reaching the consumer

size 100?

 

 

might be as low as .04 or as high as .l6. Let us take an extreme case:
in l977, 35 steers were tested for sulfa (see Table 3.2). No violations
were found, so the estimated proportion was 0. However, if the true
population proportion of steers violative in sulfa were .05. we might
still have achieved an estimate of zero because the size of the potential
error with such a small sample is :_7.22%.

This exercise has been done merely to demonstrate the potential for
inaccuracy of FSIS estimates of the violation rate, when either the sam-
ple size is small, the actual violation rate is high, or both. In de-
fense of the agency, it must be said that where high violation rates have
been observed, sample sizes have also tended to be high. In cases like
sulfa in swine, where l,000 or more samples have been taken, even a

14% true violation rate should be estimated within :_2.l%. Whether the

157

violation rate is estimated as 12% or 16%, the signal to the agency that
a problem exists is clear. In fact, the tendency to increase sample size
when violations are high may be motivated by a desire to improve accuracy
of the estimate on the part of the Residue Division. Such a motivation
was not revealed during interviews with officials. When estimates of
zero violations are assumed to be accurate with a sample of 25 or 30
animals, however, the agency may be acting (or not acting) on false in-

formation.

4.3 The Relationship Between Sampling and Violations

4.3.1 Introduction: Description of the Model

Theory has indicated that the potential violator weighs expected
costs and benefits of violation before making compliance decisions. The
probability of detection is a component of expected costs; as such it is
the purpose of this section to test the relationship between this probabil-
ity and the decision to violate. Further, a test of this relationship
will assist us in judging whether resources spent to raise this probabil-
ity have the desired effect on violative behavior. The "cost" aspect of
the expected cost calculation is assumed to be equal across all violations
in the formulation of the test. This assumption is justified on the basis
of the information provided in Chapter Three which indicates that penal-
ties imposed by the FDA and FSIS upon violative producers are almost

uniformly zero.1 In this test, therefore, we attempt to isolate the

 

1This is apart from the issue of whether the market may in fact im-
pose differential costs, such as a price effect of violations upon future
sales. This question provides an important direction for future research.

158

effect of probability of detection since it is one of FSIS's primary ac-
tions toward regulated producers.

In order to perform this test, proxies have been chosen to measure
both probability of detection and violative behavior. The former, in
the meat residue case, is the sampling rate or number of samples taken
out of the population (animals slaughtered within certain states subject
to federal inspection). The measurement tool for the latter (violative
behavior) is the violation rate or percentage of violations detected
out of the number of samples taken and analyzed.

The case of sulfa in swine is the only compound for which FSIS
changed the sampling intensity significantly over time within certain
states. Four states where large monitoring samples and extra "spotlight-
ing" samples were taken from 1980 to 1983 are examined here: Georgia,
Iowa, Indiana and North Carolina. Data on total numbers of random samples
taken during the four year period under the two programs was provided by

the Residue Division:I

The number of samples taken and violations found
in each month was available, providing 48 observations for each of the
four states. The violation rate (v) is the number of violations divided
by number of samples taken in a given month and state.

It has been illustrated that not all violations detected by sampling
are traced back to their source by FSIS: i.e. not all violators detected
are "caught." Since precise data is not available on the number of vio-

lations which were traceable in the four states studied between 1980-1983.

the violation rate "v" is used as a proxy for the probability of being

 

1Please refer to Appendix for detailed results of ordinary least
squares procedure.

159

"caught" in violation. This is possible by making the assumption that
the number caught is a constant proportion of the number of violations
detected in FSIS laboratories. The sampling rate (5) is the number of
samples taken divided by the total number of hogs slaughtered for a given
month and state.

FSIS has indicated that it uses the sampling rate as an enforcement
tool because increasing the number of violators caught is effective in
reducing violations. A mathematical expression of this statement is
that v = f(s). The primary purpose of this exercise is to test the vali-
dity of that statement. However, several other factors may help to ex-
plain the violation rate for sulfa in hogs as well and therefore have
been included in alternative specifications. Six specifications in total
were run using the Time Series Processor (TSP) package on the MSU main-

frame computer.

4.3.1.1 Alternative Model Specifications

Model one, the simplest form tested, is a regression of the current
period sampling rate (5) on current period violation rates. Model two
lags s by one month (s-l). on the hypothesis that some time may be re-
quired for producers to adjust to higher sampling rates. since withdrawal
periods (30 days for sulfa) must be observed to ensure residue free ani-
mals. Model three tests a two month lag of s in order to determine whether
the amount of time required for adjustment of producer practices is longer
than one month. This adjustment of the model will improve its fit if the
combination of the time required for producers to find out about the in-
creased probability due to stepped up sampling, and the necessary with-

drawal period of thirty days, exceeds one month.

160

Model four specifies that the violation rate (v) is a function of
both sampling done the previous month and sampling of two months previous.
This specification also results from the hypothesis that it takes one or
two months for producers to receive information on changes in FSIS en-
forcement intensity and change their production practices accordingly.
That is, information about the sampling intensity of two months previous
is reinforced by similar information for one month previous.

Models five and six include a lagged endogenous variable as an ex-
planatory variable. This is done because Dr. Spaulding maintains that
catching a large number of violators in one period affects the industry
through interaction of detected violators with potential violators.
Therefore, the violation rate in a previous period (v-l) should help in
explaining violations in the current period. The hypothesis here is
that the producer's compliance decision is based on current sampling and
prior "convictions"--albeit costless convictions-~among other producers
with whom she is associated. Model five regresses v-l and the sampling
rate of the current period (5) on v. in order to test this hypothesis.
Model six regresses v-l and s-1 on v, because as discussed earlier it is
not known how long it takes for information on FSIS sampling behavior to
be reflected in producer compliance decisions. Model six thus allows for
a month lag in both 5 and v.

Dummy variables were also included for three of the four states (one
state was excluded to prevent perfect multicollinearity). This was done
to test for differences between states in the relationship between sampling
and violations. Furthermore, dummies were added for each month in order

to check for seasonality in each of the six models. Again one month was

161

excluded to prevent perfect multicollinearity. Table 4.3 summarizes the
equations tested and the variables included in each.

Two further procedures were used in order to derive more information
from these models. The first was to run each state separately through
each of the six model specifications. It was hoped that this procedure
would help in explaining the specific differences between states, if any
exist, better than the gross measure provided by the state dummy variables.

