. =€4
. 1.7.9 .c .x
I. .,.7- 3'
3i;
It“: 1 I
5 .

:1 .fl.

A

AIR ~ # ..
.. s.» .1... 4n"..-
. .3. a». . . 5.... fl... .
H..m.q...£¥mwm§?& .

.1

nm

55

xi... . :v I
....; %
cl}... :0 .
1 1':

run.

 

 

mam LIBRARY

9 , Michigan State

. 000 . .
Unuversuty

 

 

 

This is to certify that the
dissertation entitled

POLITICAL INSTITUTIONS AND BUREAUCRATIC
AUTONOMY IN THE US. REGULATORY POLICY
PROCESS

presented by

Doo-Rae Kim

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Political Science

 

'gJIWV N» W

Major Professor’s Signature
M Ink i i . a 00 ST
U r

Date

MSU is an Affinnalive Action/Equal Opportunity Institution

--¢-c—c—o—-— -

-o-o-o-n—n— .A

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_._' —«

POLITICAL INSTITUTIONS AND BUREAUCRATIC AUTONOMY
IN THE US. REGULATORY POLICY PROCESS

By

Doo-Rae Kim

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Political Science

2005

ABSTRACT

POLITICAL INSTITUTIONS AND BUREAUCRATIC AUTONOMY
IN THE US. REGULATORY POLICY PROCESS

By

Doo-Rae Kim
The proper role of bureaucracy and its unelected officials in democratic governance has
long been a matter of controversy. One part of the debate involves the argument that
democratic control and bureaucratic autonomy are opposites: if there is democratic
control, there cannot be bureaucratic autonomy, and vice versa. This dissertation reveals
that conditions of democratic control and bureaucratic autonomy are not incompatible in
this fashion: government agencies are subject to political control to some extent but the
agencies also can strategically maneuver among competing and divided political
institutions to make autonomous policy choices. This research also shows that these
institutional impacts on both bureaucratic responsiveness and bureaucratic autonomy are
mediated by how policy preferences are distributed inside the bureaucratic agency. The
varying degrees of bureaucratic policy bias influence the institutional dynamics of
bureaucratic policy choices in predictable ways.

Narratives of legislative policymaking on various issues of occupational safety
and health and an extensive analysis of data on occupational safety and health
enforcement provide evidence for the theoretical advancement. While the level of
regulatory policy outputs of the Occupational Safety and Health Administration (OSHA)
was determined primarily by directional changes in the preference configuration of
representative institutions in the policy space, bureaucrats’ autonomous policy choices

regarding OSHA enforcement were determined by the degree of preference divergence

among those political institutions. Moreover, the magnitude and significance of the
effects of institutional interactions on regulatory behavior varied systematically with
occupational safety and health agencies’ preference distributions, thereby supporting the
general argument that bureaucratic responsiveness and bureaucratic autonomy can be
better understood by considering the interplay of institutional relations and bureaucratic

policy preferences.

To the memory of my grandfather

iv

ACKNOWLEDGMENTS
It gives me great pleasure to acknowledge those who have supported my research.
Serving as chair of my guidance and dissertation committees, Professor Thomas H.
Hammond has helped me tremendously from the beginning. Thanks to his detailed
comments on numerous versions and his generous contribution of insightful thoughts,
this work has been completed. I would also like to thank other members of my
dissertation committee. Professor Belinda Davis has given me inspiring comments and
generous help ever since I participated in her State Politics and Policy seminar. I
benefited greatly from the comments and criticisms of Professors Burt Monroe and Ken
Boyer. Thanks also are extended to Dr. Joseph DuBois of the Occupational Safety and
Health Administration for his generous help in collecting IMIS data.

During my years at Michigan State, I also benefited from support and help of
many others. As the former graduate program director, Professor Brian Silver had
offered me the generous support of scholarship. I thank the excellent staff of the
Department of Political Science for their superb help in response to my countless
requests of various sorts over the years. I also acknowledge the financial support of the
Graduate Office for this dissertation research.

I am deeply indebted to my beloved family in Korea and the US. My parents
have always given me their unconditional love. Without their unwavering trust and
support, I would have been unable to pursue my studies this far. I am very grateful to
my parents-in-law for their kind words. Their encouragement offered me great comfort.
I also sincerely thank my sister and brother for inviting me to go abroad while

shouldering personal responsibilities on my behalf. I am grateful to my little ones, Dong

Y001 and Lauren, for the happiness and joy they have brought into our home. They
merit great credit for being patient with their busy father. Growing from a delicate
young lady into a strong wife and mother, J iyoung has been a great partner in our
journey from Seoul to East Lansing. Finally, I would like to acknowledge the profound
influence of my late grandfather in my formative years. It is my honor to dedicate this

dissertation to him.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. ix
LIST OF FIGURES ............................................................................................................. x
CHAPTER 1 INTRODUCTION ....................................................................................... 1
1.1. Research Questions .......................................................................................... 7
1.2. Research Plan ................................................................................................. 1 1
CHAPTER 2 POLITICAL CONTROLS AND BUREAUCRATIC AUTONOMY ....... 13
2.]. Political Controls ............................................................................................ 14
2.1.1. Legislative Controls ........................................................................ 15
2.1.2. Presidential Controls ....................................................................... 18
2. 1 .3. Multi-Institutional Controls ............................................................ 21
2.2. Bureaucratic Autonomy ................................................................................. 27
2.2.1. Agent Problems ............................................................................... 30
2.2.2. Complexities of Hierarchy .............................................................. 32
2.2.3. Multilateral Institutional Relations ................................................. 34
2.3. Limitations of Past Research .......................................................................... 37
2.3.1. Conceptual Ambiguity: ‘Independence,’ ‘Noncompliance,’ and
‘Reciprocity’ ................................................................................... 38
2.3.2. The Omission of Collective Instituional Actions ............................ 42
2.3.3. Methodological Pitfalls ................................................................... 44
CHAPTER 3 THEORETICAL FRAMEWORK ............................................................. 48
3.1. A Spatial Model ............................................................................................. 51
3.1.1. Policy Disequilibrium and Policy Equilibrium ............................... 53
3.1.2. Policy Equilibrium Interval and Agency Actions ........................... 55
3.1.3. Heteroscedastic Distribution of Bureaucratic Policy
Preferences ...................................................................................... 62
3.1 .4. Simulation ....................................................................................... 69
3.2. Veto Players ................................................................................................... 75
3.2.1. The Majoritarian Perspective .......................................................... 77
3.2.2. The Distributive Politics Perspective ............................................. 79
3.2.3. The Partisan Government Perspective ............................................ 80
3.3. Issue Characterisitcs ....................................................................................... 82
3.3.1 . Salience ........................................................................................... 82
3.3.2. Partisan Polarization ........................................................................ 83
3.4. Statistical Model: Bridging the Formal/Emprical Gap .................................. 86
CHAPTER 4 POLITICAL INSTITUTIONS AND OCCUPATIONAL SAFETY
AND HEALTH REGULATION ............................................................... 93
4.1. The Occupational Safety and Health Act of 1970 ......................................... 94
4.2. F uctions of the Occupational Safety and Health Administration ................... 96

vii

4.2.1. Standard-Setting Rulemaking ......................................................... 96

4.2.2. Enforcement ................................................................................... 97
4.2.3. Compliance Assistance ................................................................ 103
4.3. State Occoupational Safety and Health Programs ....................................... 105
4.4. Institutional Environments of OSHA ........................................................... 109
4.4.1. Encactrnent of the OSH Act (1968-1970) ..................................... 109
4.4.2. Labor-HEW Appropriations Bills (1972-1980) ............................ 116
4.4.3. The Risk Notification Bill (1986-1988) ........................................ 123
4.4.4. Reform, Overhaul, and OSHA (1992-1996) ................................. 127
4.5. Institutional Environments of OSHA in Perspectives .................................. 132
4.5.1. The Majoritarian Perspective ........................................................ 133
4.5.2. The Distributive Politics Perspective ............................................ 136
4.5.3. The Party Government Perspective ............................................... 138
4.6. Discussion .................................................................................................... 141

CHAPTER 5 FEDERAL OCCUPATIONAL SAFETY AND HEALTH

REGULATION ENFORCEMENT ......................................................... 144

5.1. Federal OSHA Enforcement Activities ........................................................ 145

5.2. Modeling Federal OSHA Enforcement Activities ....................................... 156

5.2.1. Dependent Variable ....................................................................... 158

5.2.2. Model Specification for Bureaucratic Responsiveness ................. 158

5.2.3. Model Specification for Bureaucratic Autonomy ......................... 168

5.3. Empirical Results ......................................................................................... 173

5.3.1. Majoritarian Politics ...................................................................... 173

5.3.2. Distributive Politics ....................................................................... 178

5.3.3. Party Government Politics ............................................................ 183

5.3.4. Mediating Effects of the Agency Preference Distribution ............ 187

5.4. Discussion ................................................................................................... 192
CHAPTER 6 STATE OCCUPATIONAL SAFETY AND HEALTH REGULATION

ENFORCEMENT .................................................................................... 196

6.1. State Occupational Safety and Health Enforcement Activities .................... 198

6.2. Modeling State OSH Enforcement Activities .............................................. 204

6.3. Empirical Results ......................................................................................... 205

6.3.1. Majoritarian Politics ...................................................................... 205

6.3.2. Distributive Politics ....................................................................... 210

6.3.3. Party Government Politics ............................................................ 215

6.3.4. Mediating Effects of the Agency Preference Distribution ............ 220

6.4. Discussion ................................................................................................... 223

CHAPTER 7 CONCLUSION ........................................................................................ 227

BIBLIOGRAPHY ........................................................................................................... 236

viii

LIST OF TABLES

Table 1. OSHA Health Standards (1972-2001) ................................................................ 98
Table 2. OSHA Safety Standards (1972-2001) ................................................................. 99
Table 3. State Occupational Safety and Health Programs .............................................. 108
Table 4. Federal OSHA Regional Jurisdictions .............................................................. 148
Table 5. Federal OSH Inspections by Region and State (1992, 2000) ........................... 149
Table 6. Policy Equilibrium Intervals (The 91"‘~106th U.S. Congresses) ....................... 160

Table 7. ML Hetroscedastic Normal Regression Analysis of the Effects of
the Majoritarian Politics on Federal OSH Enforcement ................................. 175

Table 8. ML Hetroscedastic Normal Regression Analysis of the Effects of
the Distributive Politics on Federal OSH Enforcement .................................... 180

Table 9. ML Hetroscedastic Normal Regression Analysis of the Effects of

the Party Government Politics on Federal OSH Enforcement .......................... 185

Table 10. Quantile Regression Analysis of the Mediating Effects of the Agency
Preference Distribution on Federal OSH Enforcement ................................... 190
Table 11. State OSH Inspections (1992, 2000) ............................................................... 200

Table 12. ML Hetroscedastic Normal Regression Analysis of the Effects of
the Majoritarian Politics on State OSH Enforcement ..................................... 207

Table 13. ML Hetroscedastic Normal Regression Analysis of the Effects of
the Distributive Politics on State OSH Enforcement ...................................... 212

Table 14. ML Hetroscedastic Normal Regression Analysis of the Effects of
the Party Government Politics on State OSH Enforcement ............................ 217

Table 15. Quantile Regression Analysis of the Mediating Effects of the Agency
Preference Distribution on State OSH Enforcement ....................................... 222

ix

LIST OF FIGURES

Figure 1. Policy Disequilibrium and Policy Equilibirum .................................................. 54
Figure 2. Policy Equilibirum Interval and Agency Policy Choices .................................. 58
Figure 3. Heteroscedastic Distribution of Bureaucratic Preferences ................................ 65
Figure 4. Change of Policy Equilibirum Interval and Agency Policy Choices ................ 67
Figure 5. Monte Carlo Simulation .................................................................................... 71
Figure 6. Issue Characteristics and Types of Regulatory Politics ..................................... 85
Figure 7. Mapping Agencies onto Policy Continuum ....................................................... 91
Figure 8. Citations per Inspection by Violation Types (Federal, 1982-2000) ................ 152
Figure 9. Penalties per Inspection by Violation Types (Federal, 1982-2000) ................ 153
Figure 10. Issue Salience: New York Times Coverage of Occupational Safety and Health
Issues (1969-2000) ......................................................................................... 163
Figure 11. Policy Equilibrium Intervals (The 91‘°’t~106th U.S. Congresses) .................... 170
Figure 12. Citations per Inspection by Violation Types (State, 1982-2000) .................. 200
Figure 13. Penalties per Inspection by Violation Types (State, 1982-2000) .................. 203

CHAPTER 1

INTRODUCTION

Regulatory politics and policies have drawn an enormous amount of attention from
political scientists and policy researchers. The attention to regulation has grown along
with the expansion of govemment’s role in our everyday life. The US. government now
has something to say not only about protecting consumers’ interests from large firms,
providing welfare benefits, preserving the environment, and maintaining healthy and safe
workplaces but also about allowing women to terminate unwanted pregnancy,
prohibiting youths from accessing intemet pornography, and granting hopeless patients
the right to die with dignity. New research projects have emerged as these new areas
have been defined as policy “problems” that call for government intervention.

Despite the ubiquity and complexity of regulatory policies, most research has
attempted to provide an answer to one simple but most important question: “Why do the
governmental agencies intervene in the private sector of society in the way they do?”
This dissertation provides one part of the explanation: institutional preferences and rules
affect bureaucratic choices on regulatory policy alternatives in the area of workplace
safety and health. I develop a theoretical framework by which I examine how inter-
institutional dynamics and rules may affect bureaucratic policy choices. I focus this
research on two aspects of bureaucratic behavior such as bureaucratic responsiveness and
bureaucratic autonomy.

As a prelude to discussing why this is an important matter, a brief discussion

about important themes in prior explanations of such regulatory politics is in order.

First, the state of regulation can be thought to be determined by the nature of its origins.

 

J. Q. Wilson (1980: 364-3 72) explains variations in regulatory policies in terms of
patterns, actors, and consequences based on the distribution of perceived costs and
benefits associated with the proposed policy. Costs and benefits may be perceived as
being widely distributed or narrowly concentrated. When both costs and benefits are
expected to be widely distributed, “majoritarian politics” takes place. Since no definable
part of society such as an industry and a locality can get either disproportionate benefits
or avoid a disproportionate share of costs, any strong support or opposition from
particular segments of society is unlikely to occur. Thus, regulatory measures that seem
to offer a net gain to the majority will be adopted.

When both costs and benefits are considered to be narrowly concentrated on
particular segments of society, “interest-group politics” prevails. Here a certain segment
of society can benefit from the regulation at the expense of another segment of society.
Each side, the beneficiary group or the regulated group, has a strong incentive to
organize and exercise political influence to promote or avoid the measure, while the
voice of general public remains weak.

When the benefits of a prospective policy are concentrated but the costs are
widely distributed, “client politics” is expected to result. While the beneficiary group is
likely to organize in support, the large numbers of people who bear the diffused burdens
at a low per-capita rate have little incentive to organize in opposition. Although
watchdog public interest groups may emerge, client politics produces regulatory
measures that almost exclusively serve economic groups’ interests.

Finally, when the benefits are widely distributed at the expense of costs

concentrated on a small segment of society, “entrepreneurial politics” is likely to occur.

While the regulated group has a strong incentive to take actions, the beneficiaries who
may get diffused benefits at a low per-capita rate will remain inactive. In order for this
sort of regulatory legislation to be enacted, policy entrepreneurs, a group of people who
willingly devote their time and resources, will have to mobilize the latent public opinion
to promote their policy goal.

In a second body of literature, the regulatory policy process has been explained
by the distribution of influence among social interest groups. On one hand, the capture
theory or the producer-dominance model asserts that regulatory agencies are likely to
serve producers’ interests at the expense of consumers by restricting competition
(Bernstein 1955; Huntington 1952; Stigler 1971; McCaffrey 1982). According to Stigler
(1971), all firms seek to maximize profits, and profits can be increased if competition is
reduced. Government regulations that restrict entry by requiring a firm or a member of
an occupation to be licensed can be used for the firms’ benefits for two reasons. First,
since a small number of firms in any given industry expect to gain at a high per-capita
rate from regulation, the firms find it easier to organize to bear the costs of wielding I
political influence. Second, self-interested government officials seek to maximize their
votes or their wealth. The firms can supply these resources such as campaign
contributions and lucrative jobs.

On the other hand, the general group dominance theory argues that either
producer or consumer or other economic interests may become influential. According to
Peltzrnan (1976), government officials are vote-maxirnizers who arbitrate among
competing interests that seek to use government to redistribute resources. Politicians

will favor one or another interest as economic circumstances give greater urgency to the

needs of one or the other. Furthermore, politicians have to make compromises among
these competing interests to form large and heterogeneous coalition so that neither
adversary party gets all it wants.

Under what conditions, does one group emerge as influential at one point in time
while the other group becomes stronger at another point in time? There are three main
factors that may affect whether group interests can organize effectively (Rothenberg
1994: 26-32; Olson 1965; Dunleavy 1991). First, if the potential members who share
common interests are concentrated (or small in number) and the pool of the usable
resources is large, there is a greater chance that an interest group can be formed and
maintained. The small size of the membership and ample resources will minimize the
potential problem of one’s free-riding on others’ contributions so that the organization
can pursue the collective good-«the benefits from regulation. Second, the organizational
goal will reflect the voices from those who are interested in collective goods and those
who make large contributions. Any substantive gap between the organizational goal and
individual value will hurt the stability of the organization’s membership. Third, the
organizational capacities to provide politicians with valuable information and resources
are another crucial factor that determines the organization’s fate. Organizations can
prove their political value if they provide information about electoral preferences,
technological expertise, and other institutional actors’ preferences and behavior.
Resources that organizations can contribute include campaign money for elected officials
and the promise of jobs after leaving government.

In a third body of literature, the effect of institutional preferences and rules on

bureaucratic choices can be thought to explain regulatory policy implementation. The

main argument of this perspective is that regulatory policy or public policy in general
cannot be understood without systematic investigation of the nature and dynamics of
political institutions. More specifically, the argument is that institutional preferences and
rules determine the pattern of regulatory behavior. This approach collapses into three
subcategories: congress-centered, executive-centered, and multi-institutional
perspectives.

The congress-centered explanation emphasizes the effect of congressional rules,
structure, and electoral incentives on regulatory agency behavior. Numerous theoretical
models focus on statutory arrangements imposed by legislators on the agency. The main
argument is that statutes can create an institutional environment wherein the range of
permissible bureaucratic actions is defined and monitored through various procedural
requirements (F iorina 1982; McCubbins and Schwartz 1984; McCubbins, N011, and
Weingast 1987). The role of congressional committees and subcommittees has also
received attention (McCubbins 1985; Shepsle and Weingast 1987; Weingast and Moran
1983; Weingast 1984; Aberbach 1990; Knott and Hammond 2000). Especially when
committees are granted monopoly power in their jurisdiction and committees’ policy
views are distinctively different from the rest of the chamber, the congressional
committees can play a crucial role in shaping regulatory behavior.

The executive-centered explanation emphasizes the importance of the presidential
power to control agency leadership (Moe 1982, 1985, 1990). The presidential resources
for influence include formal powers such as appointment of agency heads, the OMB’s
review of agency budget and activities, and the president’s unilateral agenda-setting

power (Moe and Howell 1999; Cameron 2000; Howell 2003; Lewis 2003). The role of

the president in domestic policy areas has been considered to have grown as the
president’s organizational apparatus has expanded. The Executive Office of the
President has sufficient capacities to control policy administration by the executive
departments. In addition to this expansion of formal resources, the president can
capitalize on people’s mandate to exert informal influence on legislators (Kemell 1997).
His standing in the public can be used by the president as political capital to persuade
members of Congress to achieve presidential policy goals. In the era of partisan politics,
the president can play a role of policy magnet that can unite his party across branches
and between levels of the government.

The multi-institutional perspective advocates a broader model that includes all
key institutional actors such as the president, congressional actors, and the courts to
explain agency behavior (Moe 1985; Wood 1988; Wood and Waterman 1994; Scholz
and Wei 1986; Scholz, Twombly, and Headrick 1991). This perspective assumes that
the agency is able to respond simultaneously to discrete and even conflicting inputs from
those individual institutions. Recent development of theoretical models advances
propositions that clarify causal mechanisms of multi-institutional influence on agency
actions (Hammond and Miller 1987; Hammond and Knott 1996, 1999; Calvert,
McCubbins, and Weingast 1989; Epstein and O’Halloran 1999; Huber and Shipan 2002).
Macro rules that bind institutional actors can affect a regulatory agency’s behavior
through their immediate impacts on the likelihood of major policy change, the level of
discretion, and the range of politically-feasible bureaucratic policy choices.

Lastly, a substantial part of the variation in regulatory performance is attributable

to bureaucratic discretion and autonomy. Bureaucratic discretion and autonomy

originates from various factors. First, the limitations of formal and institutional control
mechanisms such as vague legislation, the opportunity costs of monitoring, and the
shortage of time and resources suffered by political supervisors leave detailed decisions
to bureaucrats (Dodd and Schott 1986). Second, the technological complexity of policy
problems and bureaucratic expertise reinforce the politicians’ temptations to delegate
policy authority to agency (Meier 1993). Third, the organizational adaptation of
regulatory agencies to fit idiosyncratic local environments leads to the undermining of
nationally-deterrnined policy directives (Lipsky 1980; Bardach 1977; Bardach and
Kagan 1982; Keiser and Soss 1998). Fourth, an organization’s culture, including the
agency’s past experiences and individual officials’ professional values, can result in
different regulatory outcomes (Downs 1967; Niskanen 1971, 1975; Wilson 1980, 1989;
Kelman 1981; Eisner and Meier 1990; Brehm and Gates 1997; Gonnley 1997; Carpenter
2001). In other words, the bureaucracy-centered explanation asserts that regulatory
behavior reflects various intra-bureaucracy factors due in part to incomplete supervision
by political institutions and in part to bureaucratic goals that may be different from those
of the political principals, augmented by the agency’s capacities to administer regulatory

programs in volatile environments.

1.1. Research Questions

Since regulatory politics and policies are complex social phenomena, one cannot easily
examine all possible combinations of causes and consequences. One must choose a
theoretical lens through which one can focus his or her research on some particular

aspects of the complex totality. This dissertation seeks to provide an explanation of the

linkage between institutions and a regulatory agency’s behavioral patterns in
occupational safety and health regulation.

Why are institutional preferences and rules the focal point in this research? The
first reason for focusing on institutional features is that despite the importance of the
relationship between representative institutions and bureaucratic organizations in a
modern democratic society, we still have limited knowledge of it. The question of
whether bureaucratic organizations comply with goals set by political institutions has
been at the center of academic discourse ever since Woodrow Wilson (1887) claimed
that the enterprise of administration should be separated from the normal process of
politics. Some scholars, without questioning what happens in the political environment,
delved into questions of efficient organization of administrative work involving division
and coordination (Gulick 1937), the distribution of authority among different ranks
inside bureaucratic organizations (Barnard 193 7), and the mode of administrative
decision-making (Simon 1947; March and Simon 195 8). On the other hand, others saw
government bureaucracies as saturated by so much politics that the politics-
administration dichotomy---the idea of purely administrative organizations-«was
considered impossible (Waldo 1948; Long 1949; Downs 1967). It was further asserted
that administration was well incorporated into the normal process of democracy so that
representative political institutions determined what bureaucrats would do on behalf of
the general public (Redford 1969).

Each of these views on the relationship between politics and administration left
its theoretical residue in contemporary debates on political control and bureaucratic

autonomy. The ‘political control’ perspective argues that public bureaucracies are

directed by elected leaders and that bureaucratic actions for the most part reflect
politicians’ wishes rather than the bureaucrats’ own predispositions (Weingast and
Moran 1983; McCubbins and Schwartz 1984; Moe 1982; Scholz and Wei 1986; Wood
1988; Wood and Waterman 1994). According to this view, bureaucrats provide the
public with policy services in ways that serve interests of the elected political leaders,
such as the president and congressional actors, who can use various control and oversight
tools. On the other hand, the ‘bureaucratic autonomy’ perspective asserts that various
intra-agency factors such as discretion, policy expertise, professional norms, and the
bureaucrats’ own policy preferences affect administrative decisions and that the impact
of these bureaucratic factors on policy outcomes is not outweighed by that of political
factors (Lipsky 1980; Rourke 1984; Wilson 1989; Meier 1993; Carpenter 2001).

Current discourse on the relationship between representative institutions and
bureaucratic organization remains inconclusive. The ‘political control’ perspective is
unable to clearly account for why a considerable part of the variation in regulatory
behavior has been explained by bureaucratic factors but not by institutional preferences.
The ‘bureaucratic autonomy’ perspective falters in the face of evidence that agencies’
decisions on the distribution of regulatory resources and the level of regulatory
stringency vary systematically with the preferences of elected leaders in political
institutions. In sum, we are left with insufficient understanding of the concurrence of
bureaucratic responsiveness to institutional preferences and autonomous bureaucratic
actions.

In this context, the first set of research questions of this dissertation is as follows:

How can we better understand bureaucratic autonomy amid a variety of
@olitical) institutional constraints in a democratic society? Does the presence of
bureaucratic autonomy negate the possibility of bureaucratic responsiveness? Can we
conceptualize these two seemingly incompatible processes (responsiveness and
autonomy) in an integrated fiamework?

The second reason for focusing on institutional effects on regulatory policies is
that although the literature has explored various effects of institutional preferences and
rules on regulatory policies, we are still left with insufficient and inconclusive empirical
knowledge about how these political institutional actors collectively affect agency
officials’ actions. The advancement of theoretical models has led our attention to the
nature and the mode of interactions among political institutions in exerting influence on
the agency. For instance, the US. Constitution and its separation of powers make
important policy change difficult in the absence of a joint majority of the chambers and
the president (Hammond and Miller 1987). Since a political principal can block
unilateral actions by other principals and since an important policy decision needs a
multilateral agreement, politically-feasible agency actions are constrained by joint
actions of multiple principals such as congressional actors and the president (Hammond
and Knott 1996, 1999; Calvert, McCubbins, and Weingast 1989; Epstein and O’Halloran
1999; Huber and Shipan 2002).

Past empirical research, however, has underestimated the importance of
multilateral actions of political institutions to regulatory agency actions. On one hand,
the importance of joint actions of political principals has been empirically examined to

explain various aspects of legislative decisions such as the statutory design of

10

bureaucratic discretion (Epstein and O’Halloran 1999, chapter 6; Huber and Shipan
2002, chapters 6 and 7), legislative productivity (Mayhew 1991; Krehbiel 1998), and
budgetary decisions (Brady and Volden 1998). However, none of them addresses the
effect of inter-principal interactions on how agency officials take actions to implement
policy. On the other hand, several bodies of empirical work that examined institutional
effects on agency actions left out inter-institutional relations. For instance, some studies
focus on the dyadic relationship between one principal and one agency while ignoring
the influence of other principals (Moe 1982, 1987; Weingast and Moran 1983). A host
of other empirical studies employ additive multi-institutional models that assume that
political control consists of individual institutions’ independent influences, thereby
ignoring how these political principals interact with each other to influence agency
actions (Scholz and Wei 1986; Scholz, Twombly, and Headrick 1991; Wood 1988,
1992)

Therefore, the second set of research questions is:

How do political institutions interact with each other to exert influence on
bureaucratic choices on regulatory policy alternatives? Who should be considered
pivotal among the institutional actors? How can we minimize the gap between formal

models and empirical research in the institutional study of the behavior of the regulatory

agency?
1.2. Research Plan

In the next chapter I provide a critical review of existing theories on the questions of

political control and bureaucratic autonomy and discuss limitations of past research such

11

as conceptual ambiguity, the omission of inter-institutional relations, and methodological
pitfalls. In chapter 3 I develop the theoretical framework. I present a multi-institutional
model of bureaucratic policy choices to derive propositions about how the nature and
mode of inter-principal interactions affect agency officials’ policy choices and how the
agency’s diverse preference distribution or policy bias may mediate these institutional
effects on agency ofiicials’ policy choices. Then I discuss which set of institutional
actors should be considered pivotal by focusing on three alternative views of veto
players: the majoritarian, the distributive politics, and the party government perspectives.
I propose that the relative importance of these alternative sets of veto players depends on
the characteristics of the policy issues at hand, such as salience and partisan polarization.
In chapter 4 I describe the formal structure of the Occupational Safety and Health
Act of 1970 and organizational and functional features of the US. Occupational Safety
and Health Administration. I also portray how the key institutional actors are involved
and intertwined to pursue their own policy goals through OSHA’s regulatory activities.
Chapters 5 and 6 include statistical analyses of the federal OSHA and state occupational
safety and health agencies’ inspection activities to test the hypotheses. Throughout these
two chapters, the process through which the national-level institutional preferences are
transmitted to federal and state agencies is empirically examined. Besides the main
hypotheses, whether and why federal and state agencies respond to different sets of the
national institutions are discussed. In Chapter 7 I discuss whether my research questions
have been successfully answered and what the implications of this research for the future

research are.

12

CHAPTER 2

POLITICAL CONTROLS AND BUREAUCRATIC AUTONOMY

Elected policymakers typically delegate policy implementation to bureaucratic agencies
in the modern administrative state. Bureaucratic agencies have been authorized to apply
statutory goals in individual cases, make adjudicatory decisions for disputed policy
cases, and establish administrative rules and policy standards. This delegation of
authority to unelected agency officials can take place under general conditions that
elected policymakers prefer to delegate legislative authority to administrative entities

(F iorina 1982; McCubbins 1985). First, the elected officials may create administrative
agencies to cope with technical complexity of policy problems. By creating
bureaucracies, specialized knowledge can be used to resolve the technical uncertainties
involving the impact of alternative policy actions. Second, elected policymakers use
delegation to minimize the political opportunity costs of directly dealing with policy
problems themselves. Third, politicians can reduce the political costs of making specific
decisions by shifting the responsibility to bureaucrats.

However, the delegation of policy authority to bureaucratic actors may result in
generic problems of principal-agent relationships. Due to the instability of political
coalitions on one hand and bureaucratic rigidity on the other hand, policy disagreements
among the political principals and bureaucratic agent can come into existence. In other
words, a stable contractual relationship between a political principal and a bureaucratic
agent is nearly impossible since the existing winning political coalition can be replaced
by other ones through electoral processes whereas bureaucratic organizations tend to

develop and institutionalize their own system for executing statutory mandates (Horn

13

 

1995). As a result, policy goals of bureaucracies may differ from those of political
principals. Then, the dilemma for politicians is that they must sacrifice some control to
capture the benefits of delegation. Even though the elected leaders can resolve technical
problems and decrease their opportunity costs by granting authority to bureaucratic
agencies, they still cannot avoid agency problems such as bureaucratic noncompliance
and information concealment, which can lead to agency actions that are not consistent
with what the politicians expected to obtain. Thus, politicians cannot avoid the trade-off
between “uncertainty about policy consequences” and “uncertainty about agency

behavior” (Bawn 1995: 63).

2.]. Political Controls

Although delegation of policy authority to agencies is in accordance with politicians’
interests, the politicians still want to maintain a degree of bureaucratic compliance.
Researchers who believe in the value of democratic control of the administrative
apparatus argue that political institutions and elected leaders should and can direct an
agency’s administrative decisions and performance (Redford 1969; Behn 2001). In a
democracy, representative institutions should determine on the general public’s behalf
what bureaucracy should pursue through its daily operations. Bureaucrats who can
obtain legitimacy for their use of authority only through delegation are expected to carry
out policy in accordance with the elected leaders’ wishes. The political control thesis
argues that political institutions send signals of their preferences through various
monitoring and incentive mechanisms to control bureaucratic behavior, and bureaucrats

respond to those political demands. The elected leaders are principals and bureaucrats

l4

are agents or servants; the political principals mandate policy goals, structure, and
resource levels, and control agency behavior. Researchers have examined the political
control or bureaucratic responsiveness processes in different ways. The dyadic approach
focuses on the relations between one institution, either Congress or the president, and an
agency; in contrast, the multi-institutional approach takes all those institutions into

account to explain agency behavior.

2.1.1. Legislative Controls
Legislators oversee the bureaucracy in an effort to promote policy objectives and claim
credits for promoting goals valued by their own constituencies (Fenno 1978; Mayhew
1974; Fiorina 1989; Keefe and Ogul 1997). According to F enno (1978), congressional
behavior is motivated by getting reelected, achieving influence, and making good public
policy. Mayhew (1974) and Fiorina (1989) argue that legislators, as “professional”
politicians, make self-interested policy choices to maximize their electoral credit and the
chances of reelection. Thus, legislators who almost always seek reelection and longer
careers in Congress want to oversee bureaucratic agencies when they see “a connection
between their own political lives and bureaucratic activity” (Keefe and Ogul 1997: 382).
Researchers have focused on two types of legislative controls (McCubbins and
Schwartz 1984; McCubbins, N011, and Weingast 1987). The first type is Congress’s
direct supervision through its investigative and oversight mechanisms. Congress can use
oversight power combined with a system of incentives and sanctions such as a promise
of continuous authorization and a threat of budget cuts. COngress possesses the power to

hold oversight hearings to monitor an agency’s performance and investigate its

15

as"

5”

v
0,~

ub-

01‘

.35

w\...

‘h;*
ale.

‘9"

e“.\
l

h."
I'Nu'.

'r‘p

Hp

.._c

 

wrongdoings. Congress also has the power to discontinue authorization of an agency as
a punishment of its undesired behavior. The effectiveness of this type of “police patrol”
mechanism rests on the premise that the mere existence of severe punishment of an
agency’s wrongdoing can change agency officials’ incentive systems, thereby increasing
the likelihood of bureaucratic compliance (Fiorina 1982). Even under conditions of
inactive oversight activity and the low probability of detecting undesirable agency
behavior, the agency will take actions in accordance with politicians’ wishes in fear of
the formal powers of authorization, appropriation, and appointment.