The second procedure used was the inclusion of a step function to

 

represent the granulation of sulfa. This event occurred gradually between
November 1981 and June 1982,.l as a response by sulfa producers to the

high residue levels being observed in swine. Granulation of sulfa, it
was felt, would reduce electrostatic clinging in feed equipment and re-
sulting carryover. Members of the animal feed industry maintain that
granulation of sulfa was responsible for the eventual reduction of sulfa
violations in hogs. For this reason, it was tested as an explanatory
variable in the six complete models (i.e. those which included all four

states together using dummy variables).

4.3.2 The Results of Ordinary Least Squares Regression

This section will report only the results of the regression proce-
dure of each of the six models. Where possible, generalizations have
been made over all of the models when results were similar. Discussion
of the implications of the results is left for the following section.

2

All six of the model specifications resulted in very low R and ad-

justed R2 statistics. The former ranged from .09 to .13, indicating that

 

1Interview with Dr. Schmidt, FDA, BVM, May 1 1984.

162

TABLE 4.3 Alternative Model Specifications1

 

Model One:

Model Two:

t-l
Model Three:

t-2
Model Four:

v = a + bs + cs

t-l t-2
Model Five:

v = a + bvt_1 + cst

Model Six:

II
D)
+

th-i + cSt-i

 

1This table demonstrates the primary differences between models 1-6.
Each model also included dummies for states and months, as well as a step
function for granulation. The error term is implied.

163

only 9-13% of the variation in the dependent variable (the violation rate)
was explained by the models as specified. Adjusted R2, which takes ac-
count of the number of variables included in the model was much lower,
ranging from -.008 to .025.

As a result of these low R2's all six models fail the F test at the
95% level of confidence used to measure the significance of the explana-
tory power of the models. There are differences between the F statistics
for the various models, however. Models five and six have F's in the
vicinity of 1.23, while those for equations 1-4 are approximately .8-.9
(degrees of freedom are almost identical between the equations).

"T" tests for significance of the coefficients on the independent
variables are also failed by most of the included variables at the 95%
level. Among the month dummies, most have negative coefficients, but
none are significant. That is, the effect of the included months on v
is not significantly different from that of the excluded month. Strong
seasonality in violative behavior is therefbre not indicated.1

In three instances, significant t statistics are reported for vari-
ables. In equation five, the lagged endogenous variable (v-l) is nega-
tive and significant. In equation six, both the violation rate of the
previous month and the intercept shifting dummy for Iowa are significant
and negative. When separate state regressions were run, coefficients for
v-l in Indiana were significant in equations five and six; no significant

t's were reported for other states or other equations.

 

1When each state was run as a separate equation, monthly patterns
did emerge in Iowa. Violations were significantly lower in March, April,
June, July and November than in the excluded month. In Indiana, viola-
tions were higher in November in equations 2-6.

164

Although the equations as specified did not support the hypothesis
that the sampling rate is a strong factor in determining the violation
rate, the behavior of variables is worth mentioning. While the coeffi-
cient on s was positive in equation one, in all of the other equations,
which included lags for 5, its coefficient was negative.

A final comment can be made about the inclusion of the granulation
variable to explain the violation rate for sulfa among hogs. No signi-
ficant differences were observed between the complete models run with
and without this variable. In some cases, R2 statistics were higher
(which is to be expected) when the "9" variable was included, while ad-

2 and F statistics were lower. In other cases, adjusted R2 was

justed R
slightly higher when "9" was included, but none of the differences ex-

ceeded .01 in magnitude.

4.3.3 Interpretation of the Results

If the objective of this exercise had been to describe and predict
producer behavior with regard to violation of residue tolerances for
sulfa in hogs, we would have to conclude that the exercise had failed.
Failure of the F test at a 95% confidence level indicates that the vari-
ables chosen, when related to the dependent variable in the forms speci-
fied, did not explain a significant amount of the behavior of that depen-
dent variable. In this case we could say that the sampling rate, whether
current, lagged by one month, lagged by two months, or combined with the
violation rate of previous months, does not appreciably affect the viola-
tion rate.

Fortunately, it was not our objective to fully explain violative

behavior. This research takes the perspective of the agency which has

 

165

certain enforcement tools at its disposal. It has the option of allocat-
ing resources toward the use of a particular tool, or not--and using the
resources elsewhere. In a sense, we are attempting to say something
about the opportunity cost of using sampling as an enforcement tool, by
measuring its effectiveness.

The results of the techniques used here do not support the hypothe-
sis that increasing the probability of detection (within the range demon-
strated in the sulfa case) effectively deters violation.1 No strong ne-
gative relationship between the sampling rate and the violation rate has
been shown. If it is safe to assume that other things remained equal
during this time period, the FSIS could be assured that use of laboratory
and financial resources to raise sampling was ineffective as an enforce-
ment tool. This conclusion is accurate unless some exogenous factor
would have raised the violation rate even higher in the absence of in-
creased sampling. No such exogenous factor has been uncovered by this
research.

Several other points can be made, however, which mitigate the above
conclusion. Firstly, the change in sign between equation one and equa-
tions two through six indicate that while the sampling rate and the viola-
tion rate may be positively associated at first, a month later they are
negatively associated. Since the t statistics are not significant,
strong conclusions may not be drawn about this phenomenon. In fact,

with the information that has been provided earlier in this thesis, we

 

1It must be recognized that the geographical basis of this conclu-
sion is limited to the four states studied and the time period 1980-1983.
Different results might be observed in areas or time periods when condi—
tions determining violative behavior differed significantly.

166

might suspect that other things are happening which are positively cor-
related with the increase in sampling which may in fact be creating the
apparent negative effect on violations.

This tentative idea is further supported by the significance of the
previous month's violation rate in explaining the violation rate in the
current month. Dr. Spaulding indicated that his objective in increasing
sampling was to put pressure on the industry to police its own members.
When a high violation rate in one month is strongly correlated with a
low rate in the next, the question arises as to whether it is in fact
the sustained level of sampling which creates deterrence, rather than
monthly changes which may be viewed as temporary by producers. One
might conclude that it is the interaction of the detected violators with
the potential violators that reduces violations. Such interaction might
deter violators because livestock and meat tends to be an undifferentiated
product. A high violation rate among hog producers may motivate members
of the industry to take more care to reduce such violations due to their
potentially negative effect on consumer confidence. This kind of reac-
tion is produced by the externalities of violations upon other producers.