The other type of legislative controls is procedural controls. Congress can induce
agency actions within certain permissible bounds by using various administrative
procedures such as record keeping, information disclosure, notice-and-comments, and
citizen participation. These various procedural constraints through administrative due
process as codified by the Administrative Procedure Act of 1946 were originally
installed to impose uniform standards on the exercise of bureaucratic discretion.
Administrative due process can be used to serve legislators’ informational and electoral
interests (McCubbins, Noll, and Weingast 1987). Legislators can use administrative
procedures to minimize informational disadvantages in dealing with agencies. Notice-
and—comment rulemaking and freedom-of-information requirements can facilitate the
role of affected interests in alerting politicians to agency misdemeanors. This system of
“fire alarms” provides an efficient way to allocate resources to the most salient policy
areas. Secondly, legislators can “stack the deck” in favor of legislative interests by
creating decision-making criteria and opportunities for participation through which

agency officials are held responsible to the winning legislative coalition’s constituents.

l6

While both types of legislative controls-«direct monitoring and procedural
requirements-«can be employed by legislators, scholars have examined their relative
advantages. Some scholars assert that procedural controls are more efficient and
effective than direct supervision by helping reduce politicians’ opportunity costs of
monitoring and by shaping institutional environments that induce agency officials to
behave in predictable ways (McCubbins and Schwartz 1984; McCubbins, N011, and
Weingast 1987). But others contend that the choice between oversight and procedural
tools depends on the legislators’ level of knowledge about policy. Those who are
knowledgeable about a certain agency and policy (i.e., committee members) may prefer
oversight since they can concentrate their resources on potential problems, while those
who lack such expertise will prefer procedural controls (Bawn 1997).

The role of congressional committees in shaping regulatory behavior has received
special attention (McCubbins 1985; Shepsle and Weingast 1987; Weingast and Moran
1983; Miller and Moe 1983; Aberbach 1990). A body of literature has examined the
possibility of preference congruency between a committee (or subcommittee) and an
agency (Huntington 1952; Weingast 1984; Knott and Hammond 2000). The importance
of a committee’s role rests on the premise that Congress consists of loosely-coupled and
functionally-specialized committees (or subcommittees); these congressional subunits
exert legislative monopoly power over issues in their jurisdictions. These quasi-
independent committees can protect their jurisdictions over administrative agencies by
forming strong policy coalitions or “iron triangles” with interest groups and the agencies.
The policy triangle produces policy outcomes based on mutual benefits of the legislators,

constituency groups, and the agency. Congress as a whole distributes policy benefits

17

through these policy coalitions in various areas to serve local and industrial interests.
Legislative activities for distributive programs or “pork-barrel” policy certainly
contribute to legislators’ home districts’ wellbeing and help secure greater political

supports (Fenno 1978; F iorina 1989).

2.1.2. Presidential Controls
The unbalanced attention of the literature on legislative delegation and control has been
criticized for ignoring the presidential role in directing agency behavior (Moe 1982, 1985,
1987). The main argument of the presidential control perspective is that congressional
dominance is not true since presidents are systematically ignored by the congressional
dominance models even though they often play a major role in shaping agency behavior.
Moe (1985: 1101) expresses his beliefs about the ineffectiveness of congressional
control, due mainly to the Congress’s complex web of multiple decision-making nodes:
Yet these [congressional] powers are wielded by various committees,
subcommittees, and chairs in both Houses. Thus, some congressional actors may
be highly interested in influencing the NLRB, whereas others choose to focus
their resources elsewhere; some of those interested in influence may be quite
conservative, others quite liberal; and particularly as the cast of characters
changes and the commitment of actors in the various institutional bodies ebbs and
flows, serious control efforts may shift from one committee to the next and back
again over time. Within this complicated context of competing principals, the
NLRB is faced with conflicting demands and pressures, but also with
Opportunities to avoid compliance by shifting responsibility and playing
congressional actors of against one another. (emphasis added)
Recent theoretical models highlight the president’s superior ability to take
unilateral actions, which may offer opportunities for him to exert the greatest influence

when interacting with other institutional actors such as Congress and the courts

18

(Cameron 2000; Howell 2003; Lewis 2003). For instance, Howell (2003: 14-15)
describes the president’s unilateral powers as follows:
The most important is that the president moves policy first and thereby places
upon Congress and the courts the burden of revising a new political
landscape. . . .If they choose not to retaliate, either by passing a law or ruling
against the president, then president’s order stands. Only by taking (or credibly
threatening to take) positive action can either adjoining institution limit the
president’s unilateral powers. ....The second important feature of unilateral power

is that the president acts alone. There is no need to rally majorities, compromise
with adversaries, or wait for some interest group to bring a case to court.

(emphasis added)

The US. presidency also has improved its policymaking capacities. The role of
the president in policymaking has grown in the last decades as its functions have been
expanded and institutionalized. Since the creation of the Executive Office of the
President (EOP) in 1939, the presidential staff bureaucracy has grown in size,
specialization, and responsibilities. The development of the institutionalized presidency
has increased the presidential power (Edwards and Wayne 1999). Thus, many agree that
the presidency as an institution has become an independent policymaking powerhouse.
Some scholars even coined the term “Executive Hegemony” to emphasize the president’s
enhanced power in both administrative and legislative arenas (Spitzer 1993).

Article II of the US. Constitution provides the president with a broad range of
authority over the administration of government, the task of executing the law, and
oversight of the executive departments. As the chief executive, the president can employ
two strategies to control agencies: personnel management to staff agencies with loyal
executives, and centralized fiscal management and supervision via the BOP (West 1995:
77-83). The most powerful presidential appointments fall under the “executive

schedule” wherein the president can fill numerous positions such as department

19

'1‘)“
55..
'

24“}.
‘h\ r.
s

.4....

'...\ .

TI"

\
n-..’

s

IA'y‘

ts».

leadership, agency and bureau heads, and commissions with loyal executives. Over
these political appointees, the president is granted unilateral removal power. In addition,
the Civil Service Reform Act of 1978 enhanced presidential power over the US. civil
service system by allowing the president to fill up to 10 percent of the jobs in the Senior
Executive Service by political appointment. Next, the president’s fiscal powers can be
used for centralized management. The president via the BOP, especially the Office of
Management and Budget (OMB), can establish presidential policy priorities among
agency programs and pursue the administration’s objectives by auditing and evaluating
agency programs during executive budget preparation. Moreover, the president has
discretionary controls over the use of appropriated money such as the authority to
transfer budgeted funds within and among agencies and the authority of impoundment
including deferral and rescission of budgeted funds.

Although the president has these weapons at his disposal, the important question
is whether the president or his staff organizations are seriously committed to bringing the
president’s policy preferences to bear. There is some evidence of increased interest of
the president and the BOP in monitoring agency policy and programs. For instance, the
president’s administrative involvement, through policy review programs, has been on the
rise (West 1995: 85-90). Richard Nixon’s executive order began the “Quality of Life
Review” program, which required all proposed environmental regulations of the EPA to
be submitted for comment to other agencies. Gerald Ford’s Executive Order 11821 (as
amended by ED. 11949) expanded his predecessor’s program and required that all major
regulations be assessed by an Inflation Impact Statement (118) which was subject to

review by the Council on Wage and Price Stability (COWPS) in the BOP. Jimmy Carter

20

further expanded the review program by requiring all major rules to be justified by cost-
benefit analyses (Regulatory Analyses). Ronald Reagan’s Executive Order 12291
required cost-benefit analysis and the Office of Information and Regulatory Affairs
(OIRA) review for all regulatory proposals. This OIRA review program was kept
through the Bush (senior) and Clinton administrations.

In addition to these formal powers, the president can use informal channels of
influence on the national policymaking process. For instance, Moe (1982: 201) remarks
that “Many individuals within the commissions may give great weight to the president’s
policy positions not because he wields rewards and sanctions, but simply because he
holds the office of president and, in their minds, has a right to expect compliance.” This
sense of compliance with presidential wills has also been observed in the legislative
arena. The president often uses his resources to persuade legislators to support his own
policy goals. The president’s standing in the public can be transformed into the
president’s political capital to influence individual legislators’ voting behavior (Kemell
1997). The president as a party leader can also mobilize broad partisan support for his
own policy agenda to achieve legislative successes in domestic policy areas (Bond and

Fleisher 1990).

2.1.3. Multi-Institutional Controls
The dyadic approach---legislative and presidential control---has been criticized on the
grounds that it focuses on one particular institution while ignoring other institutions and

that this imbalanced attention may yield only biased inference about institutional

21

determinants of regulatory behavior. For instance, Moe (1985: 1095) expresses this
dissatisfaction as follows:

It is plain from decades of research on bureaucratic politics that public agencies

are anchored in networks of relationships with executives, legislative committee,

and constituency groups. . . ..Although this is no secret, popular models of
regulation as well as quantitative empirical work have tended to focus only on
very small parts of the whole---in the former case for reasons for clarity and
mathematical tractability, and in the latter because of data collection and
measurement problems (and because they are often guided by these same
models). . . .It is important to remember that [these research strategies] threaten to
yield biased inferences about the causes of regulatory behavior. They clearly
omit factors whose causal effects may overwhelm or distort the “special”
relationships on which they singularly focus.

In contrast, theoretical and empirical multi-institutional models have also been
advanced. The key argument is that political institutions should be considered together
to better understand institutional influence on bureaucratic behavior. Theoretical multi-
institutional models focus on how political institutions interact with each other to
influence agency actions (Hammond and Miller 1987; Calvert, McCubbins, and
Weingast 1989; Hammond and Knott 1996, 1999; Epstein and O’Halloran 1999; Huber
and Shipan 2002). These multi-institutional models build on the macro rules that bind
separate institutions together, such as bicameralism and the presidential veto. Hammond
and Miller (1987) show why and how the US. Constitution and its separation of powers
induce policy stability. Under the system of bicameralism and the executive veto, policy
change is difficult without the agreement of a joint majority of the chambers and the
president. This is because the set of undominated policies---the core---will not be
decreasing in size with the added veto players and their dissimilar policy preferences.

This finding about the nature of multi-institutional policymaking opens the door

to a new area of research: how the interactions among these institutions affect

22

bureaucracy. Since major policies are made by collective efforts of the institutional
actors, bureaucratic choices should be bounded by the joint actions of the multiple
institutions. Calvert, McCubbins, and Weingast (1989: 589) make this point clear:

[T]he actual [bureaucratic] choice of policy is traceable not to bureaucratic

preferences but to the preferences of legislative and executive politicians. ...Even

though the agency may be the sole active decisionmaker, policy outcomes are
traceable to the preferences of all institutions and to the constitutional process in
which they act.

Hammond and Knott (1996: 163) also provide an explicit view:

In our view, control of the bureaucracy is function of the interactions of the

president and Congress. . . .Whatever the extent of constraints on an agency, one

cannot single out any one institution as primarily responsible for these
constraints. Instead, control of the bureaucracy must be seen as a systematic
matter: the president, House, and Senate collectively control the bureaucracy.

(emphasis in original)

This multi-institutional framework has been employed by a host of empirical
studies on regulatory policy, especially economic, occupational safety and health, and
environment protection regulation (Moe 1985; Scholz and Wei 1986; Scholz et a1. 1991;
Wood 1988, 1992; Wood and Waterman 1994; Shipan 2004; Whitford 2005). These
empirical studies develop an additive multi-institutional model, wherein political
influence is assumed to consist of individual and discrete institutional effects. Wood and
Waterman (1994) make a clear remark on this progress: “the simple dyadic images
depicted by past research should now give way to an image of bureaucracies as
continually adapting to multiple, concurrent, and diverse stimuli” (101). Their study
examined various executive sources of political influence, including a new presidential

administration, presidential appointment of a new agency head, executive order for

reorganization and the tone of presidential statement. For the part of Congress, budget

23

 

 

x.‘ n

..Ir -,

appropriation, congressional oversight hearings, and the enactment of new legislation
were considered.

This multi-institutional research on regulatory agencies provides empirical
evidence for one of the main arguments advanced by the multiple-principal fi'arnework:
all key institutional actors should be considered together to explain bureaucratic
decisions and behavior. As assumed by positive theories, various centralized formal
controls that can be used by the president and congressional actors over an agency have
been examined by the multi-institutional research. By adding political institutions to the
explanatory equation, the multi-institutional approach at least overcomes the omitted
variable problems that plagued the dyadic approach.

In addition to its inclusion of multiple institutions, the multi-institutional research
program has discovered causal mechanisms that link institutional preferences and agency
behavior. First, empirical studies systematized the interactions between political
institutions and government bureaucracies as “stimulus (signal)-and-response” relations
(Wood 1988, 1992; Scholz and Wei 1986, Scholz et a1. 1991; Wood and Waterman
1994). Wood and Waterman (1994) depict bureaucracy as an adaptive entity responding
concurrently to stimuli of different types. They identify three different stimulus types:
discrete, event, and tonal. “Discrete” events are stimuli that occur just once but are
expected to have effects that last for some time. The appointment of a new agency head,
a large one-year budget cut, the enactment of new enabling act, or a landmark judicial
ruling can be considered this type of stimulus. “Event” processes are sequences of
discrete event stimuli that pass through time. This type of stimuli includes a set of tirne-

ordered budgets, all congressional hearings, and all relevant rulings by the courts.

24

I
a
mo [

L. z.

arg‘
..liH

MC".
5...

I. 4‘
H_‘_

I
'n-o

Ira, .
r1.

.3.
i.

I .
v...

“Tonal” stimuli are those that develop gradually over time rather than being manifested
through each discrete event, such as gradual change in the public mood and the news
media attention.

Bureaucracy in turn responds to these multiple stimuli. External stimuli are
distributed across time and so are bureaucratic responses. Bureaucratic response to
stimulus can take place instantly (zero-order), occur some time later (lagged), or be
distributed across time. The difference in bureaucratic response can be accounted for by
three factors: technology, rationality, and politics. Technical factors include intra-
organizational dependence and bureaucratic inertia that tend to increase response time.
Bounded rationality of political and bureaucratic actors may generate ambiguous, weak,
and conflicting signals and responses. Lastly, divergent interests among politicians and
bureaucrats may lead to slow and incremental bureaucratic responses.

Another important advancement of the multi-institutional research program is
that it provides systematic knowledge about various channels through which the top-
level politicians’ preferences reach front-line officials at the bottom ranks of government
hierarchy. First, centralized hierarchical control is a part of the causal mechanism by
which the national politicians’ policy preferences influence lower-level agency officials
(Moe 1985). According to this view, it is unlikely that agency officials at lower ranks of
the government bureaucracy take their cues directly from politicians, given the complex
structural and incentive system of bureaucratic organizations. Those political cues can
be delivered to the rank and file of the bureaucracy primarily through the mediation of
top agency officials. Moe (1985) describes this ‘two-tiered, strictly hierarchic system’:

“Political authorities attempt to control the behavior of their immediate subordinates (the

25

[NLR] Board), and the Board in turn attempts to control its own organizational
subordinates, the staff” (1100). The hierarchical control channel, thus, rests on the
effectiveness of supervision-compliance relations within bureaucratic organizations.
Ideology or policy preferences embedded in the organizational decisions at the top of the
bureaucratic hierarchy provide signals to lower-level bureaucrats regarding daily
operations.

The flow of national political influence through local channels to field
bureaucrats has also been examined by researchers (Scholz and Wei 1986; Scholz,
Twombly, and Headrick 1991). Scholz and Wei (1986) provide a very clear view of this:

Once the national policy is set, a programmatic agency, however, is likely to

respond to the more subtle concerns of congressmen for their particular districts

by initiating more intense enforcement efforts in areas and industries where
congressmen hope to maintain strong labor support and by developing more
cooperative enforcement programs in areas and industries where business

backing is important to congressmen (1252).

The impact of local channels on street-level bureaucratic behavior rests on the
assumptions that top agency officials want to maintain support from particular elected
officials and that field officials are inclined to capitalize on elected officials’ willingness
to provide local leadership. To the extent that required local resources are more
problematic than central resources, the effect of local channels can even outweigh that of
centralized channels via the formal hierarchy.

Scholz, Twombly, and Headrick (1991) extend this bottom-up explanation fiirther
by focusing on the influence of local partisan activities on OSHA enforcement as
follows:

[P]artisan activities of elected officials and their electoral coalitions in the local

arena provide important systematic influences on bureaucratic behavior,
particularly in circumstances in which conflict reduces the ability of central

26

institutions to exercise political control. We emphasize the role of nonlegislative

or “home-style” activities of legislators and their support coalitions in electoral

districts of local, state, and federal legislatures (830).

As the authors argue, the dependence of a regulatory agency upon local political
support and resources can divert implementation from the national policy goals set by
central leaders, thereby generating variations across regions and localities. However, it
may be also true that the national elected officials can exert influence through these local
channels on the street-level bureaucrats to pursue the national policy goals. The national
politicians through their home-style activities can get involved in ‘daily battles’ of
implementers in the field. That is, the prominent elected officials are not just distant and

minor participants of bureaucratic operations. The national-level political signals may be

strong enough to reach the bottom ranks of bureaucratic hierarchies.

2.2. Bureaucratic Autonomy

The principal-agent framework has been used by most of the institution-based
explanations of bureaucratic behavior to rediscover the importance of democratic
hierarchies in shaping bureaucratic behavior. The thrust of this view is that democratic
institutions can control what unelected officials in governmental agencies do to make
bureaucratic outcomes consistent with what the general public may wish. However, the
principal-agency relationships, especially the stable and ordered hierarchical relations
among political institutions and bureaucratic agencies, have come under suspicion. This
criticism emphasizes the irnperviousness of the “fourth branch of government” to
political control. For instance, Rourke (1986: ix-x) succinctly stated this view as follows:

[T]he actual role of bureaucrats may deviate widely from their theoretical role as
servants of policy. Bureaucrats may help to create as well as to carry out the

27

public will by generating new policy initiatives which the public accepts. In

some areas of policy their expertise may even entitle them to act at their own

discretion, limited only by the vaguest set of guidelines laid down by the White

House or Congresses. So, as is often the case with the master-servant

relationship, the activities of some bureaucratic servants may very much resemble

those of a master.

From this view, the external checks by political institutions and elected leaders
over specific policy areas and agencies are considered at best superficial and perfunctory.
Capitalizing on the ineffectiveness of political controls and bureaucratic insulation from
political institutions, bureaucrats may use their discretion to pursue their own goals that
may not be in accordance with those of the principals. According to Rourke (1984),
bureaucratic ‘power’ rises from four factors: expertise, constituency, vitality, and
leadership. Bureaucratic expertise confers power through superior knowledge of a
problem or policy. Constituency goes to the core of political relationships through the
ability of bureaucracy to mobilize political support or curb political opposition. Vitality
refers to the professional commitment of bureaucratic personnel to job, program, and
organization. Leadership will bring greater expertise to an organization, effectively
mobilizing constituencies, and improving personnel commitment to make the
organization more vital.

Similarly, Carpenter (2001) emphasizes the importance of bureaucracy’s
entrepreneurial efforts to define the functions for an organization, to mobilize external
supports and resources, to maintain the highest-level of expert knowledge, and to defend
the mission and goals of an agency. From this perspective, bureaucratic autonomy is a
result of political struggle of innovative bureaucratic organizations in the jungle of

politics. Carpenter makes a very strong point about when and how we can observe

bureaucratic autonomy. According to him, bureaucratic autonomy occurs when

28

bureaucrats take sustained patterns of “actions consistent with their own wishes, actions
to which politicians and organized interests defer even though they would prefer that
other actions (or no action at all) be taken” (4). He then suggests that the general
conditions under which bureaucratic autonomy emerges are:

0 Autonomous bureaucracies are politically diflerentiated from the actors who seek
to control them. They have unique preferences, interests, and ideologies which
diverge from those of politicians and organized interests.

0 Bureaucratic autonomy requires the development of unique organizational
capacities---capacities to analyze, to create new programs, to solve problems, to
plan, to administer programs with efficiency, and to ward off corruption.
Autonomous agencies must have the ability to act upon their unique preferences
with efficacy and to innovate. They must have bureaucratic entrepreneurs.

o Bureaucratic autonomy requires political legitimacy, or strong organizational
reputations embedded in an independent power base. Autonomy first requires
demonstrated capacity, the belief by political authorities and citizens that
agencies can provide benefits, plans, and solutions to national problems found
nowhere else in the regime. These beliefs must also be grounded in multiple
networks through which agency entrepreneurs can build program coalitions
around the policies they favor. (14: emphasis in original)

Some scholars argue that bureaucratic interests are not necessarily self-serving.
Instead, bureaucracies may function as a “representative” institution where the
competing interests of diverse social groups can be compromised and a stable set of
solutions can be pursued without the direct mediation of the elected leaders and political
institutions. This follows from the contention that “political” questions differ from
“administrative” questions only in who decides them, not in differences in content
(Meier 1993).

These observations may accurately reflect what autonomous bureaucracies may

look like. Nonetheless, while these could be a systematic description of the

characteristics of autonomous bureaucratic organizations, they may not be causal factors.

29

So the question is: due to what factors will bureaucracies be able to pursue their own
goals, develop their own policy capacities, obtain external resources and support, and

even represent the interests of various social groups?

2.2.1. Agent Problems

In the framework of the principal-agent model, bureaucratic autonomy has been
attributed mainly to agent “problems.” The major problem is asymmetric information
combined with conflict of interest among political principals and bureaucratic agents.
Problems of asymmetric information include hidden preferences (adverse selection),
hidden actions (moral hazard), and policy uncertainty (bureaucratic expertise) (Miller
1992; Moe 1984). A bureaucratic agent can misrepresent his or her true policy
preferences to a principal. Given limited information about the agent’s quality and
worldview, a principal may choose a wrong person for a bureaucratic position. The
mistakenly chosen person could lack the capacity to carry out the task or perform the
assigned job in the direction that the principal has not expected. Moral hazard can occur
when the principle cannot obtain complete information about her subordinate’s behavior.
The bureaucratic agent can conceal his performance simply by cheating or by free-riding
on his colleagues’ team efforts. Policy uncertainty refers to the impact of unexpected
external shocks on policy outcomes about which only the policy irnplementer can know
in detail. The principal may obtain at best inaccurate (or probabilistic) clues about what

happens in the real world.

30

These agent problems have been considered in the context of the multi-
institutional framework. For instance, Calvert, McCubbins, and Weingast (1989: 595)
argue that:

Imperfect information could arise at many points in the process. The elected
authorities might not know exactly the true preferences of the agent. Indeed,
policymaking by an agency often starts with the gathering of information about
the policy problem to be addressed, information presumably not known to the
elected authorities at the time of appointment. It may not even be clear in
advance what the ultimate policy alternatives will be. . . .Any slippage between the
expectations of the appointers and the preferences of the appointee creates the
possibility that the agent’s preferences will have an independent effect on the
ultimate policy choice.

In a similar vein, Moe (1985: 1101) describes the possibility of limited political
control over the NLRB decisions as follows:
Added to this is the ever-present information asymmetry: the NLRB knows far
more about the content and direction of its own behavior, from the lowest-level
staff investigatory decisions to formal Board decisions, than these political
authorities can hope to ascertain, even should they adopt costly and extensive
monitoring methods. . .. Because the Board has its own interests to pursue, both as
an organization (budget, slack, autonomy) and as a collection of individuals
(career, ideology), the authorities can expect partial compliance at best.
Regarding the consequence of vague procedural legislation and centralized
monitoring system’s limited effects, Scholz et al. (1991: 832) also remark on behavior of
OSHA inspectors as follows:
Observers of regulatory enforcement consistently comment on the broad
discretion that each inspector must deal with in determining how closely to
scrutinize a given establishment, whether observed conditions constitute a
violation, and whether a violation is intentional and should be cited or
“accidental” and should be dealt with informally.
A stylized case of bureaucratic noncompliance was provided by Wood’s study on

EPA (1988). He asserts that a bureaucratic agency can effectively resist political

pressure that is in conflict with its policy preference even during a period of limited and

31

resources and political constraints. Hierarchical control is effective only when there is a
consensus among political principals and bureaucratic agency. Wood (1988: 227-8)
concludes that:
[C]onsiderations of hierarchy, although important, have obvious limitations for
explaining outcomes in some implementation policy processes. For clean air, a
principal-agent model would predict that the Reagan administration, given the
most Republican Congress since the 19505 and extraordinary political influence,
should have been able to shift the preferences of the environmental
bureaucracy. . . .But in the end EPA’s revealed preferences were completely
opposite from what the model predicted. . . .[E]nforcements were pursued more
vigorously than at any time in the agency history and in a manner inconsistent
with the ideological dispositions of elected political institutions.
2.2.2. Complexities of Hierarchy
Even if there is no agent problems in the relationship between political institutions and
political appointees in bureaucratic agencies so that top agency officials are under tight
supervision of or in complete consensus with political authorities, there are still
possibilities that the hierarchical control inside the bureaucratic organizations are
ineffectual. This is due to the complexity of bureaucratic organizations and many other
latent problems of internal control in hierarchies. This phenomenon can result especially
when the superior-subordinate relations in a formal organization are unstable. In the first
place, the weakness of superior-subordinate relations can be attributable to various agent
problems such as adverse selection (misrepresentation of human resource quality), moral
hazard (exploitation of informational advantage in part of operational units), and the
inseparability of team-based activities inside the agency (Miller 1992). These factors
tend to weaken the influence that the superior can exert on the subordinate.

A more fundamental issue may be whether hierarchical organizational structures

guarantee clear-cut lines of command. In fact, the non-linearity in organizational

32

' .
1.3".
955;.

m:

I r
LA

I.’

lie“

..
5”3.‘
Hub

’3‘. .

.b.u.

’WI-w

A‘Vclu.

\-]~”
T “Jr

the

‘i'o‘.

,1
"a.“
i

.
It.
\‘I-

\“‘

 

a
.-
I
.I.

“slit.

decision-making has been examined in an extensive body of the literature (Barnard 1938;
Simon 1946; March and Simon 1958; Lindblom 1959; Landau 1969; Allison 1971;
Cohen, March, and Olson 1972; Hammond and Miller 1985; Hammond 1986; Heimann
1993; Brehm and Gates 1997). There are several factors that can be thought to hamper
effective hierarchical controls. Unclear lines of command in a formal organization may
result from the conflict between different sources of influence such as expertise and
formal authority (Hammond and Miller 1985). More specifically, there are various
situations in which the formal-superior cannot control the expert-subordinate even in
hierarchical organizations. The non-linearity of organizational decision-making also
may take the form of bilateral relationships between the superior and the subordinate
(Barnard 1937; Brehm and Gates 1997). The bilateral interactions may negate the strict
command-and-compliance relations by emphasizing mutual adjustment and reciprocal
influences between different ranks of a hierarchy (Landau 1969). F urtherrnore, when
organizational goals are too ambiguous and the level of organizational unity is declining,
subunits of the organization are likely to engage in competition for organizational
resources and influences to promote their own parochial interests rather than
organizational goals (Lindblom 1959; Allison 1971).

These phenomena may stem from the limited human capacity or bounded
rationality (Simon 1946). Organization may not be considered as a unitary rational actor
but a cooperative human effort to increase human capacities to cope with complex
environments. Although organization can improve the limited human capacity to some
degree, organization itself also may trigger problems; organized activities can be

characterized more accurately by satisfaction, mutual adjustments, the mobilization of

33

selective and sequential attention, and redundancy (March and Simon 1956; Landau
1969). The irony is that within hierarchical systems, the interactions among elements of
the same system can be much more complex than the interactions between different
systems. Due to this complexity, bureaucratic organizations may not be a reliable

control mechanism.

2.2.3. Multilateral Institutional Relations
The essence of the multi-institutional explanation of bureaucratic behavior is that the acts
of the institutional actors are so intertwined that political influence is exerted as a
concerted effort. However, due to their divergent policy preferences and goals, it may be
difficult for these institutional actors to send consistent signals to the agency. In the
separation of powers system, multiple principals may compete for greater influence on
the agency. For instance, Moe (1984: 768) remarks that:
Bureaus are “partial agents” of various governmental principals, without being
under the complete authority of any one in particular, and without any common
understanding of how authority is legitimately divided among the competing
principals. . ..American politics is, by its nature, a context of competitive
principals, it is hardly paradoxical that politicians impose constraints “on
themselves”. In fact, politicians impose constraints on one another in a
competitive effort to see to it that their own interests are protected from the
intrusions of politician-opponents. This is rational for individual politicians and
groups of politicians, but the net result is that politicians in general have a more
difficult time controlling the bureaucracy. This can only tend to strengthen the
foundations of bureaucratic autonomy.
What is the impact of policy conflict among the political principals on
bureaucratic autonomy? There are some theoretical and empirical answers for this

question. Several bodies of empirical work on bureaucratic behavior in the area of

regulatory policy have found that inter-principal competition and its inconsistent signals

34

result in bureaucratic noncompliance. For instance, as Moe (1985) put it, “[A]ll political
authorities have formidable bases for influencing the NLRB in desired directions, but
compliance is nonetheless problematic, which results partly from institutional conditions
that they can do little about: the ambiguity and competitive arrangement of governmental
authority” (1102). Similarly, Kelman (1981) observes that “When the president and the
relevant congressional committee give an agency different signals, as in the OSHA case,
the bureaucrat who is trying to be responsive is in a quandary. It also creates
opportunities for agencies to play both against each other” (105). Bureaucratic
noncompliance can also take place as a symbolic response to political demands while
pursuing bureaucratic goals. Scholz and Wei (1986) remark in their analysis of OSHA
enforcement activities that “[B]ureaucrats will respond to political demands by changing
lower-cost, ‘symbolic’ output that may help generate the desired political support, even
if it may have no effect on accidents, but will respond to task factors with ‘instrumental’
output that agency professionals consider to be more likely to affect outcomes, even
though it is also more costly” (1255-6). Tsebelis (2002) also provides some empirical
conjectures on the relationship between institutional fragmentation and bureaucratic
autonomy. His analysis of the independence of the central bank in EU countries shows
that institutional fragmentation as measured by the number of institutional and partisan
veto players tends to increase the Central Bank Independence index scores.

Although these descriptions reveal some patterns of bureaucratic response in
conflict-ridden political environments, they do not provide a clear causal mechanism for
the phenomenon. Theoretical developments on this subject have begun only in recent

years. Hammond and Knott (1996, 1999) have developed a multi-institutional model of

35

bureaucratic autonomy. According to Hammond and Knott, bureaucratic autonomy can
arise from institutional rules, such as executive veto, bicameralism, committee gate-
keeping and judicial review, and the preference configuration of these institutions.
Based on a veto-player game which approximates the decision rules that require a
multilateral agreement, they demonstrate the existence of a set of equilibrium policies,
which may be called the core. Knowing that any policy choice inside the core cannot be
replaced by any decisive coalition of veto players, a farsighted strategic agency manager
will choose the best policy (i.e., one closest to her ideal point) from the set of equilibrium
policies. As Hammond (2003) put it, “The existence of a set of equilibrium policies, and
change from one equilibrium policy to another, without fear that its chosen policy will be
upset by any decisive coalition of elected officials” indicates the degree of bureaucratic
autonomy (76). A larger core will give the agency head substantial room for “unilateral
policy change.” Since the core gets larger with greater divergence among veto players,
policy conflict or institutional fragmentation should lead to bureaucratic autonomy.
While Hammond and Knott present the institutional and political conditions
under which bureaucratic agencies can take unilateral choices, Epstein and O’Halloran
(1999) and Huber and Shipan (2002) focus on institutional and political factors that
determine the level of bureaucratic discretion in legislations. Both models begin with the
question of why there are considerable variations in statutory restrictions on bureaucratic
discretion. But they come to different conclusions. Epstein and O’Halloran (1999),
based on a gridlock interval analysis, propose that while policy conflicts between the
congressional committee and the chamber floor will lead legislators to write less detailed

law (more discretion for bureaucrats), policy conflict between branches, the president

36

and Congress, will lead to more detailed law (less discretion for bureaucrats). But the
former impact will diminish as the latter impact increases. In other words, heightened
inter-branch conflict will generally decrease bureaucratic discretion, which is not
consistent with the results of the empirical research already mentioned.

Huber and Shipan (2002) attribute variations in legislative delegation to political
and institutional factors: whether the legislators possess information and time to write
detailed legislation (legislative capacity), to what degree politicians distrust the agency
(policy conflict), and whether legislators have reliable non-statutory mechanisms that can
induce desirable policy outcomes (non-statutory factor). They examine the independent
impact of institutional arrangements such as the executive veto and bicameralism by
holding these factors constant. In a model of the executive veto, a legislator can avoid
the presidential veto only if she writes a high-discretion bill allowing the bureaucrat to
implement a policy preferred by the president to the status quo. In a model of
bicameralism, due to increasing bargaining costs, preference divergence between
chambers will inhibit a restrictive bill that may secure only one charnber’s interest.
Therefore, overall policy conflict between branches or between chambers is expected to

increase bureaucratic discretion.

2.3. Limitations of Past Research

Although an extensive body of the literature has examined various effects of institutional
preferences and rules on regulatory agency actions, there are several weaknesses that
should be addressed further. In this section, I discuss major limitations of this previous

research, such as conceptual ambiguity of the notion of bureaucratic autonomy,

37

underestimation of the importance of inter-institutional relations, and methodological

pitfalls.

2.3.1. Conceptual Ambiguity: ‘Independence,’ ‘Noncompliance,’ and ‘Reciprocity’
The notion of bureaucratic autonomy as a behavioral pattern remains ambiguous in the
literature. Researchers tend to describe an autonomous bureaucracy in terms of their
own images of bureaucracy such as a politicized institution (Rourke 1984; Meier 1993),
a policy entrepreneur (Carpenter 2001), bureaucratic noncompliance (Wood 1988), or an
agency’s responsiveness to both institutional and task factors (Scholz and Wei 1986;
Scholz, Twombly, and Headrick 1991).

The view of bureaucracy as a politicized institution has a long history. Norton
Long (1949: 250) argued that “the lifeblood of administration is power.” Bureaucracies
may cultivate their own bases of support to maintain their status in the broader political
system. This view was echoed by later scholarly work on contemporary bureaucracies.
Rourke (1984) and Meier (1993), for instance, contend that bureaucracies can capitalize
on their better knowledge of policy problems and implementation technologies, cohesive
professionalism, effective leadership to bring resources and authority to the organization,
and capacities to mobilize external support to curb political opposition. Moreover,
bureaucracies can represent diverse social interests without the mediation of political
institutions. When viewed as such politicized institutions, bureaucracies can be
described as “the fourth branch” of government that functions as if they are by

themselves legitimate governance institutions.

38

The biggest problem with this view is that the major principles of a democratic
polity are violated. In a democratic society, political power can be legitimate only when
it comes out of the general public’s will, which is expressed and delivered through
electoral processes. Only elected officials in the representative institutions have electoral
mandates on which they are expected to base their policymaking. Unelected officials,
however, can exert autonomy only affer they are granted or delegated authority by the
representative institutions. Bureaucratic decision-making thus should be considered in
the context of the broader political system. If bureaucrats exert independent influence on
policy outcomes without the consent of the elected officials, they do mischief to the
democratic principles. Furthermore, if bureaucracies are independent powerhouses, why
should we observe bureaucratic responsiveness to politicians’ wishes? There has been
ample evidence that bureaucrats consider institutional preferences when they carry out
policy decisions made by the elected officials. In other words, bureaucrats cannot do
whatever they want to do due to political constraints on bureaucratic actions.