One way of creating this dynamic is to follow the course that FSIS
has taken in the past: increasing sampling so that a larger number of
violators are caught, creating Dr. Spaulding's "critical mass" of detected
and identified violators. Yet other options also exist. If the key is
to detect a larger number of violators, an alternative approach might be
to improve the connection between detecting violations in meat samples
and tracing, informing, and prosecuting those who cause the violations.
As pointed out earlier, when expected costs of violation are a function

of inspection, detection, and a follow-up which imposes costs. success 1"

167

deterrence depends on all three. If the probability of tracing a viola-
tion to its source remains at approximately 50%, as cited for veal in
Chapter Three, and the probability of any punishment being effected by
FSIS/FDA is close to zero, then spending money to raise the probability
of inspection alone may not be as effective a use of resources as possible.
If the same number of samples were taken, but most of the violators
were caught and penalties imposed, the absolute number of violators de-
tected and punished would increase. The Residue Division has been using
variable v as a proxy for the number of violations traced due to the posi-
tive relationship between the two. However, the violation rate may go
up and down independently of sampling and these exogenous changes in v
may affect the absolute number of violators caught. The point here is
that s does not need to be changed to raise the number of violators
caught. Other things do need to be changed, however; specifically, the

ability of the enforcement system to trace and punish violators.

4.3.4 The Limitations of Regression Models

Use of a mathematical statement and measurement technique suffers
certain limitations at two points: firstly, when the relationships pro-
posed are being translated from verbal into mathematical terms, and sec-
ondly when the mathematical results are interpreted, or translated back
into English. For example, the mathematical statement may not be in a
form which completely captures the nuances of the social relationships
being measured. We have used a linear form to measure the relationship
between sampling rates and violation rates. If the relationship is ac-
tually curvilinear, quadratic, hyperbolic, or logarithmic, the format

used may have failed to measure it accurately.

168

A linear form was used because the range of variation in the explan-
atory variable (5) is very small. We have observed a change in the prob-
ability of detection from some ten samples in a hundred thousand hogs
slaughtered up to perhaps 150 samples from the same amount slaughtered.
This represents an increase from a probability of .0001 of being detected
to a probability of .0015. At this small range, it was assumed that a
linear form would closely approximate a curvilinear relationship, if one
existed.

The range issue must also be kept in mind when extrapolating the
results of a specific model to general cases. The test performed here
neither supports nor disproves Becker's theory that the probability of
detection is inversely related to violations of the law. It merely con-
cludes that hog producers probably do not react in a linear fashion to
the sampling intensity created by the FSIS when using practices which
create residues. A plausible defense of Becker's theory is that the
expected costs created by FSIS and FDA are so low that they are below
some threshold of observance by livestock producers. This is true be-
cause no matter what the probability is of being detected, if the costs
of being detected are zero then expected costs of violation are zero.

No counterweight exists to the expected benefits of violation. When
these rise, violations will also rise. Future research should inquire
into the factors determining these expected benefits, which may differ
over time and between producers. They may have greater explanatory
power of violative behavior than the sampling rate does.

Finally, it should be mentioned that the choice of 95% level of con-
fidence is arbitrary. It was used in the standard statistical tradition.

If less confidence were required about one's conclusion, different

169

conclusions might be drawn about the results. However, most of the co-
efficients fail t tests not only at the 95% but the 90% and lower levels

as well.

4.4 Conclusions

Section 4.2 indicated some of the technical/administrative/resource
constraints which, in addition to the social harm considerations mentioned
in Chapter Three, determine the allocation of agency resources. It also
described some of the trends in allocation and some new directions FSIS
is taking in enforcement. The final topic was a discussion of the poten-
tial for inaccurate estimates of the violation rate under some of FSIS's
present sampling policies. This demonstrates a situation in which admin-
istrative or other forces have counteracted the considerations of the
social harm which might result from inattention to residues in certain
species/substance pairs.

Section 4.3 tested a statistical model which hypothesized that FSIS
sampling has an effect on violative behavior among producers. Results of
the test do not support the further use of stepped up random sampling
as an enforcement tool. The agency itself may have recognized this fact,
which could explain its experimenting with new enforcement alternatives
such as veal calf certification.

The results of this chapter call into question the performance of
the monitoring system on both of its objectives. As an information sys-
tem, adjustments to constraints in capacity have reduced the statistical
reliability of some estimates of the amount of violative meat reaching
consumers. As an enforcement system, sampling (unless used as surveil-

lance of a particular operation) has little effect on producer behavior,

170

and fails to prevent meat which is known to be contaminated with chemical
residues from reaching consumers. The policy implications of these con-

clusions will be examined in the final chapter.

CHAPTER FIVE
REVIEW OF CONCLUSIONS AND POLICY IMPLICATIONS

5.1 Introduction: Objectives

This chapter will review some of the issues raised by this research,
many of which have been touched upon in the conclusions and "implications"
sections of Chapters Three and Four. The objective is to assess whether
the enforcement system fbr chemical residues in meat allocates resources
effectively and "efficiently" in protecting the public health. Further-
more. some of the policy implications of the research findings will be

discussed.

5.2 Enforcement Programs

5.2.1 Defining the Residue “Problem"

FSIS enforcement efforts for residues rely heavily upon sampling,
both random and purposive. Several reasons exist for this, the first
being that the agency's legal mandate does not stretch far beyond the
slaughter/processing stages. Investigation and the imposition of penal-
ties on livestock producers is the function of FDA, whose varied respon-
sibilities, limited funding, and internally determined priorities have
mitigated its ability or willingness to take such actions.

Secondly, however, we can explain FSIS's response to residues by
looking at the way in which it has chosen to frame and measure the prob-
lem. Its primary (though not only) method of acquiring information about

its enforcement "problem" is through its calculation of the "violation

171

172

rate.” When the percentage of samples taken which are above tolerance
levels is "high" (how high varies between substances), FSIS recognizes
the existence of a "problem" and launches a program to "solve" it. Solu-
tions, in turn, are also conceptualized in terms of the violation rate:
lower it, and the residue "problem" is solved.

But, what does such a conceptualization mean for the consumer? From
the consumer's perspective, the violation rate actually represents the
extent to which the enfbrcement system has failed to protect the public
from potential harm by chemical residues. Since meat which is discovered
to be violative under random monitoring has been sold and consumed by the
time the residues are detected, the violation rate (if accurate) actually
indicates the amount of contaminated meat which the public has already
consumed.

If the information provided by the monitoring system resulted in
penalties being imposed upon those who have broken laws regarding market-
ing of tainted meat, we might conclude that both aspects of deterrence
(probability and cost of detection) were being implemented by this system.
Yet, we have observed that the only time costs are imposed is when the
actions of one livestock producer affect other producers such as feed
mills, storage facilities, or other livestock establishments. This oc-
curs when a residue discovery is traced to its source and the potential
for civil action exists. Using the violation rate as an indication of
the problem distracts the attention of the agency from other aspects of
residue problems. It gives no indication of who is to blame for the
existence of residues and inadequate information for the imposition of
costs of detection in violation. In this sense, sampling by the agency

in general does not perform any enforcement or deterrent function.