The notion of bureaucratic noncompliance has also been used to describe a
particular behavioral pattern that does not match political institutions’ policy preferences
(Wood 1988; Wood and Waterman 1994; Eisner and Meier 1990). For instance, after
observing bureaucratic noncompliance in EPA regulatory activities, Wood and
Waterman (1994: 126) argue that:

Bureaucratic resistance to duly elected politicians may actually sometimes be

more consistent with democracy and public preferences than bureaucratic

responsiveness may be.

However, observing bureaucratic noncompliance with policy directions is not

sufficient evidence for autonomous bureaucratic choices. It remains unclear whether the

39

bureaucrats intended to perform assigned tasks in the opposite direction or they just
failed to accomplish policy goals due to bureaucratic inertia or incompetence. In other
words, the reasons why we oftentimes observe bureaucrats doing things differently from
what they are told to do remain puzzling. Is it bureaucratic resistance or bureaucratic
failure?

Bureaucratic responses to both political and bureaucratic factors have often been
used as evidence for bureaucratic capacity to find equilibrium solutions amid
inconsistent and conflict-ridden external forces. For instance, Scholz and Wei (1986)
contend that:

The image of public bureaucracy is that of an organization that responds

rationally to political demands but does so in a complex, federalist environment

in which statutory commands and oversight by central institutions provide only
one set of conflicting signals. The role of federal agencies in the American
policy process is not simply one of translating central political decisions into
organizationally efficient routines. . . .Instead, the creative role of the bureaucracy
requires the development of organizationally feasible tasks that will gain and
maintain sufficient support from critical actors in multiple operational arenas
without undermining central support needed for formal budgets and statutory
adjustments.

This view bases its explanation on the assumption that bureaucratic behavior is a
function of political inputs and bureaucratic discretion. The problem here is that there is
no sound logic for distinguishing bureaucratic discretion from bureaucratic behavior.
Simply put, it is erroneous to put bureaucratic discretion into the explanatory equation as
an independent factor along with the political input variables, thereby ignoring the fact
that bureaucratic discretion is also one aspect of bureaucratic behavior.

Still there is another view that relations between democratic institutions and

bureaucracy is reciprocal. When viewed as a ‘creative’ organization, bureaucracy may

receive signals of political institutions only selectively to fit its organizational needs and

40

capacity. If politically imposed policy goals are balanced with bureaucratic demands, the
relationship between political institutions and bureaucracy can be considered as being
reciprocal. As Wood and Waterman (1994: 126) put it:

[B]ureaucracies are more than vacuous receptacles of democratic power

responding in any direction political principals want them go. Rather,

bureaucracies also have power in their own right and sometimes use that power to
alter outcomes in their relations with other actors. . ..[R]elations between
politicians and the bureaucracy are bidirectional, with politicians sending signals
and bureaucracies responding at some times and with bureaucracies sending
signals and politicians responding at other times. (emphasis added)

In a similar vein, Krause (1999: 12) argues that:

Policy administration is the product of joint (endogenous) interaction between

governmental organizations and political institutions, subject to environmental

considerations.

This perspective emphasizes the role of bureaucratic feedback to politicians so
much that the relationship between political institutions and bureaucracy is considered to
be bilateral. This kind of view underestimates the importance of asymmetry between the
bottom-up bureaucratic influences and the top-down political influences. Although
bureaucratic inputs may be one possible factor that elected policymakers should consider
to set policy goals, it is only one of numerous factors. On the other hand, political inputs
including general policy directions may be important more than any other factors for
agency officials to set their own guidelines for policy implementation. A very informed
observer, Herbert Kaufman (1981: 166), notes that:

Members of Congress and their staffs have been known to defer to the judgment

of the leaders of the agencies, accepting their reports and recommendations

despite competing pressures from other quarters; influence ran in both directions.

But the relationships were not symmetrical. Congress could rarely be led by the

chiefs if it was strongly unwilling; the reverse was not equally true. Congress’s

displeasure therefore was not risked often or casually by the chiefs, and its favor
and respect were diligently nurtured. (emphasis added)

41

So what kind of behavioral patterns can be considered to be autonomous
bureaucratic behavior? The concept of bureaucratic autonomy still remains ambiguous
in the existing literature. As Hammond (2003: 76) decries:

Earlier definitions referred to the general ability of a bureaucracy to do what it

wants, but the definitions did not embed the bureaucracy in any particular

political context. This left it unclear as to whether the bureaucracy could adopt
any policy it wanted or just some policies, and if just some policies were feasible
the definition did nothing to specify what particular policies were feasible and
why.

That is, at the center of the ambiguous conceptualization of bureaucratic
autonomy lies confusion about the relationship between political accountability and
autonomous bureaucratic behavior. All those terms used by researchers, such as
“Independence,” “Noncompliance,” and “Reciprocity”, build on the view that
bureaucratic autonomy is an antithesis of political accountability in varying degrees. Do
autonomous bureaucracies always violate the principles of democracy? Is there some

possibility that politically accountable bureaucracy can also exert some degree of

autonomy?

2.3.2. The Omission of Collective Institutional Actions

Although theoretical models increasingly emphasize the joint actions of multiple
institutions in affecting bureaucratic decisions and actions, previous empirical research
seems to be based on an insufficient understanding of the nature of interactions among
political principals and agency officials. Most empirical research on political control did
not appropriately consider how multiple principals interact with each other. Early works
focused only on the dyadic relationship between one political institution (i.e., either

Congress or the president) and one agency (Weingast and Moran 1983; Moe 1982,

42

1987). Analyses of the dyadic interactions have produced only inconsistent results.
Depending on their focus, it is Congress but not the president or the president but not
Congress to whom agencies are held accountable. Considering the fact that all those
institutions have formal authority over agency actions, the dyadic approach leads to the
omission of important independent variables. Since the dyadic models examine one
institution while completely neglecting the others, the validity of any inference from the
under-specified models is in question.

As previously noted, recent empirical studies tend to use additive multi-
institutional models that include all political institutions. Apparently these models are
not under-specified. However, most of these studies underestimate the importance of the
fact that these institutions interact with each other as they try to influence the agency. In
other words, this approach builds on the erroneous assumptions that the relations among
political institutions are nonreciprocal and that political influence flows through multiple,
disjointed, and independent channels. Moe (1985: 1109) once concluded that:

We have been able to estimate the impacts of each of the three governmental
institutions [---the president, Congress, and the courts] while controlling for the
other two, and each accounts for a significant portion of the variance, which adds
substantially to our confidence in assessing political control.

This kind of conventional empirical design does not fit our understanding of
inter-institutional relations. We have learned from positive theories that the relative
alignment of an agency with one particular institution cannot be identified by the
covariate relationship in a multiple regression model (Hammond 1998). Instead, it is
determined by the location of the agency’s ideal point and the distance between the

status quo and all political institutions’ ideal points. That is, there are situations in which

changes in agency behavior responding to an individual principal do not lead to a close

43

alignment of the agency and the principal. For example, pro-business shifts of
bureaucratic actions corresponding to an inauguration of a Republican administration do
not necessarily imply that the agency’s position is closest to the Republican president if
we do not consider the agency’s position relative to other political institutions. If the
agency’s position has been closely aligned with a Democratic-controlled Congress, the
agency’s position can be closer to Congress than to the president even after the agency’s
pro-business policy shifts. Since the additive multi-institutional model considers
political influence as consisting of independent streams of discrete institutional
preferences, there is no way to examine interactions among separate but intertwined
powers. Hammond and Knott (1996: 120, 126) criticize this careless treatment of inter-
principal interactions in empirical research as follows:
[M]ost major components of the literature lack an explicit theory of how the
president, Congress, bureaucracy, and courts interact to make public policy.
Lack of an explicit theory makes it difficult to know what would constitute
disconfirming evidence for any hypothesis about who controls the
bureaucracy... .Some of these [empirical] studies have focused on the influence of
just one institution at a time, and even the broader, multi-institutional studies
rarely have tested theories that specifi/ the nature of the interactions among these
institutions. This leaves the reader unsure as to whether key variables have been
considered, or even whether the proper statistical measure has been constructed
for evaluating data about influence over policy outcomes. (emphasis added)
This gap between theoretical models and empirical research has not been bridged

successfully; the formal-empirical divide in the literature on political control and

bureaucratic autonomy still remains substantial.

2.3.3. Methodological Pitfalls
How the extent of autonomous bureaucratic behavior can be empirically tested is even

more ambiguous. By and large, past research has examined the presence of bureaucratic

44

autonomy from the perspective of a political stimulus and a bureaucratic response. This
stimulus-response system can be expressed succinctly in the following form:

Policy Outputs (Bureaucratic Behavior) = f (X, T)

Bureaucratic behavior or policy output is considered to be a function of a set of
political inputs (X) and a set of bureaucratic discretion variables (T). In the dyadic
approach, X includes variables representing the policy preferences of one institution--
either the Congress or the president; in the multi-institutional approach, X includes the
policy preferences of all seemingly pivotal institutions. If coefficients on both X and T
turn out to be significant, one might conclude that not only do bureaucrats respond to
institutional preferences X but they also exert discretion to adjust their task to factors in
T (Scholz and Wei 1986). If coefficients on T but not on X turn out to be significant, one
might conclude that institutional preferences do not matter and only bureaucratic factors
do matter (Eisner and Meier 1990; Wood 1988).

This stimulus-response system, however, entails methodological pitfalls when it
is applied to testing hypotheses of politics-bureaucracy relations. The major problem is
that it is extremely difficult for us to determine the extent to which bureaucrats are held
accountable for democratic control mechanisms. If the net impact of political inputs can
be articulated only by holding constant bureaucratic discretion factors, we should assume
that the amount of bureaucratic discretion is independent from institutional preferences.
However, we already know that bureaucratic discretion is not exogenous to institutional
factors. Theoretical models have demonstrated that the amount of bureaucratic
discretion may be determined by various factors such as the level of policy conflict

between branches of the government and the transaction costs to write a detailed

45

 

legislation. Second, although bureaucratic responsiveness and bureaucratic discretion
are both components of bureaucratic actions, the stimulus-response system as used in
past empirical research treats bureaucratic discretion as one determinant of bureaucratic
actions. This is odd since the amount of bureaucratic discretion or the level of
bureaucratic autonomy is an essential part of bureaucratic behavior. Both bureaucratic
responsiveness and bureaucratic autonomy (or discretion) should be explained by some
other factors including the political institutions’ policy preferences.

The other major problem is that empirical works include both some measures of
the institutions’ preferences (i.e., interest group ratings, partisanship, and so forth) and
some measures of the exercise of control tools (i.e., change in budget, appointment of
new agency head, reorganization, and so forth) in the set of political inputs (X).
Although both are important factors that influence agency actions, these should not be
treated as if they are exogenous to one another. In fact, the likelihood of a use of control
tools is also determined by the preference configuration of the institutional actors. The
amount of money allocated to an agency carmot be independent from the policy
preferences of the president and Congress. If the president and Congress do not want to
maintain the current scope and extent of the agency’s program, they will reduce ftmds for
it. If the president and Congress do not want to expand a regulatory agency, they will
agree to choose a person who can streamline the agency on behalf of her principals.

Unless these uses of control tools are seen to reflect the political principals’
preferences, there is no logical ground for one’s expectation that the principals’
preferences will make a noticeable difference in agency actions. When we predict some

systematic relationship between institutional preferences and agency actions, we should

46

assume that agency officials will keep watch on changes in institutional preferences
because the former will be afraid of punishment by institutional actors on bureaucratic
actions that are not in accordance with institutional preferences. If we take this kind of
endogeneity for granted, we should consider either the preferences of institutional actors
or uses of control tools but not both at the same time. In sum, we cannot examine the
effects of the preferences of institutional actors while controlling for uses of control tools

such as budgets and appointments, and vice versa.

47

CHAPTER 3

THEORETICAL FRAMEWORK

Positive theories of bureaucracy have asserted that agency actions are bounded by the
elected officials’ preferences since politicians define policy goals and the set of feasible
policy alternatives for an agency (McCubbins, N011, and Weingast 1989; Calvert,
McCubbins, and Weingast 1989). Building on the principal-agent framework, these
models assume that the amount of bureaucratic discretion is a function of two factors:
informational asymmetry and policy uncertainty. First, following Weberian depiction of
bureaucratic secrecy, informational asymmetry has received enormous attention (Bendor,
Taylor, and Van Gaalen 1987; Banks and Weingast 1992). An informational imbalance
between the political principals and the agent is thought to originate from the agent’s
policy expertise and the politicians’ monitoring costs. Second, policy uncertainties, such
as post-policymaking random shocks, are assumed to be revealed to policy implementers
but not to policymakers, so that politicians end up with only limited knowledge about
“real world” policy outcomes.

In addition to the agent problems, a host of studies have focused on multi-
institutional relations to explain varying degrees of bureaucratic autonomy. One line of
research emphasizes the institutional constraints on the level of legislative delegation.
Epstein and O’Halloran (1999) argue that policy conflict between different branches of
the government will motivate legislators to impose heavier restrictions on agency
actions. In contrast, Huber and Shipan (2002) contend that legislators opt for writing a

high-discretion bill allowing the bureaucrat to implement policy preferable to other

48

institutional actors amid inter-institutional policy conflict in order to minimize the threat
of executive veto and bicameral bargaining costs.

In this context, Hammond and Knott (1996, 1999) make a unique contribution to
our better understanding of bureaucratic autonomy in multi-institutional environments.
Unlike other models, their model implicitly assumes that the agency has already been
given lots of legal discretion. Rather, the focus of their model is on the question of
whether the agency is able to take full advantage of the legal discretion it has been given.
Hammond and Knott pay much attention to the very nature of the inter-principal
interaction and the role of a strategically-sophisticated agency manager within the
fiamework of multiple veto players.1 The institutional rules goveming inter-institutional
relations such as the executive veto and bicameralism tend to lead to policy stability (or
maintenance of current policy) since a major policy change can take place only if there is
an agreement among key institutional actors on replacing the status quo policy with a
new policy.

Focusing on this theoretical expectation about policy change (or disequilibrium),
Hammond and Knott demonstrate the existence of a set of status quo policies which
those principals cannot agree to replace with other alternatives. Knowing this, a strategic
agency head can choose a policy closest to her own ideal point in the set of equilibrium

policies. Hanunond and Knott (1996: 144) argue that the potential variability of

 

' As they put it, “[T]he two major reasons for bureaucratic autonomy. . .are asymmetric information and
multiple principals. The most general model should, of course, include both factors. However,
incorporating asymmetric information would entail mathematical complexities which, for reasons for
tractability, would require simplifications elsewhere in the model, especially a reduction in the number of
institutions considered. In the face of this trade-off our choice is to maintain a relatively complete set of
institutions” (1996: 127).

49

politically-feasible policy options for the agency indicates the degree of bureaucratic
autonomy as follows:

[Pjolitical autonomy [of an agency] means that the agency director can adopt a
new policy without being reversed by the president or Congress. The key to
understanding agency autonomy, then, is whether there exist any policies that the
president and Congress cannot upset if chosen by the director. It follows that the
most appropriate measure of an agency ’s political autonomy is simply the size of
the set of equilibrium policies produced by the president and Congress.

(emphasis in original)

As the size of the set of equilibrium policies---the core---increases, so does the
extent of bureaucratic autonomy. The size of the core, regardless of policy
dimensionality, is non-decreasing or increasing with a greater preference divergence
among the veto players. The size also is non-decreasing as new veto institutions are
added.2 The extent of bureaucratic autonomy rests on some interactions of institutional
preference configurations and institutional fragmentation (or the number of veto
institutions). For instance, the added veto institution can increase the core only if it is a
preference outlier; and an increase in preference divergence among outlying institutions
can increase the size of the core even with a fixed number of veto institutions.

Viewing bureaucratic actions in this way clarifies the conceptual confusion about
the notion of bureaucratic autonomy in the literature. As Hammond (2003: 77) put it:

[A] bureaucracy can be more or less autonomous, depending on the size of the set

of equilibrium policies. Moreover, by relating the extent of bureaucratic

autonomy to the size of a set of equilibrium policies, there is always a boundary
to the set. This boundary sets limits on what the bureaucracy can and cannot do:
it can move from policy to policy within this equilibrium set, but it cannot sustain

a policy that lies outside this equilibrium set. [T]he preferences of the elected
ofi‘icials will always collectively constrain the range of bureaucratic

 

2 These propositions build on findings of Hammond and Miller (1987) in their APSR article, “The Core of
the Constitution.” Hammond and his coauthors have continued to extend them in a variety of
policymaking processes (Miller and Hammond 1990; Hammond and Hill 1993; Miller, Hammond, and
Kile 1996; Knott and Hammond 2000; Hammond 2003; Hammond and Butler 2003). Very similar
propositions have been presented in several bodies of work by Tsebelis (1995, 1999, 2002).