173

5.2.2 The Costs of Residue Violations

When residues are untraceable and no costs are imposed upon the
livestock producers who create them, it seems that the only group whose
interests are threatened is meat consumers. These consumers are a dif-
fuse group, and the damage done to any individual by the existence of
residues in meat is generally less than the large cost which would be
incurred in attempting to control or punish violators. That consuming
chemical residues imposes a health "cost" on the public is implied by
the existence of laws which control the levels of residues in foods.
(We will not debate here the validity of tolerance levels and their
safety margins, but accept as given that in the process of setting these
tolerances, acceptable and unacceptable levels of "social harm" have
been established by FDA and EPA.)

The potential for imposition of health costs upon a diffuse, un-
organized group necessitates a government "watchdog" agency, which is
supported by public tax funds. The few dollars of each individual's
tax payment which go to support FSIS, FDA, and EPA are theoretically
supposed to prevent the public good characteristics of food safety from
causing the classically observed “underproduction" problem inherent in
such goods.

This research indicates that costs are rarely if ever imposed on
meat producers when they break the laws controlling the purity of their
product. It has also been demonstrated that some of the actions chosen
by FSIS to combat residue violations are not effective in controlling
compliance decisions. In the face of low or zero cost, even an increase
in the probability of detection achieved by increasing sampling has not

been strongly correlated with a reduction in violations. Whether or not

174

a conscious choice to do so has been made, the enforcement system as it
has been mandated by Congress and implemented by the agencies has impli-
citly imposed several types of costs on consumers: firstly, the tax

cost of maintaining the agency, secondly, the potential health costs of
consuming residues, and thirdly, higher meat prices which may result if
slaughterers pass on the risk of losses due to condemnations to the con-
sumer. The potential for these three types of costs indicates that con-

sumers, slaughterers, and non-violating producers all pay the costs of

violations, and hence these costs remain largely external to the violator.
The Residue Division can point to the fact that, to the extent that moni-
toring is accurate, it indicates that few violations actually occur.

For most substances, fewer than one percent of samples tested are above
tolerance. Where higher violation rates exist, it is admitted by FSIS
that merely sampling at a higher rate is not adequate to deter violations.

New approaches to sidestep the legal constraints are being developed.

5.2.3 Enforcement Via the "Market"

In defense of the Residue Division of FSIS, it must be pointed out
that some aspects of this problem are recognized, and steps are being
taken in new directions in an attempt to resolve them within the existing
legal framework.

The new certification program for veal calves is an example. While
it cannot require animal identification and drug treatment histories on
all animals, it recognizes that these are desirable tools in creating a
real deterrent to improper animal drug use. In using on-site testing
and creating market incentives for the use of voluntary certification,

it will increase the fraction of violations which are traceable to their

175

sources. Whether increasing the chances that excess sulfa in veal calves
will be traced to its source without imposing a financial penalty will
better control violations remains to be seen. What the Residue Division
is hoping for is that easy identification and immediate condemnation will
cause slaughterers to impose costs which it cannot by lowering the prices
offered for uncertified animals. This plan is consistent with its en-
couragement of self—policing in the industry and its hopes that vertical
integration in the large animal industries will improve its ability to

enforce tolerances.

5.2.4 Indirect Enforcement: Persuasion vs. Penalties

The question which arises out of this market-oriented approach is,
why should enforcement programs designed to prevent residue crimes be
forced to rely on such indirect mechanisms? A similar critique can be
made of FSIS attempts to "educate" the producer who is presumed to vio-
late tolerances out of ignorance. Criminals of other types pay fines to
the SEC, go to jail for fraud, or lose their licenses for violating re-
strictions. Yet, farmers who violate the law which sets standards for
the purity of their product receive little more than a warning. Even
repeat violators suffer no more severe consequences than an injunction.
Agency officials and producer group representatives fail to characterize
the residue violator as a criminal. Violations, it is alleged, are pri-
marily due to ignorance of the effects of chemical use on livestock and
their eventual human consumers. Yet it might also be said that when
ignorance of safety standards is costless, no motivation exists for pro-
ducers to educate themselves about the potential hazards of improper

chemical use. When the producer is unmotivated to find out about these

176

hazards, FSIS finds that it must launch an education program. The tax-
payer thus pays for both ineffective enforcement and education of producers.

It has been alleged that the relationship between the USDA and the
producers it regulates is one of cooperation rather than conflict (see.
e.g., Wellford, 1972). Officials of the Residue Division have indicated
that a "cooperative" atmosphere is desirable because it is more fruitful;
i.e., deters better (Spaulding, July 1983 interview). This approach may
have developed partly in response to the concentrated nature of the meat-
producer clientele which is a function of their undifferentiated product.
Because spillover effects of violators' actions affect other producers,
whether violators or not, it is assumed to be in the best interest of
the industry to reduce its own violations. Yet individuals within the
industry respond to micro rather than macro-motives; a social trap exists
because individuals do not absorb the entire cost of their actions.

This attitude toward the regulated group differs markedly from that
taken by other regulatory bodies toward the groups they regulate. The
evidence of this thesis raises questions about the extent to which con-.
sumers are protected and tolerance violator-criminals are deterred through

the imposition of costs of noncompliance (i.e. punishment).

5.2.5 The Value Judgements of Enforcers

Any individual faced with the problem of how to go about enforcing
the law is required to make decisions which have distributive effects
upon various segments of the population. This occurs because administra-
tive offices which carry out the directives of Congress are given some
latitude in implementing these directives. When the Residue Division

chooses to concentrate on certain species and substances because they

5‘.

177

display residue "problems,“ effects may be felt by those industries
which produce the particular animals or substances.

Although decision makers may act as if problems define themselves
or are somehow exogenously determined, from the perspective of others
this may appear untrue. For example, as a chloramphenicol-sensitive in-
dividual I may perceive of prevention of chloramphenicol residues as the
most important priority of the FSIS. Yet as a producer of that substance,
I may perceive of FSIS attention as unwelcome, unfair, and costly.

By the same token the Residue Division must make choices about which
issues to devote resources to that are not based solely on objectively
determined, scientific fact. but on value judgements. These judgements
are affected by the goals of agency decision makers. It cannot be in-
ferred from the evidence presented here whether the Residue Divisions
goal is to enforce tolerances and protect consumers, or to forestall
negative media coverage of itself and protect the meat producers' incomes.

It has been the choice of both FSIS and FDA not to impose costs on
animal producers directly. Legislative constraints to such actions
have been pointed out. If such constraints were really binding upon FDA
and FSIS desires to punish violators, it would seem that these constraints
would be under attack from within the agencies themselves. The way in
which the costs of violation have been parceled out to consumers, viola-
tors. and taxpayers is not solely explained by technological and legal
factors, but also must include a willingness to accept the status quo by

administrators.

178

5.3 Policy Implications

If the U.S. government is seriously concerned about the health haz-
ards of residues in meat, several potential structural changes could im-
prove the system's enforcement capabilities. Some of these alternatives

are discussed below.

5.3.1 Imposition of Real Costs on Violators
There are two separate steps involved in imposing costs. First,
violators must be identified. Secondly, a fine or other punishment must

be imposed.

5.3.1.1 Tracing

The existence of a mandatory. standardized animal identification
and certification system would enable FSIS to determine an animal's pre—
vious ownership. Records of ownership and dates and types of drug
treatment, including subtherapeutic feed applications, should be kept
for each animal. Although this system would impose certain costs, adding
paperwork and other transactions costs to the existing expenses of animal
maintenance, it would also reduce other costs. The time required to
trace violations by FSIS regional offices would be reduced, and, if it
succeeded in reducing violations, implicit health costs to consumers of
meat could also be reduced.

A primary advantage to imposing recordkeeping upon animal producers
is that an educational function would also be served. Keeping records
of treatment would bring the existence of regulations controlling use of
chemicals into sharper focus. It would also serve as a reminder that
failures to observe withdrawal periods which result in the sale of con-

taminated meat could cause adverse consequences, both health and financial.

 

179

FSIS and FDA will be aware of the data and nature of treatment, creating

at least the potential for prosecution.

5.3.1.2 A Change in Sampling Approach

Residues are created by animal producers, but the sampling frame
used by FSIS is the list of federally inspected slaughter establishments.
One way of making sampling and testing more immediate to the producer
would be to sample randomly among producers rather than among slaughterers.
Sample taking could still occur at slaughter plants, but with reference
to the source of the animal. The data requirements of this approach
would be large, including information on the marketing channels used by
each livestock producer.

Clearly, this is a much more viable option in certain industries
than others. FSIS is already taking this approach to some extent in
the poultry industry which is highly concentrated and more vertically
integrated than other meat industries. Sampling programs can, therefore,
be tailored to individual slaughter/processor facilities, which tend to
purchase consistently from partially or wholly owned subsidiary poultry
growing establishments.

Some 466,410 operations produced hogs in 1983, while in the beef
industry, 1,585,200 establishments operated. (Crop Reporting Board.
USDA, 1984). To attempt to sample each producer even once per year
would increase the number of samples from some 25,000 a year to over
2,000,000 for these two species alone. A two tiered system would have
to be developed whereby some producers--perhaps only the largest--were
subject to producer specific sampling. An animal identification program.

combined with legislated changes enabling FDA to impose fines, might be

180

successful enough that this potentially expensive alternative would be

unnecessary.

5.3.1.3 Enabling the System to Impose Penalties

Fragmentation of responsibilities for various aspects of enforce-
ment means that different portions of the same investigation are carried
out by different organizations. This allows the Residue Division of
FSIS to curtail its activities on a particular case when information
about violators is handed over to FDA for "fbllow-up." Difficulties in
defending tolerances in court, combined with the lack of any specified
administrative penalty for violations, allows the FDA to end investiga-
tions routinely with a warning to producers not to repeat their "mis-
takes.” Severely limited funding is another reason given by FDA officials
for the small proportion of violations which are investigated.

The existence of an animal identification/certification program
could improve the legal prospects for prosecution of violators. With
such a system, it would be easier to determine the source of a residue.
It would not, however, solve the problems of defending tolerances in
court. The combination of these barriers explains the lack of imposed
penalties to this point.

Combining all enforcement procedures within one agency and making
provisions for the assessment of administrative penalties for violators
of tolerances are major structural and legal changes. Yet, if an effec-
tive deterrent is to be created to violation of tolerances by animal
producers, these changes may be required. Merely increasing the number
of samples taken under the present structure has not been shown to be

effective. As pointed out by the GAO (GAO, 1979, p. 50), the agencies

 

181

should be authorized to impose civil money penalties on producers of vio-
lative meat. Furthermore, statutory changes should be made in the Food,
Drug, and Cosmetic Act which would make misuse of animal drugs a crime.
As presently specified, only the marketing of tainted products is a
criminal offense. Pleas of ignorance by farmers are often the only out-

come of the existing law.

5.3.2 Statistical Assistance

The experience of this researcher indicates that the Residue Divi-
sion is in need of a full-time statistician with access to FSIS computers.
Presently, statistical assistance is provided by another division (Mathe-
matics and Statistics) within the Science Program on a "time available"
basis. This is inadequate for the needs of the Residue Division and has
resulted in confusion and inaccessibility of data.

The primary duty of the Residue Division is to monitor the levels
of residues in the meat supply, a task which is carried out by statistical
inference. When data is in disarray, it cannot be interpreted with con-
fidence. Furthermore, an individual whose job is to make use of the data
could use it to the fullest, adding more detail to the information de-

rived from it while monitoring its quality and accuracy.

5.3.3 Improving Technology

Some of the most "profitable" investments of FSIS in terms of deter-
rent-creating potential per dollar spent, have been in the area of im-
provements in testing capability. The existence of on-site testing for
antibiotics and sulfas have permitted FSIS to establish closer links to

meat producers; because they allow for more extensive testing and quicker

.- - IL n- ..K 1‘“!
I 9- 33 A

182

identification of problems. Costs can be imposed because carcasses need
be held only overnight, and hence can be condemned and removed from the
market before consumption.

Therefore, a final policy recommendation is that FSIS continue to
concentrate its efforts on expanding capabilities for on-site testing.
This would be particularly useful in the area of chlorinated hydrocarbons.
The development of a swab test and inexpensive live animal and feed tests
for these toxic substances would greatly improve the capacity of both
the enforcement system and farmers themselves to prevent (rather than

merely reacting to) contamination incidents.

5.4 Limitations of the Research

Since the topic of this thesis was limited primarily to the enforce-
ment programs of the FSIS for residue tolerances in meat, several impor-
tant questions raised have not been answered. The following is a listing

of these questions. which also comprise some suggested research directions.