50

choices. . ..[But] as long as the bureaucracy selects some new policy from inside
the boundary the disagreements among the politicians will keep them fiom
upsetting the bureaucracy ’s choice and imposing some other policy. (emphasis
added)
3.1. A Spatial Model
I now present a spatial model of bureaucratic policy choices to highlight basic
relationships among policy actors and the underlying flow of causality. This model,
which I call “Multiple-Principals and Large-N-Agents” or MPLNA, extends Hammond
and Knott’s spatial model of bureaucratic autonomy in order to incorporate a large but
finite number of agency officials. Agency actions can be seen as the aggregate sum of
individual members’ actions. Take an example of regulatory enforcement activities of
the Occupational Safety and Health Administration (OSHA). OSHA’s regulatory
performance or regulatory stringency has been measured by the sum of individual
enforcement officers’ activities such as the number of inspections, the number of
violations cited, or the amount of penalties (Scholz et a1. 1986, 1991). Therefore, rather
than assuming an agency as a unitary actor represented by a single head, we can consider
an agency as a distribution of a large number of individual officials, each of whom has
somewhat independent decision-making authority in the field. The hierarchical structure
of bureaucratic organizations will be left aside despite its importance in shaping
regulatory behavior (Moe 1985; Padgett 1981; Carpenter 1996). But not all intra-
bureaucracy factors will be ignored; in particular, agency-level policy bias, viewed as a
skewed distribution of bureaucratic policy preferences, will be included in the model.

Let me begin by assuming that there are multiple veto players i = l, 2,.., k, whose

ideal points are denoted by V,- and a large number of agency officials j = 1, 2,. . ., n,

51

whose ideal points are A]? Veto players are individual or collective actors whose
agreement is necessary for a change of the status quo. It follows that a change in the
status quo requires a unanimous decision of all veto players. In contrast, agency officials
do not possess veto power. Each player’s utility profile is assumed to be single-peaked
on his or her most-preferred policy and symmetric in a one-dimensional policy space, X
= [0, 1]. In the context of regulatory policy production, 0 indicates a complete absence
of regulatory activities and 1 indicates the maximal level of regulatory activities. At the
agency level, the aggregate distribution of preferences (ADP) of the agency officials is
assumed to be Beta-distributed such that A,- ~ Beta(a, b) in the domain of [0, 1].4 If the
two Beta-distribution parameters are restricted to a = l and b = 1, this means that the
agency officials’ ideal points are assumed to be spread evenly over the whole domain of
the policy space. We can take advantage of the flexibility of Beta distribution when we
consider various distributional-shapes of bureaucratic preferences in the latter discussion.
Agency official j ’s policy choice is denoted by C}. Lastly, complete information is
assumed: each veto player knows the other veto players’ ideal points; each agency
official knows the veto players’ ideal points; and no random factor interferes between

policy choices and outcomes.

 

3 It is implicitly assumed that a veto player is a representative member of each political institution for
simplicity. For instance, under the bicameral-executive veto system, veto players can be thought to be the
president, and the two chamber median legislators in the House and the Senate. In this case, the number of
veto players is considered as k = 3. However, as discussed latter, the number of defacto veto players can
be greater than 3 depending on different views of who the key congressional actors are.

‘ Since the range is confined to [0, 1], the standard Beta distribution is used here. That is, PDF

=f(x) = (x""(1— x)"“)/B(a, b) and CDF = F(x) = Lxxa‘l (l — x)”‘l dx/ B(a, b) ,

1
where B(a,b) = J; x"'1 (l - 10”"l dx ; O S x S l;a,b > 0. Latter in this section, I discuss several

situations where agency officials’ preference distributions can vary. With this Beta-distribution it is easy
to get a variety of distributional shapes by changing restrictions to the two shape parameters, a and b, so
that we can incorporate agency officials’ heteroscedastic preference distributions into the spatial model.

52

 

3.1.1. Policy Disequilibrium and Policy Equilibrium
Let me first consider the conditions of policy change (disequilibrium) and policy stability
(equilibrium). Following Hammond and Knott, the condition of policy disequilibrium is
a non-empty winset or W(SQ) =# Q; in other words, for a given status quo there exists a
set of policy alternatives that make all veto players better-off. Figure 1(a) presents this
condition. For simplicity, consider three veto players, i = 1, 2, 3, whose preferences are
ordered as V, < V2 < V3 as illustrated in Figure l. The winset of an arbitrary status quo,
W(SQ), is [SQ, V1+|V1-SQ|], which is the intersection of three winsets for the veto
Playcrs, W1(SQ) = [SQ V1+IVrSQlL W2(SQ) = [SQ, V2+IV2-SQII, and W3(SQ) = [SQ,
V3+| V3-SQI]. As long as W(SQ) is not empty, three veto players will agree on replacing
the status quo with any alternative inside the winset.

Next, the condition of policy equilibrium is an empty winset, that is W(SQ) = O.
In other words, if there is no other policy alternative increasing all veto players’ payoffs,
the status quo is maintained. In fact, the status quo with an empty winset is an element
of the core, a set of policy alternatives that are not dominated by any other alternatives,
given a profile of actors’ preferences and decision rules (Davis, DeGroot, and Hinich
I972; Hammond and Miller 1987; Friedman 1990; Tsebelis 2002). This is due to the
very nature of interactions between veto players: any decisive coalition of veto players
cannot change the status quo without a multilateral agreement of all veto players.
Figure 1(b) shows this stalemate. Any proposal to replace SQ with some other
alternative, let us say to the left toward veto player l’s ideal point, V], will face veto by
other two veto players. By the same token, a proposal favoring V2 will face veto player

l’s opposition

53

 

Figure 1. Policy Disequilibrium and Policy Equilibrium

(a) Condition of Policy Disequilibrium: A Non-Empty Winset

 

W3(SQ)

 

W2(SQ)

 

 

 

 

 

W1(SQ)
I

o SQ V, V,+|V,-SQ| V2 V3 V2+|V2-SQ| V3+|V3-SQ| 1

 

 

—

(b) Condition of Policy Equilibrium: An Empty Winset

 

W3(SQ)

 

W2(SQ)

 

 

 

 

 

 

WKSQI

0 VI‘IVI'SQI V I SQ V2 V3 V2+IV2-SQI V3+IV3-SQI 1

54

 

and the attempt favoring V3 cannot avoid being vetoed by veto players 1 and 2. This
equilibrium will be maintained for any status quo as long as it lies inside the interval of
[V1, V2]-

We can derive the following properties from the logic of policy (dis)equilibrium
discussed above.

Property 1. Policy disequilibrium: a non-empty winset: If SQ<min( V.) or
SQ>max( V.) for any status quo policy SQ, the winset W(SQ)= fl W.(SQ)¢ O.

Property 2. Policy equilibrium: an empty winset: If min( V.)S SQSmax( V.) for
any status quo policy SQ, the winset W(SQ)= fl W.(SQ)= Q.

These properties together suggest that any policy alternative lying inside the
interval of [min( V.), max( V.)] cannot be defeated by any decisive coalition of veto

players, and thus these undominated policies remain stable. The interval of [min( V.),

max( V.)] is called the Policy Equilibrium Interval (PEI) hereafter.

3.1.2. Policy Equilibrium Interval and Agency Actions
What do these theoretical properties of the policy equilibrium interval suggest for the
agency officials’ decisions? The policy equilibrium interval cannot change the
preference profiles of bureaucratic agents. However, it can influence bureaucratic policy
choices given the hierarchical arrangement governing the relationship between veto
players and bureaucratic agents.

More specifically, agency officials can make policy choices while striving to
meet two conditions. First, agency officials will attempt to avoid political upset of their
policy choices. In order for this condition to be met, agency officials should choose

alternatives from the set of equilibrium policies such that min( V.) S C,- _<_ max( V.).

55

 

Agency officials know that if their chosen policy lies outside the interval the politicians
could agree to replace the would-be status quo with a new policy. However, as long as
agency officials adopt policy options inside the policy equilibrium interval of [min( V.),
max( V.)], these bureaucratic choices are politically feasible following the logic of policy
equilibrium---the lack of an agreement among the veto players. Second, agency officials
will attempt to maximize their payoffs. For this condition, agency officials should adopt
the option which is closest to their own ideal points such that minlAj — (3,]. The only way
to minimize the loss of bureaucratic utility is to choose the options closest to their most
preferred policy among the options inside the policy equilibrium interval. In sum, these
conditions suggest that agency officials will adjust their choices to avoid a prospective
political upset and also try to minimize their loss of utility.

We can examine these strategically-sophisticated agency officials’ policy choices
through two cases: (1) when the agency officials’ ideal points lie outside the policy
equilibritun interval, and (2) when the agency officials’ ideal points lie inside the policy
equilibrium interval. First, when agency officials’ ideal points lie outside the PEI, they
will choose the policy that constitutes the boundary or “limit” of the PEI. For any
agency officials whose ideal points lie to the left of [min( V.), max( V.)], min( V.) is the
policy option that can maximize agency officials’ payoff, given that they cannot choose
their own ideal points. And for those whose ideal points lie to the right of [min( V.),
max( V.)], max( V.) can maximize agency officials’ payoffs. Therefore, we can deduce the
following properties.

Property 3. Agency’s policy choices with a non-empty winset:

(1) If Aj<min( V.), agency officials’ policy choices (C1) are min( V.), the lower limit
of the policy equilibrium interval.

56

(2) If A,>max( V.), agency officials’ policy choices (C1) are max( V.-), the upper
limit of the policy equilibrium interval.

Second, for those agency officials whose ideal points lie inside the policy
equilibrium interval, agency officials’ ideal points can be chosen without fearing
political upset.

Property 4. Agency’s policy choices with an empty winset: If min( V.)S A, 5
max( V.), agency officials’ policy choices (C1) are A], their ideal points.

What do these relations among elected policymakers and agency officials imply
for bureaucratic actions in regulatory policymaking processes? Changes in the location
and size of the policy equilibrium interval [min( V.), max( V.)] will affect bureaucratic
choices and actions. For example, there are possibilities that the elected political leaders
are replaced through electoral processes by others who have different policy preferences
and that this political upheaval changes the policy equilibrium interval. In Figure 2(a)
and (b), V, fails to get reelected and her position is taken over by V}. (> V3). As a result,
the previous PEI, [V1, V3], is replaced by a new interval, [V2, Vl']. Consequently, both
the location and the size of the interval have changed: the interval moves to the right and
the interval gets smaller. The replacement of V; by Vl‘, which yields the shift of the
policy equilibrium interval from [V1, V3] to [V2, V,’], will redefine the set of politically-
feasible policy options for the agency and the extent to which agency officials can take
unilateral actions. More succinctly, the agency’s policy choices can be presented as

follows before and after the replacement of V, by V,’:

VlifAj<Vl erifAj<V2
Cf: AjiftqujsII, —> Cj=iAjifV2sAjsz
V3ifV3<Aj - LVI'ifVI.<Aj

 

57

 

Figure 2. Policy Equilibrium Interval and Agency Policy Choices

(a) Policy Equilibrium Interval: [V1, V3]

V. A,-

V3

 

 

 

 

 

 

(b) Policy Equilibrium Interval: [V2, V2 *]

 

 

 

 

 

 

 

 

 

 

 

 

V2 A} Vl
I
0 V2 V3 V1*
(c) Policy Equilibrium Interval: [V2, V2 *]
V2 A ,- V,‘
I
0 V2 V3* V1*

58

 

 

 

 

Hence, we can examine the effect of the policy equilibrium interval on agency
actions in two aspects: (1) the effects of the horizontal location of the policy equilibrium
interval and its movement on the overall level of regulatory policy outputs produced by
the agency, and (2) the effects of the size of the policy equilibrium interval and its
change on the variability of agency choices. First, the location of the policy equilibrium
interval has an impact on how much regulatory policy output the agency produces. In
Figure 2(a) and (b), the policy equilibrium interval moves in the pro-regulation direction
with the replacement of V, by Vl‘ after the election. Since both the lower and upper
limits of the policy equilibrium interval shift toward the pro-regulation direction, agency
officials will have to choose from a set of stronger regulatory options. Given an

assumption that agency officials’ ideal points are spread evenly in the policy space, the
total policy outputs produced by the agency as a whole (= Z; C I. ) is expected to

increase. In formal terms, the overall level of policy outputs can be expressed by the
conditional mean of agency officials’ policy choices given veto players’ ideal points (V.),
agency officials’ ideal points (A!) and the shape parameters (a and b) of the aggregate
distribution of A,- such that E(C,-| V., A}, a=l, b=1). If the policy equilibrium interval
moves in the pro-regulation direction, the conditional mean of agency officials’ policy
choices will increase. This observation is consistent with the ‘political control’
perspective: changes in the preference configuration of the political principals will lead
to corresponding changes in agency actions in a systematic manner. Thus, we can
present the first hypothesis as follows.

H1: As the policy equilibrium interval moves in a pro-regulation (anti-

regulation) direction, the level of an agency’s regulatory policy outputs will
increase (decrease).

59

However, it is noteworthy that not all veto players’ electoral fates have such an
impact on the agency’s policy choices. In Figure 2(b) and (c), veto player V3 is replaced
by V 3', where V3<V3‘<V1'. Although V3‘ prefers a stronger regulatory policy than V3,
this political change does not result in any change in the location and size of the policy
equilibrium interval. This observation suggests that only the political institutions that
constitute the outer limits of the policy equilibrium interval and their policy preferences
matter for agency officials’ policy decisions. Thus, the conventional view that
bureaucratic actions are a product of the sum of all individual principals’ independent
influences ignores the importance of the inter-principal interactions. Although each veto
player possesses formidable weapons of control (appointment, budget appropriations,
oversight hearings, and so forth), they can use these only when there is a multilateral
agreement on punishing agency officials’ undesirable behavior. Furthermore, since the
political institutions’ relative positions on a policy can change as a result of an election,
the political institution that constitutes one of the limits of the policy equilibrium interval
also can be different at different points of time. Therefore, it seems fallacious to assert
that a particular political institution (i.e., the president, the Senate, or the House) is
always superior to others in regard to its influence on bureaucratic decision-making and
actions.5

Second, the size of the policy equilibrium interval can determine the variability of
policy options from which agency officials can choose. In Figure 2(a) and (b), after the

replacement of V; by VII, the size of the policy equilibrium decreases, and the range of

options given to the agency officials gets narrower. Under the condition that the agency

 

5 Hammond (1998) provides a discussion of why the conventional empirical research design based on the
additive model-«including all individual political institutions in one regression equation-"may not be
consistent with the analytical results of this kind of spatial model.

60

officials’ ideal points are distributed evenly in the policy space, a smaller policy
equilibrium interval indicates lower variability. That is, as the size of the policy
equilibrium decreases, so does the potential variability of autonomous bureaucratic
choices. The variability can be expressed in technical terms as the conditional variance
of agency policy choices given the veto players’ preference configuration, the agency
officials’ preferences, and distributional parameters such that Var(Cj| V., A}, a=1, b=1).
Thus, we can expect that as the size of the policy equilibrium interval increases, the
conditional variance of the agency’s regulatory actions will increase. In other words,
consistent with the ‘bureaucratic autonomy’ perspective, bureaucrats can maintain a
degree of political autonomy to the extent that political principals disagree on replacing
the status quo policy.

H2: As the policy equilibrium interval gets larger (smaller), the variability of
an agency’s regulatory choices will increase (decrease).

These hypotheses (H1 and H2) together depict an image of bureaucracy that
continuously adjusts its behavior in response to the political institutions’ preferential
changes and at the same time tries to find some room within these political constraints so
that its bureaucrats can develop and adopt a preferred policy. A similar depiction, in
fact, can be found in many instances from existing body of literature. For instance,
Scholz and Wei (1986: 1264) remark that:

The role of federal agencies in the American policy process is not simply one of

translating central political decisions into organizationally efficient

routines. . ..Instead, the creative role of the bureaucracy requires the development

of organizationally feasible tasks that will gain and maintain sufficient support

from critical actors in multiple operational arenas without undermining central
support needed for formal budgets and statutory adj ustrnents. Thus, bureaucracy
plays the important and seldom-recognized role of integrating political demands

made at various levels of the American federalist system by incrementally
adapting central policies to fit into varied and changing local conditions.

61

The spatial model and propositions derived from it clarify the latent process
underlying the interactions among democratic institutions and agencies. Bureaucratic
responsiveness to institutional preferences and autonomous bureaucratic operations can
take place simultaneously. These two processes together comprise the bureaucrats’

behavioral repertoire.

3.1.3. Heteroscedastic Distribution of Bureaucratic Policy Preferences
What roles do agency officials’ policy preferences play in the policy process? Even
Woodrow Wilson (1887) recognized the importance of policy preferences of individual
policy implementers when he claimed that “the administrator should have and does have
a will of his own in the choice of means for accomplishing his work. He is not and ought
not to be a mere passive instrument” (212). In fact, students of contemporary
bureaucracy echo this assertion. James Q. Wilson (1989) contends that the performance
of a regulatory agency can be affected by agency officials’ personal motivations such as
organizational maintenance, professional norms, and political concerns. Rothenberg
(1994) shows us that the Interstate Commerce Commission’s regulatory decisions reflect
its members’ ideology. Brehm and Gates (1997) also propose that the attitudes of front-
line workers on their tasks in public service organizations are one of the most important
determinants of organizational performance.

How can we characterize agency officials’ policy preferences? Some researchers
consider agency officials as biased toward more stringent regulation when it comes to
environmental quality and job safety (Meier 1993; Rourke 1984; Kelman 1980, 1981).

But others have assumed that bureaucrats seek to maximize budget (N iskanen 1971),

62

 

discretionary resources (Miqué and Bélanger 1974; Niskanen 1975), or leisure-time
(Brehm and Gates 1997). In other words, researchers’ depiction of bureaucratic
preference depends on their own views in particular ways. Each of these accounts of
bureaucratic preferences, however, is not consistent with more general perspectives that
stress motivational and preferential diversity inside the bureaucracy (Downs 1967). At
the individual level, bureaucrats may have a huge array of different personal tastes and
preferences. This individual-level preferential diversity leads to the difficulty
characterizing the bureaucracy based simply on either purely purposive (i.e., good public
policy) or purely materialistic (i.e., money and resources) motivations.