5.4.1 The Costs and Benefits of Proposed Changes

While some of the suggested policy directions have been assessed
(e.g., Kramer 1982. p. 407, on cost of animal identification), more infor-
mation is needed on the potential costs and benefits of such actions and
their distribution. Merely suggesting structural changes does not solve
distributional issues such as the way in which an animal identification
program would be implemented, and who would bear its costs. Such a study

is needed before any changes occur.

 

183

5.4.2 The True Determinants of the Compliance Decision

The model presented in Chapter Four failed to explain a large pro-
portion of the variation in the amount of violative behavior observed.
Further research is required on the actual determinants of producer be-
havior when it comes to obeying laws. This would not only be useful as
an end in itself, but also as a means to understanding how enfbrcement

agencies can best create deterrence.

5.4.3 The Role and Activities of FDA in Enforcing Tolerances

Limited information about the role played by the FDA has been in—
cluded here. A clearer understanding of the competing priorities within
this division of HHS would give a broader perspective and comprehension
of some of the barriers to effective enforcement. Of primary importance
is the legal structure which constrains enforcement actions such as the
imposition of costs on violators. Also of interest is the level of
funding received by FDA, the number of other functions it performs, and
the way in which funding is allocated between them. Since labelling is
FDA's responsibility, and often the sole source of information about
proper substance usage, further study is needed on the contributions of

mislabelling to residue problems and misuse.

APPENDIX

Right Hand
Variable

C (Y Intercept)

5 (Sampling
Rate)

st-l (Sampling
Rate, Previous
Month)

St-2 (Sampling
Rate, 2 Months
Previous)

Vt,1 (Violation
Rate, Previous
Month)

St (State
In ercept
mmmy.GA)

St (State
Ingercept
Dummy, IA)

St (State
Ingercept
Dummy, 1N)

53 (State
51 pe Oumny ,
GA)

582 (State
Slope Dummy,
IA)

583 (State
Slope Dummy,
1M)

6 (Granula-
tion)

 

TABLE A.1

184

For Complete Form of Models 1-61

Model 1 Model 2
Coef. T Coef. T
.11 2.93 .14 3.47
.11 .88 -- ~-

-- -- -.15 -1.13
.01 .44 ..Ol -.42
.06 ~1.32 -.07 «1.84
.03 -.89 p.03 -.98
.10 -.72 .18 1.31
.29 .20 .37 .30
.16 .67 .14 .59
.03 1.13 .01 .60

 

 

jRounded to nearest

hundredth.

 

Model 3
Coef. T
.13 3.42

-.08 -.63
.Ol .02
-.O7 ~1.73
-.04 -l.16
.08 .62
.32 .27
.24 .97
.02 .67

 

Model 4
Coef. T
.14 3.52

-.14 -1.06
-.05 -.40
-.Ol -.44
-.08 «1.86
-.04 -l.23
.24 1.56
.33 .18
.06 .23
.01 .49

 

Estimated Coefficients and T Statistics

Model 5
Coef. T
.13 3.37

-010 .82
-.18 -2.31
.02 .61
«~06 -1.54
-.03 -.96
-.O9 -.70
.28 .20
.17 .70
.02 1.14

 

.13

.18

.Ol

.08

.03

.16

.41

.16

.01

Model 6
Coef.

.15

T
3.83

-.99

-2.31

-.23

~2.06

-1.06

.71

-__

 

 

Right Hand Model 1
Variable Coef. T
M1 (Month

Dummy, Jan.) -.08 -l.55
n2 (Feb.) -.08 -l.63
M3 (Mar.) --.01 -.16
M‘ (Apr.) -.08 -1.71
nu5 (ng) -.oo -.08
M5 (June) -.03 -.73
M7 (July) -.07 -l.64
n3 (Aug.) -.05 -l.05
M9 (Sept.) -.08 -l.75
M10 (Oct.) -.04 -.94
n1, (Nov.) -.03 -.7o

1

Rounded to nearest

 

1535
TABLE A.1 (Cont.)

Model 2 Model 3 Model 4
Coef. T Coef. T Coef. T
-.06 -n.35 -.06 -l.2 -.oa -n.é9
-.06 -l.21 -.07 -1.4 -.05 -1.07
.00 .on -.01 -.13 -.oo -.oz
-.08 -l.72 -.08 -1.73 -.08 -1.67
-.01 -.12 -.on -.16 -.on -.19
-.03 -.73 -.03 -.75 -.03 -.76
-.08 -l.76 -.07 -1.66 -.09 -l.93
-.os -l.02 -.04 -.97 -.04 -.94
-.07 -l.56 -.07 -1.59 -.07 -l.53
-.04 -.98 -.04 -.91 -.04 -.9o
-.03 -.72 --03 -.69 -.04 -.a3

hundredth.

 

 

 

Model 5
Coef. T
.08 -1.55
.08 -1.77
.01 -.22
.07 -1.65
.01 -.29
.03 -.65
.07 -l.69
.05 -l.24
.08 -1.83
.05 -1.13
.33 -.77

 

Model 6
Coef. T
.06 -l.33
.07 -l.39
.00 .07
.07 -l.65
.01 -.31
.03 -.63
.08 -l.80
.05 -1.20
.07 -l.64
.05 -1.18
.03 -.77

186

TABLE A.2 F Statistics, R2, and Adjusted
R2 for Complete Form of Models 1-6

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

02 .1037 .1016 .0970 .1126 .1322 .1300
Adjusted R2 -.0001 -.0025 -.OO76 -.0150 .0258 .0233
F2 .9991 .9762 .9271 .8823 1.2421 1.2179

Degrees of Freedom 19,164 19,164 19,164 23,160 20,163 20,163

 

1Rounded to nearest ten-thousandth.

ZSignificant r at a = .05, v1 = 20, v2 . - is 1.57. At a = .10. Significant
F . 1.42 (Bhattacharyya and Johnson. 1977, p. 607).

 

BIBLIOGRAPHY

BIBLIOGRAPHY

"Agriculture Department to Investigate a Meat Plant in Denver." New_
York Times. September 20, 1983.

 

American Veal Association. Vealer U.S.A. November 1983.

Ashenfelter, Orley and Smith, Robert S. "Compliance with Minimum Wage
Law." Journal of Political Economy. Vol. 87, #2 (1979): 333-350.

pm” ”Muff

Bartlett, Randall. Economic Foundations of Political Power. New York:
The Free Press, 1973.