In order to take into account the possibility of diverse bureaucratic preferences, I
assume that the aggregate distribution of bureaucratic preferences on regulatory tasks be
heteroscedastic. This equation of agency preference with different shapes of
distributions of the agency’s officials reflects the fact that the overall level of regulatory
performance of an agency is actually the sum of policy outputs that those front-line
workers have produced. Individual agency workers can be classified according to their
preferences on regulation as follows (Miqué and Bélanger 1974; Niskanen 1975). First,
agency officials who prefer lenient regulation to strong regulation may stop producing
additional regulatory policy outputs at the point where the difference between
politicians’ total payments and the agency’s total production costs reach the maximum.
This maximization of slack resources leads to under-production of regulatory policy
outputs lower than the optimal level. Second, agency officials whose preferences are
skewed toward stringent regulation may continue to produce regulatory outputs as long

as politicians pay for a unit of additional production. Since policy production continues

63

until the marginal payment reaches zero, over-production takes place higher than the
optimal level. Lastly, neutral officials are those who produce policy outputs under the
condition that politicians’ total payments covers total production costs, which leads the
agency to maintain the optimal level of regulatory policy outputs.

The aggregate distribution of bureaucratic preferences can vary in time and space.
The distribution of agency officials’ preference can change in a single agency as the
membership composition changes for a long period of time. The distribution of
bureaucratic preferences can also vary across different agencies, depending on agency-
to-agency idiosyncrasy. To illustrate these variations of the aggregate distribution of
bureaucratic preferences, I focus on the skewness of the distribution of agency officials’
preferences. Figure 3 portrays three aggregate distributions of policy preferences of
randomly sampled agency officials (n = 100). The upper-left panel of Figure 3 presents
negatively-skewed agency preferences in favor of the maximal regulation. In this
agency, a disproportionately large portion of agency workers prefer stronger regulation
to less regulation. The right-upper panel presents unskewed agency preferences. This
agency consists of workers whose ideal points are distributed evenly in the entire range
of the policy space. The bottom panel of the figure portrays positively-skewed
distribution of agency preferences in favor of the minimal regulation.

These different shapes of the aggregate preference distribution of the agency can
affect how an agency takes actions in two ways: the level of regulatory policy
production, E(C,-| V., A}, a, b), and the variability of the agency’s policy choices, Var(C,-|

V.-, A], a, b). Intuitively, the effect of the location of the policy equilibrium interval on

64

Figure 3. Heteroscedastic Distribution of Bureaucratic Preferences

(b) Unskewed

(a) Skewed to the Maximal Regulation

 

 

 

 

 

 

 

 

 

P
I5.”
.m
14.0.
m
13.6
PM.
:0
a _ lq 4 J _
n v m N P o
3:09.20 .3832
-9.
..8.
-..m
m
8%
PM
.Iu
[4.0.
w
[3.6
:2.”
:0
Al a q A q —
m w n N P o
8:02.33832

(c) Skewed to the Minimal Regulation

 

 

Preferences

 

 

—4

d
.1

— - _
c n w w o
3:02 Co .3532

65

the overall level of the agency’s output production will be larger when the agency’s
preferences are skewed than when they are unskewed. Figure 4 portrays two agencies’
actions when the same change of the policy equilibrium interval occurs, from [0, 1] to [0,
0.95]. Take an example of an agency skewed in the pro-regulation direction in Figure
4(a). If the upper limit moves in the deregulation direction, a crowd of agency workers
should adjust their choices and the agency’s overall level of regulatory output
productions will decrease substantially. By contrast, in an agency with unskewed
preference distribution as portrayed in Figure 4(b), the same change in the policy
equilibrium interval will trigger only some marginal adjustment in agency choices. The
amount of change in the policy output level caused by an identical change in the location
of the policy equilibrium interval is greater in the former case than the latter. We can
derive the following hypothesis about the conditioning effect of agency preferences on
the location of the policy equilibrium interval and the production level of a regulatory
agency:

H3: The effect of the location of the policy equilibrium interval on the

production level of a regulatory agency will increase (decrease) as the

skewness of agency preference distribution increases (decreases).

On the other hand, the effect of a change in the size of the policy equilibrium
interval on the variability of the agency’s regulatory choices will be smaller when the
agency’s preferences are skewed than when they are unskewed. As seen in Figure 4(a),
if agency officials’ preferences are skewed, their choices are limited to a smaller range of
stronger regulatory policy options and the variance of agency choices is less sensitive to

change in the size of the policy equilibritun interval. If agency officials’ preferences are

distributed evenly as portrayed by Figure 4(b), change in the size of the policy

66

Figure 4. Change of Policy Equilibrium Interval and Agency’s Policy Choices

(a) Skewed Agency Preference Distribution (n = 1000)

[0,1] [0,095]

 

 

L
T

l
l'

100 120 140
I
r

100 120 140
I

80
l
I
80
1

Frequency

60
Frequency

60
J

I
l

40
I
l
40
l

 

 

 

0
0

 

 

 

8: ' ......1.......ulnnmniullIIIIIIIIIIWIIIIIIIIIIIIIIIllllllllllllllllllllllllllllllllll

0.1

I I
.8 .9 1 0 .1

2.3 .4 .5 .6 .7 .2 .3 .4 .5 .6.
Agents' Policy Choices Agents' Policy Choices

(b) Unskewed Agency Preference Distribution (n = 1000)

[0,1] [0,n95]

 

8: ....... ...l..uulunmnililllllllIllIIIIIIIIIIIIIIIIIIIIIIImill”Illll'lllllllllllll” .—

7.8.91

 

I
l

100 120 140
L L
I I
100 120 140
1

Frequency
60 80
I L
l 1
Frequency
60 80
1 l

0
1
fi
40
n

20
1
1

20

 

 

 

r
_ II|I|l||Illll|||III|I|II|I||I|||IIII|IIIIIIIIIIl||II|||III||III||I|IIIIIIIIIIIIIIIIIIIIIIIIIIIIIllli _ _ IIIIIIII|||II||||II|||II||IIII||II|I||I|||II|I|IIIIIllIIl|||IlI||IllllIlIlIlllIlIIIIIIIIIIIIII

0
0

 

 

 

7.8.91 0.1.2

0.1.2.3.4.5.6.
Agents' Policy Choices

.3 .4 .5 .6 .7 .8 .91
Agents' Policy Choices

67

equilibrium interval may lead to more substantial changes in the variance of agency
choices. Thus, we can predict the following relationship:

H4: The effect of the size of the policy equilibrium interval on the variability

of a regulatory agency’s policy choices will decrease (increase) as the

skewness of agency preference distribution increases (decreases).

These mediating effects of diverse agency preference distributions between the
policy equilibrium and agency actions (H3 and H4) are worth further discussion. First,
the influence of institutional preferences on agency actions should be considered as
conditional on how agency officials’ preferences are distributed. If the analyst ignores
various shapes of agency preference distributions, he or she may erroneously expect that
the effect of the location of the policy equilibrium interval on the level of regulatory
outputs and that of the size on the variability of regulatory actions remain constant.
However, when the mediating role of agency preference distributions is taken into
account, the policy equilibrium interval’s effects should be expected to vary to the extent
that the agency preference distribution is skewed.

Second, consistent with the general assumption of the principal-agent theory that
that a principal and an agent may have heterogeneous preferences, the agency’s
preferential skewness or policy bias can make a subtle difference in institutional effects
on agency actions. When the agency is extremely biased toward either deregulation or
regulation, the location of the policy equilibrium interval has the hypothesized effects on
the agency’s actions if the change in the location of the policy equilibrium interval has
been brought about by a movement of the interval’s limit in the agency’s favored side.

For instance, with an agency having an extremely pro-regulation bias, a movement of the

upper limit of the policy equilibrium interval will result in a substantial increase in the

68

amount of policy outputs while the lower limit’s movement will lead to only marginal
change in policy outputs. Therefore, the impact of political influence on agency actions
can be explained more accurately when we take into account not only how much the

agency is biased but also in what direction the agency is biased.

3.1.4. Simulation

In order to examine the effects of the policy equilibrium interval and the agency
preference distribution on agency actions, I conduct Monte Carlo simulations in a one-
dimensional policy space wherein a complete absence of regulation is 0 and the maximal
production of regulation is 1. First, I assume that the base policy equilibrium interval is
[0, l], which is the entire policy space, and then I make the upper limit of the policy
equilibrium interval move leftward from 1 toward 0 while holding the lower limit
constant at 0 in each trial. Accordingly, the location of the policy equilibrium shifts
leftward toward deregulation and the size of the policy equilibrium decreases
corresponding to the upper limit’s movement. Second, I assume that the base
distribution of agency preferences of 1000 officials is A,- = Beta(l , 1) in the domain of [0,
1], which is an unskewed-unifonn distribution, and then I increase the negative skewness
(pro-regulation bias) of the agency preference distribution by manipulating Beta-

distribution parameters.

3.1.4.1. The Effect of the Policy Equilibrium Interval on Agency Actions

To examine the effect of policy equilibrium interval on agency actions, I change the

upper limit of the policy equilibritnn interval by an increment of 0.001 for each of 1000

69

trials while holding the agency preference distribution unskewed by fixing parameters a
and b at 1. In each trial, agency officials’ policy choices (C1) are truncated into the
policy equilibrium interval of [0, l — (t/1000)], where t = 1, 2,. . .,1000. More

specifically,

C ={Aj ifAj Sl—(t/IOOQ

J 1—(t/1000 if A}. >1-(t/100(b

And then, the conditional mean, E(C,~| V., A}, a = 1, b = 1), and the conditional
variance, Var(Cj| V., A], a = l, b = 1), of the agency officials’ policy choices in each trial
are calculated and plotted.

The left panel of Figure 5(a) portrays the effect of the policy equilibrium interval
on the mean and variance in this simulation. As proposed by hypothesis 1, the leftward
shift of the policy equilibrium interval tends to decrease the conditional mean of the
agency’s regulatory production level. At the same time, consistent with hypothesis 2, a
decrease of the size of the policy equilibrium interval also tends to decrease the
conditional variance of agency choices (regulatory variability).

Now consider the right panel of Figure 5(a). Here the policy equilibrium
interval is held constant at [0, l] and the negative skewness of agency preference
distribution gets increased by 0.001 for each trial such that A ,- = Beta(l — (t/1000), 1). As
the agency preference distribution gets increasingly skewed in the stronger regulation
direction, the conditional mean of regulatory production, E(C,~| V., A j, a = 1— (t/1000), b =
1), tends to increase with the fixed policy equilibrium interval. At the same time,
however, the agency’s policy bias tends to decrease the conditional variance of agency

choices, Var(C}| V., A], a = 1— (t/1000), b = 1), suggesting that agency choices are

70

Figure 5. Monte Carlo Simulation

(a) Policy Equilibrium Interval “or” Agency Preference Distribution

Policy Equilibrium Interval Agency Preference Skewness
[0. 11-> [0. 0] Beta(1.1) -> Beta(.0001.1)

 

 

g-q - q-- l-
024 - 02‘ t-
Qn - ”.- r
as] - 8".- -
c c

fiQ-i

>0-
éq-
g..-
“-1

v.1

 

 

 

 

o-l

 

 

 

 

 

 

 

(b) Policy Equilibrium Interval “and” Agency Preference Distribution

Mean Variance

 

 

 

 

 

 

 

 

 

 

 

 

I I I r l I l I l l l T I
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 O .1 .2 .3 .4 .5 .6 .7 . .9 1
Upperlelt'sLel‘lward Shll't Upper leit'sLeflwardShltt
—— Beta(1,1) ----- Beta(.75,1) —— Beta(1.1) ----- Beta(.75.1)
---------- Beret.5.1) —-—-- Beta(.25,1) aerated) -'-" Betat.25.1)

 

 

 

 

 

 

71

constrained into pro-regulation policy measures as the agency preference distribution
becomes skewed with the fixed policy equilibrium interval.

Several observations can be made about these results. First, changes in the
political environment of a regulatory agency can bring about substantial changes in
policy outcomes. The horizontal movement of the policy equilibrium interval, which
represents a change in the composition of elected policymakers from those who advocate
stronger regulation toward those who support deregulation, tends to lead the regulatory
agency to decrease its policy outputs. Second, these changes in the policy preferences of
political institutions affect the amount of bureaucratic autonomy by increasing or
decreasing the range of politically-feasible policy options from which agency officials
can unilaterally make their choices. Third, the agency can make autonomous choices to
the extent that there is a disagreement on an ideal level of regulatory policy outputs
between the veto players. Unless all veto players have identical views on regulatory
policy, agency officials always have some options from which they can choose while
reflecting their own policy preferences without fearing a potential political threat. Last,
the agency’s regulatory policy outputs can change in response to changes in the agency’s
policy preference distribution, holding constant other factors including the politicians’
preference profile. With the location and size of the policy equilibrium interval fixed, an

increase in the agency’s pro-regulatory bias can increase regulatory policy outputs.

72

3.1.4.2. The Conditional Effect of the Policy Equilibrium Interval on Agency
Actions

Next, I examine how the agency’s preference distribution can mediate the impact of the
policy equilibrium interval on the agency’s policy choices. I create four different agency
preference distributions: Beta(0.25, l), Beta(0.5, 1), Beta(0.75, l), and Beta( 1, 1). The
first represents the greatest pro-regulation bias among the four, while the last represents
complete neutrality. With these heteroscedastic distributions of agency preferences, I
make the upper limit of the policy equilibrium move leftward toward deregulation by an
increment of 0.001 for each of 1000 trials. Conditional means and variances of those
four different agencies are plotted for comparison in Figure 5(b).

The left panel of Figure 5(b) presents the conditional effect of the location of the
policy equilibrium interval on the level of regulatory policy outputs. Consistent with H3,
the effect of the policy equilibrium interval on the conditional mean of policy outputs
tends to increase as the agency’s preference distribution becomes skewed. The rate of
change in the conditional mean of agency choices in response to the upper limit’s shift is
greatest when the agency preference distribution is skewed to the greatest degree
(Beta(0.25, 1)). In contrast, the conditional mean of agency policy choices is the least
sensitive to changes in the location of the policy equilibrium interval when the agency
preference distribution is unskewed (Beta(l, 1)). In other words, there is a positive
relationship between the skewness of the agency preference distribution and the effects
of the policy equilibrium interval’s location on the conditional mean of regulatory policy

outputs.

73

The right panel of Figure 5(b) presents the conditional effect of the size of the
policy equilibrium interval on the variance of the agency officials’ policy choices. The
size of the policy equilibrium interval has the greatest effect when the agency preference
distribution is unskewed (Beta(l , 1)). Its effect is smallest when the agency preference
distribution is most skewed (Beta(0.25, 1)). That is, there is a negative relationship
between the skewness of the agency preference distribution and the effects of the policy
equilibrium interval’s size on the conditional variance of agency policy choices.

These theoretical experiments reveal subtler relations between preferences of
political institutions and a regulatory agency’s actions. First, even a given amount of
change in the politicians’ preference profile does not always bring about the same change
in the agency’s regulatory policy outputs. It is likely that political institutions’ effects on
agency actions are greater when the agency is skewed toward one of the limits of the
policy equilibrium interval and when the political institution establishing the limit
moves. By contrast, a change in the politicians’ preference profile may not substantially
change the agency’s policy outputs if the agency preference distribution is unskewed or
if the agency preference distribution is extremely skewed to one limit but the political
institution constituting the other limit moves.

Second, when there is a substantial disagreement on policy between the agency
and the political institutions, any change in the politicians’ preference profile may not
have a significant effect on the degree of bureaucratic autonomy. For instance, there is a
possibility that the agency preference distribution is extremely skewed so that agency
officials’ ideal points lie outside the policy equilibrium interval. In this case, even if the

policy equilibrium interval is large so that there are many feasible policy options for the

74

agency, the agency officials will end up having to choose fi'om a very limited range of
options. That is, the conflicts among the principals and agents tend to deprive the agents
of the potential opportunity to capitalize on an inter-principal disagreement even if there
is a substantial preference divergence among the principals.

Third, the second observation suggests that as the preference divergence among a
set of veto players and agency officials increases as a result of the agency’s policy bias,
we should predict agency officials’ behavioral compliance with changes in the
preferences of the veto players. This can be logically concluded from the fact that an
increase in the agency preference skewness (policy bias) leads to a greater effect of the
PEI’s location on the conditional mean of the agency’s policy outputs. When cheating is
not assumed to be an available option for bureaucratic agents, the policy disagreement
among principals and agents will reinforce the latter’s sensitivity to the former and at the
same time make inter-principal conflict increasingly irrelevant to the degree of

bureaucratic autonomy.

3.2. Veto Players

Having demonstrated the importance of joint actions of veto players to agency actions,
we need to determine which of those institutional actors should be included in the set of
veto players. In order to clarify the membership of the veto-player set, we should discuss
why a particular institutional actor should be considered along with other particular
institutional actors. Surprisingly, previous empirical research has not yet raised this
question. Researchers have considered numerous possible sources of political influence,

not only the institution-level actors such as the president, the House, and the Senate, but

75

also other intra-institutional actors such as congressional committees, legislative parties,
and congressional leadership. So which of these actors should be considered pivotal in
policymaking process? Which of these actors has a greater chance of being perceived as
crucial by agency officials?