Baron. Barry R. and Baron, Philip. "A Regulatory Compliance Model."
Journal of Contemporary Business. Vol. 9, #2 (1980): 139-150.

Becker, Gary S. "Crime and Punishment: An Economic Approach." Journal
of Political Economy. Vol. 76, #2 (1968): 169-217.

Bhattacharyya, Gouri K. and Johnson, Richard. Statistical Concepts and
Methods. New York: Wiley, 1977.

 

Booz, Allen, and Hamilton, Inc. "Study of the Federal Meat and Poultry
Inspection System." Unpublished Study Commissioned by U.S. Depart-
ment of Agriculture, July 1977.

Breit, William and Elzinga, Kenneth G. "Antitrust Penalties and Attitudes
toward Risk: An Economic Analysis." Harvard Law Review. Vol. 86.
#4 (1973): 693-713.

 

Cole, Roland J. and Sommers, Paul. "Business and Government Regulation:
A Theory of Compliance Decisions." Journal of Contemporary Business.
Vol. 10, #1 (1981): 143-153.

Cyert, Richard and March, James. A Behavioral Theory of the Firm.
Princeton, N.J.: Prentice-Hall Inc., 1963.

Diver. Colin. "A Theory of Regulatory Enforcement." Public Policy.
Vol. 28, #3 (1980): 257-299.

 

Federal Environmental Pesticide Control Act. U.S. Code. Vol. 7, sec.
136. 1972.

Federal Food, Drug and Cosmetic Act. U.S. Code. Vol. 21, sec. 301,
1938.

187

188

Kessler, David A. "Food Safety: Revising the Statute." Science. Vol.
223, #4640 (March 9, 1984): 1034-1040.

Kramer, Carol. "An Economic Analysis of Food Safety Regulation: The
Case of Sulfa and the Swine Subsector." Ph.D. Dissertation. Michi-
gan State University, 1982.

Kunreuther, Howard and Slovic, Paul. "Economics, Psychology, and Protec-
tive Behavior." American Economic Review. Vol. 68, #2 (1978):
64-69.

 

McKean, Roland N. "Some Economic Aspects of Ethical-Behavioral Codes."
Political Studies. Vol. 27. #2 (1979): 251-65.

 

Marshall, Eliot. "Scientists Endorse Ban on Antibiotics in Feeds."
Science. Vol. 222, #4624 (November 11, 1983); 50].

Murphy, Jerome T. Gettingythe Facts, A Fieldwork Guide for Evaluators
€nd PolicyyAnalysis. Santa Monica: Goodyear Publishing Co., Inc.,
980.

 

National Association of Federal Veterinarians, American Academy of Veteri-
nary Preventive Medicine. "Proceedings of AAVPM Seminar: Every-
thing You Wanted to Know About Residues, But Didn't Know Who to Ask."
Las Vegas, Nevada. October 18, 1983.

Pennsylvania State University, College of Agriculture, Cooperative Exten-
sion Service. ”RAP: Sulfa Residues in Calves." Prepared by C.A.
Rossiter and L.J. Hutchinson. Unpublished document. No date.

Piwoni, Richard and Kliebenstein, Jim. Marketing Strategies for Veal
Calves. Madison, Wis.: University of Wisconsin Extension. No date.

 

Roberts, Tanya. Benefit Analysis of Selected Slaughterhouse Meatlnspec-
tion Practices. N.C. Project 117. Working Paper Series, WP-7l.
September 1983.

 

Shaffer, James. "Notes for a Theory of Personality for a Theory of Con-
sumer Behavior." Unpublished document. No date.

Stigler. George J. "The Optimum Enforcement of Laws." Journal of Poli-
tical Economy. Vol. 78, #3 (1970): 526-536.

 

U.S. Congress, General Accounting Office. "Chan es Underway to Correct
Inadequacies in Florida's Meat and Poultry nspection System. GAO/
RCED-83-70. December 30.1982.

 

U.S. Congress. General Accounting Office. Improving;Sanitationggg_
Federal Inspection at Slaughter Plants: How to Get Better Results
for the Inspection Dollar. CED-81-118. July 30, 1981.

U.S. Congress, General Accounting Office. Monitorin and Enfo§§jng_Food
Safety - An Overview of Past Studies. GAO7RCED-83-153. September
9. 1983.

U.S.

U.S.

U.S.

U.S.

U.S.

U.S.

U.S.

U.S.

U.S.

U.S.

U.S.

U.S.

U.S.

U.S.

189

Congress, General Accounting Office. Problems in Preventing the
Marketing of Raw Meat and Poultry Containing Potentially_Harmfu1

Residues. HRD-79- 10. April 17,1979.

 

Congress, General Accounting Office, USDA's Oversi ht of State Meat
and Poultryylnspection Could be Strengthened: GAOgRCED-84-23.

 

October 21, 1983.

Congress, Office of Technology Assessment. Drugs in Livestock Feed.
Vol. 1: Technical Report. (Washington, D. C: Government Printing
Office. 1979).

Department of Agriculture, Crop Reporting Board, Statistical Report-
ing Service. Cattle. January 30, 1984.

Department of Agriculture, Crop Reporting Board, Statistical Report—
ing Service. Hogs and Pigs. March 20, 1984.

Department of Agriculture, Extension Service, Agricultural Programs,
Livestock and Veterinary Sciences. "Residue Avoidance Program:
Summary." Unpublished document. September 30. 1983.

Department of Agriculture, Food Safety and Inspection Service. An
Analysis of Exemption Provisions of the Meat and Poultry Inspection

Laws. Marchfl983.

Department of Agriculture, Food Safety and Inspection Service.
Budget Information 1983 - 1984. Unpublished document. No date.

Department of Agriculture, Food Safety and Inspection Service. flgw_
to Perform the Live Animal Swab Test for Antibiotic Residues. Agri-
culture Handbook No. 601. January 1983.

Department of Agriculture, Food Safety and Inspection Service. Resi-
due Avoidance Program Bulletin. January 1983.

 

Department of Agriculture, Food Safety and Inspection Service. "Resi-
due Control in the Nation's Meat and Poultry Food Supply.” Notes
for an oral presentation. Unpublished document. No date.

Department of Agriculture, Food Safety and Inspection Service.
Statistical Summary, Federal Meat and Poultry Inspection for Fiscal

Year 1982. March 1983.

Department of Agriculture. Food Safety and Inspection Service, Meat
and Poultry Inspection Program. Meat and Poultry Inspection Direc-
tory)for July 1981. (Washington, D. C: Government Printing Office,
981 .

Department of Agriculture, Food Safety and Inspection Service, Meat
and Poultry Inspection Program. “Receiving, Modifying and Stabiliz-
ing the S.T.0.P." Training Module. Unpublished document. No date.

U.S.
U.S.
U.S.

U.S.

U.S.

U.S.

U.S.

U.S.
U.S.

U.S.

U.S.

U.S.

190

Department of Agriculture, Food Safety and Inspection Service,
Meat and Poultry Inspection Program. "Training Modules for Meat
and Poultry Inspection Officers." Unpublished document, 1981.

Department of Agriculture, Food Safety and Inspection Service,
Policy and Program Planning, Policy Analysis Staff. Secretary's
Discretion in Processing Plants: Background Information. 1982.

 

Department of Agriculture, Food Safety and Inspection Service.
Residue Evaluation and Planning Division. "The Economics of Residue
Control," by Dr. John Spaulding. Unpublished document. 1981.

Department of Agriculture, Food Safety and Inspection Service,
Residue Evaluation and Planning Division. "Issuance Clearance:
Testing and Disposition of 0-3 week-old Calves Suspected of Con-
taining Sulfa and/or Antibiotic Residues.“ Unpublished document.
February 9, 1984.

Department of Agriculture, Food Safety and Inspection Service.
Residue Evaluation and Planning Division. "National Residue Moni-
toring Program." Unpublished document. No date.

Department of Agriculture, Food Safety and Inspection Service.

Residue Evaluation and Planning Division. "A Residue Approach for

the 80's? Remarks by Dr. John Spaulding before the American Bac-
teriological and Chemical Research Corporation Symposium, Gainesville,
Florida, March 1982.

Department of Agriculture, Food Safety and Inspection Service,
Residue Evaluation and Planning Division. "USDA's Residue Control
Program. Considerations in Designing a Residue Control Program:
Comments on EC Proposal." Unpublished document. December 3. 1981.

Department of Agriculture. Food Safety and Quality Service. "Cooper-
ative Residue Programs." Unpublished document. No date.

Department of Agriculture, Food Safety and Quality Service. "Im-
proving the F505 Residue Program." Unpublished document. July 1980.

Department of Agriculture, Food Safety and Quality Service. Meat
and Poultry Inspection,,l980. Report of the Secretary of Agricul-
ture to the Committee on Agriculture, House of Representatives, and
Committee on Agriculture, Nutrition and Forestry,U.S. Senate. 1981.

Department of Agriculture, Food Safety and Quality Service. Report
on the PCB Incident in the Western United States. 1980.

Department of Agriculture, Food Safety and Quality Service, Meat
and Poultry Inspection Program. "Performing the Swab Test for

antibiotic Residues." Training Module. Unpublished document. No
ate.

 

191

U.S. Department of Agriculture, Food Safety and Quality Service, Patho-
logy and Epidemiology Division. An Analysis of the Federal Meat
and Poultry Inspection Program. February 20, 1977.

U.S. Department of Agriculture, Packers and Stockyards Administration.
Packers and Stockyards Resume. Vol. XVII, #3. December 28, 1979.

U.S. Department of Health and Human Services, Food and Drug Administra-
tion, Study Group on FDA Residue Programs, FDA MonitoringyPrograms
for Pesticide and Industrial Chemical Residues in Foodl June 1979.

Viscusi, W. Kip and Zeckhauser, Richard J. "Optimal Standards with In-
complete Enforcement." Public Policy. Vol. 27, #4 (1979): 437-456.

Wellford, Harrison. Sowingthe Wind. New York: Bantam Press, 1972.
Williamson, Oliver: Markets and Hierarchies. Chapter 2: The Organi-

zational Failures Framework, pp. 20-40. New York: The Free Press,
1975.

192

LIST OF GOVERNMENT OFFICIALS AND
INDUSTRY REPRESENTATIVES INTERVIEWED

Bureau of Veterinary Medicine, Food and Drug Administration.

 

 

Ballitch, Edward. Director, Division of Compliance. September 14, 1983.

Beaulieu, Andrew. Director, Division of Surveillance. September 14,
1983.

Carnevale, Richard. Deputy Associate Director for Scientific Evaluation,
Division of Scientific Evaluation. September 14, 1983.

Crandall, Max. Associate Director, Division of Surveillance and Com-
pliance. September 14, 1983.

Frappaolo, Philip. Branch Chief, Compliance Programs and Hearings Branch.
September 14, 1983.

Graber, George. Director, Division of Animal Feeds. September 14, 1983.

Pugliese. Frank. Special Assistant, Division of Scientific Evaluation.
September 14, 1983.

Ransdell, Homer. Branch Chief, Case Guidance Branch, Division of Compli-
ance, September 16, 1983.

Schmidt, Robert. Branch Chief, Medicated Feed Branch, Division of Animal
Feeds. May 1. 1984.

Settepani, Joseph. Deputy Associate Director, Division of Human Food
Safety. September 14, 1983.

White, William. Chemist, Division of Human Food Safety. September 14,
1983.

Food Safety and Inspection Service, U.S. Department of Agriculture.

Burke, William. Branch Chief, Quality Control Branch, Mathematics and
Statistics Division, Science Program. September 13 and November 23,
983.

Connaughton, Harold. Regional Supervisor of Meat Inspectors. Southern
Michigan area, Meat and Poultry Inspection. March 23, 1983.

193

Cordle, Marilyn. Deputy Director, Residue Evaluation and Planning Divi-
sion, Science Program. September 15, 1983.

Leese, William. Assistant Director, Emergency Program Staff, Compliance
Division. November 25, 1983.

Levy, Bert. Director, Mathematics and Statistics Division. Science Pro-
gram. September 13, 1983.

Petersen, Ann. General Counsel's Office. September 15, 1983.

Rellosa, F. R. Staff Assistant, Northeastern Regional Office, Meat and
Poultry Inspection. September 12, 1983 and March 15, 1984.

Rubin, Leslie. Animal Physiologist, Residue Evaluation and Planning
Division, Science Program. November 23, 1983.

Spaulding, John. Director, Residue Evaluation and Planning Division,
Science Program. July 29, and November 23, 1983, and March 12, 1984.

Wissman, Ken. Data Management, Technical Services, Meat and Poultry
Inspection. September 15, 1983.

Industry Representatives

 

Adams, John. National Milk Producers Federation. September 8. 1983.

Frederick, Howard. American Feed Manufacturers Association. September
9. 1983.

Hodges, James. American Meat Institute. September 8, 1983.
Kimbel, Steven. Animal Health Institute. September 13, 1983.