The president can be considered as a permanent member of the veto-player set as
most work in the multiple-principal framework includes an independent executive veto
player (Hammond and Miller 1987; Calvert, McCubbins, and Weingast 1989; Hammond
and Knott 1996, 1999). The president possesses various powers such as the executive
veto over congressional bills, appointment and removal of top positions in the civil
service system, and budget requests. In addition to these formal powers, the president
has ample discretion and resources to take initiative actions to influence the make-up of
policies (Moe 1982; Moe and Howell 1999; Howell 2003; Lewis 2003). Presidential
influence on individual legislators also is considered substantial (Kemell 1997; Bond and
Fleisher 1990). It is almost impossible to ignore the pivotal role that the president may
play in important policymaking. It may be legitimate, then, to treat the president as a
political institution and a veto player rather than treating him as a part of the executive
agency (i.e., Shipan 2004).

However, Congress is different due to its complicated intra-institutional structure.
Congress consists of two separate chambers, each of which is based on different sets of
rules. And the operation of the chambers is decentralized to the system of congressional
committees. Thus, as Hall (1996: 2) put it:

Participation in Congress is seldom universal. It is never equal. Although most

(sometimes all) members vote when specific decisions come to a formal roll call

on the chamber floor, floor voting is only one and probably not the most
important form of participation in the legislative process. Building a coalition for

76

a legislative package, drafting particular amendments, planning and executing

parliamentary strategy, bargaining with or persuading colleagues to adopt one’s

point of view---all these activities weight more heavily than voting in the

decision-making calculus of most bills.

However, there is no clear agreement among researchers regarding which of these
various congressional actors should be highlighted in the legislative policymaking
process. We can examine a few major perspectives on the composition of important

congressional actors in the congressional literature, from which alternative sets of veto

players can be constructed.

3.2.1. The Majoritarian Perspective

The majoritarian perspective emphasizes the importance of majoritarian rules and the
median members in the chamber (Maass 1983; Krehbiel 1991, 1998). According to this
view, congressional decisions are determined for the most part by politics on the floor
rather than other intra-congressional organizations such as the congressional committee
system. Even if congressional committees retain policy expertise and information in
their jurisdiction, they are organized in the manner that can efficiently serve their parent-
chamber’s needs. In other words, it is the chamber as a whole that creates and maintains
organizational structure of Congress according to its interests.

The key strategy that can be employed by the chamber to maintain a degree of
discipline inside Congress is to manipulate the distribution of preferences in each
congressional organization (i.e., committee) through a variety of procedural tools. The
chamber can appoint members of the committees in order to minimize the possibility that
committees can challenge their parent chambers in two ways (Mass 1983; Krehbiel

1991). First, the chamber may assign members of committees whose median member

77

has preferences that are the same as the chamber’s median member. Then, committees’
decisions are likely to reflect the wishes of the chamber as a whole. Second, closely
related to the first, the chamber may appoint members of committees that are
representative of the full chamber. Decisions of these heterogeneous committees, thus,
are unlikely to deviate from what one can observe on the floor.

In addition to the Chamber’s ex ante manipulation of committee membership, the
chamber possesses formal authority to reverse committees’ decisions if the former finds
the latter inconsistent with its interests. Even if committees make legislative decisions
that may serve their own particularistic interests and send them to the floor, the chamber
as a whole can protect its policy goals simply by amending and defeating the
committee’s proposals on the floor. That is, the chamber can use floor votes as
safeguards against committees’ decisions that contradict the chamber’s preferences.

According to this perspective, policymaking in Congress is governed by the
principles of majoritarian decisions---the will of majorities of the chambers determines
policymaking in Congress.6 While leaving out the subtle difference of procedural rules
between the House and the Senate (see Binder 2003 for the evolution of the Senate to a
majoritarian institution), the median member of each chamber can be thought to best
represent institutional preferences. Then, we can consider the majoritarian veto-player
set as consisting of policymakers whose preferences represent their respective

institutions, such as the president, the House floor median, and the Senate floor median.

 

6 In fact, the extensive literature focuses on the supermajoritarian rules especially in the Senate such as 3/5
vote rule to invoke cloture for filibuster (Krehbiel 1991; Brady and Volden 1998). However, we cannot
underestimate the importance of simple majority rules in the congressional process. Both the House and
the Senate were originally created as majoritarian institutions in face of the fear that under the
supennajoritarian rules that had been used in the Continental Congress, “it would be no longer the majority
that would rule; the power would be transferred to the minority” (Federalist 58). In fact, individual
Senators’ exercise of their right to filibuster can be discouraged by various factors including Senate
leaders’ efforts, and most bills are processed under the simple majority rules in Congress (Binder 2003).

78

After all, this perspective may best portray the most fundamental feature of the
separation of powers and bicameralism: political influences are exerted on bureaucratic
actions as a result of a multilateral agreement among the majorities of veto institutions

(Hammond and Miller 1987).

3.2.2. The Distributive Politics Perspective

The distributive politics perspective highlights the role of congressional committees in
addition to the majoritarian median members (Fenno 1973; Weingast 1984; Shepsle and
Weingast 1987; Weingast and Marshall 1988; Miller and Moe 1983; Aberbach 1990;
Hall 1993). This perspective contends that each committee has strong interests and
autonomy in policy decisions under its jurisdiction. As Fenno (1973: xiii) put it:

[C]ommittees are autonomous units, which operate quite independently of such

external influences as legislative party leaders, chamber majorities, and the

President of the United States. . . .[E]ach committee is the repository of legislative

expertise within its jurisdiction; ...committee decisions are usually accepted and

ratified by the other members of the chamber;. . .committee chairmen can (and
usually do) wield a great deal of influence over their committees.

Congressional committees are said to be characterized by homogeneous
preferences among their members on issues under their respective jurisdiction and to
have outlying preferences (e.g., high demander) compared to the rest of the chamber.
That is, congressional committees are non-representative of the chamber as a whole. The
independent and non-representative committees can be seen as a result of a few factors.
First, the self-selected assignment process may reinforce the preferential bias of the
committees. The assignment of members to committees is thought to be determined

primarily by individual legislators’ reelection motivation and their constituents’

particular interests. Second, the committees’ insulation from the rest of the chamber can

79

stem from the committees’ symbiotic relationship developed with external actors such as
interest groups and agencies over a long period of time. Congressional committees may

make decisions in accordance with interests of these external principals rather than those
of the chamber and party caucuses.

Since the chamber defers to the committees’ monopoly over issues in their
jurisdictions and since the norm of mutual forbearance among different committees tend
to develop, committees can wield a strong gatekeeping power. Although the committee
proposal can be amended on the floor under an open rule, a risk-averse committee
median will vote not to send a bill to the floor if the expected floor amendment to the
committee bill will be worse than a status quo policy (Hammond and Knott 1996: 135-
6); that is, without consent of the committee median, no new policy can be passed.
Furthermore, the conference committees can be used by committee members to prevent
noncommittee members from amending their legislation, which may be called “ex post
veto” (Shepsle and Weingast 1987).

In sum, we can think of the set of veto players consisting of five key institutional
actors: the president, the two floor medians, the House committee median, and the
Senate committee median. The interactions among these actors may influence the
agency’s behavior in a way that can be distinct fiom how the majoritarian set of actors

influences the agency’s behavior.

3.2.3. The Party Government Perspective

The party government perspective emphasizes the importance of partisan goals and party

leadership in legislative decisions (Rohde 1991; Cox and McCubbins 1993). This

80

 

perspective contends that agents of party caucuses, including the party leadership, the
speaker, committee chairs, and party whips, can mobilize various parliamentary powers
to enhance their partisan interests. Cox and McCubbins (1993: 2) succinctly summarize
this perspective as follows:

[P]arties in the House---especially the majority party---are species of “legislative

cartel.” These cartels usurp the power, theoretically reside in the House, to make

rules governing the structure and process of legislation. Possession of this rule-
making power leads to two main consequences. First, the legislative process in
general---and the committee system in particular---is stacked in favor of majority
party interests. Second, because members of the majority party have all the
structural advantages, the key players in most legislative deals are members of
the majority party, and the majority party’s central agreements are facilitated by
cartel rules and policed by the cartel’s leadership.

In general, the influence of the parties on legislative policymaking can be
identified in two ways. First, party caucuses may assign their loyal contingent to
committees so that committee medians reflect the caucuses’ median. Through this
partisan selection process parties can ensure that committee decisions are in accordance
with partisan preferences (Cox and McCubbins 1993). Second, party caucuses may try
to discipline pivotal legislators such as committee and floor medians. Especially when
an issue is highly salient to party members and policy preferences become homogenized
within each party and polarized between parties, the majority party leadership can force
the committee medians and the majoritarian medians to be closely aligned with the
majority party median (Aldrich and Rohde 1999).

From this perspective, it is partisan politics that exerts the most decisive influence
on legislative processes. The behavior of the majoritarian and distributive politics set of

veto players in Congress may actually be governed by the concerted partisan efforts to

pursue partisan goals. One can expect the legislative majority party to force policy

81

outcomes to be in its own interest when inter-party conflict is intense. Thus, the set of
veto players can be thought to consist of five pivotal actors: the president, the House and
Senate majority party medians in the floor, and the House and Senate majority party

medians in the committee.

3.3. Issue Characteristics

Policy researchers have argued that the regular participants in policymaking can be
predicted by issue characteristics such as salience, complexity, and partisan interests
(Gorrnley 1986; Eisner, Worsham, and Rinquist 2000). These factors can influence the
motivation and interests of important policy actors to participate in the policy process.
Therefore, participants in policymaking can differ across policy areas and over time
depending on issue characteristics. In this section, I discuss the salience of the issue and
the degree of partisan polarization that may affect the motivation of elected officials to
attempt redirection of agency behavior. The influence of the three sets of veto players---
majoritarian, distributive, and party government sets---may be contingent upon changes

in salience and partisan interest of a policy issue.

3.3.1. Salience

The salience of an issue can change. Public attention and attitudes toward policy issues
may change at an almost imperceptible pace over a long period of time (Stimson 1991).
It may take generations to observe the shift of public attention from one issue to another
one. However, some monumental events may break the long-term equilibrium and bring

new issues to public attention. Disastrous accidents, scandals, or some great

82

 

achievements can suddenly attract intense public attention in a very short period of time
(Kingdon 1995; Baumgartner and Jones 1993; Wood and Waterman 1994). These
changes in issue salience have predictable policy effects. Heightened public attention to
particular policy issues can provide incentives for politicians to take actions about the
issue and collect information about the govemment’s previous policies and possible
policy changes (Gormley 1986; Kingdon 1995). Especially when an issue receives
national media attention and emerges on the national agenda, political leaders will
eagerly follow the sequence of issue development and seize the crucial moment to take
clear position. Politicians’ opportunity costs of engaging in such highly salient policy
issue will decrease because the issue can increase the politicians’ visibility and political

stakes, which compensates for the politicians’ time and resources spent on this issue.

3.3.2. Partisan Polarization

Policy issues that polarize partisan interests and increase partisan unity will motivate
political party leaders to step in to make some voices. Traditionally, some policy areas
such as welfare, education, environment, and labor have been considered as
battlegrounds between the competing partisan interests. These issues tend to mobilize
partisan interests to induce policy outcomes that increase partisan benefits. Partisan
interests can be mobilized at various levels from the local to the national. Local partisan
networks consist of core party activities, local officials, trade groups, and local offices of
the federal officials (Fenno 1978). These extensive partisan networks may compete with
each other to insert their interests into the policy implementation process (Scholz et al.

1991). When those partisan policy issues reach the national agenda, a vertical

83

integration of grass-root party networks may take place. Under these conditions, party
leaders at the national level will stand up to mobilize the national-level bases of partisan
supports.

In recent years, the role of the national party committees in aggregating local
partisan interests has grown. The national party committees control campaign money,
possess modern campaign expertise, and expand their role in local candidate selections.
The nationalization of the US. party system provides the opportunity for party leaders
not only to represent core party members’ ideology but also to put pressure on individual
legislators to modify their behavior (Fiorina 1989; Jacobs 2001). With enhanced intra-
party powers, the leaders of the majority party in Congress can use their superior
parliamentary powers to accomplish partisan goals (Rohde 1991).

These characteristics of policy issues thus can be expected to have considerable effects
on which of the competing sets of veto players will actively participate in policymaking
processes. In other words, the relative importance of the three sets of veto players may
be contingent on the saliency of the issue at hand and to the extent the issue polarizes
partisan interests (Maltzman 1997; Epstein and O’Halloran 1999, chapters 7 and 8;
Aldrich and Rohde 1999; Hurwitz, Moiles, and Rohde 2001). These issue characteristics
may affect the motivations of elected officials to participate in the policy process. Figure
6 presents my expectations about the effects of issue characteristics on the type of
regulatory politics. When the issue is highly salient, the chamber or the majority party
will be likely to force the committee median to be closely aligned with either the floor
median or the majority party median. When the issue divides interests between parties,

the majority party will be likely to make the majority party contingent in the committee

84

Figure 6. Issue Characteristics and Types of Regulatory Politics

 

Partisan Polarization
Weak Strong
Low Distributive
Issue Salience
High Majoritarian Party Government

 

 

 

85

 

and the chamber median more closely aligned to the majority party median.
Thus, when the issue is highly salient and divides the parties, it is likely that the
PEI resulting from interactions between the president and majority party medians in the
chambers and the committees will be most influential in affecting agency actions. When
the issue is highly salient but does not divide the parties, the president and two chamber
medians will make up the PEI. When the issue is neither salient nor party-dividing, the
PEI resulting from the interactions among the president, the chamber medians, and
outlying committee medians will be most influential.
H5: The effects of the majoritarian, distributive, or partisan PEI’s on an
agency’s regulatory actions vary according to the salience and the partisan
polarization of the policy issue.
HS-l: As issue salience increases and polarization decreases, the
majoritarian set of veto players will be more likely to affect the level and
variability of an agency’s regulatory actions.
H5-2: As issue salience and polarization decrease, the distributive set of veto
players will be more likely to affect the level and variability of an agency’s
regulatory actions.
H5-3: As issue salience and polarization increase, the partisan set of veto
players will be more likely to affect the level and variability of an agency’s
regulatory actions.
3.4. Statistical Model: Bridging the Formal/Empirical Gap
My spatial model of multiple principals and bureaucratic autonomy has proposed
that joint actions of political institutions affect bureaucratic actions in two ways: while
the location of the policy equilibrium interval (due to political institutions’ policy

preference) affects the level of regulatory policy outputs, the size of the policy

equilibrium interval (policy disagreement among political institutions) affects the

86

variability of bureaucratic choices. Therefore, empirical testing should focus on the
effect of the policy equilibrium interval on agency officials’ actions in two aspects: the
conditional mean of policy outputs and the conditional variance of policy choices.

The Maximum Likelihood Heteroscedastic Normal Regression Model (Franklin
1991; King 1998) is chosen to accomplish the goal of an appropriate empirical testing of
the spatial model. The Heteroscedastic Normal Regression Model is a good match to the
nature of propositions since it enables us to estimate determinants of not only the
conditional mean (like OLS) but also the conditional variance.7 If the dependent variable
Y.. is regulatory policy actions of an agency i at year t, we can use the following

likelihood function:8

:7”) ——7] (1)

i=1 :-1 H:_l271t0':e

Unlike the likelihood function for the OLS, this function includes the non-

constant variance (0*2 ) 1n addition to the mean ( it“). Since our purpose is to estimate

the institutional factors’ effects on both the mean and variance, we can reparameterize

the likelihood function as follows:

( X a)
L: ..._
l'lfl f ”expat/.13) zl expmm l (2’

 

 

 

 

7 For the statistical approach that focuses on the stochastic components in the political science literature,
see Franklin (1991), Jacoby (1988), Tsebelis (1999), Alvarez and Brehm (1995), and Paolino (2001).

' This likelihood function depends on the Normal distribution. With the Normal distribution, we should
assume that the agency officials’ preferences are distributed symmetrically and centered on its mean.
However, in the previous discussion of spatial model, the agency officials’ preferences were assumed to be
spread with the Beta distribution. Despite this difference, simulations show that the predictions regarding
the conditional mean and variance remain the same with these two different distributions.

87

 

Now, we have one equation for the conditional mean ,U..= ..a and the other equation for

the conditional variance 0i: exp( W..fl). In other words, the conditional mean (the level

of regulatory policy outputs) and the conditional variance (the variability of agency’s
regulatory choices) are assumed to be determined by two different sets of variables X..
and W.., respectively. Then we can estimate parameter vectors [3 and a by maximizing

the following log-likelihood function:

_1 (Yr: —Xtta)2
:72: I. .1 fl will

2i=li=12t=li=l

 

 

 

 

Model specifications for these two equations can be straightforward. First, the
conditional mean equation can be specified in a partitioned form as follows to test the
hypotheses about the effect of the location of the policy equilibrium interval on the level

of regulatory policy outputs or bureaucratic responsiveness:

Policy Output Level ([1,. ) = a.) + alX... + a2X..2 (5)
where X... includes a set of measures on the location of the policy equilibrium interval in
the policy space (i.e., lower and upper limits) and issue salience and X..2 is a set of other
control variables. More specifically,

X... = Location of the PE], Issue characteristics,

Location of the PEI *Issue characteristics
X..2 = Control variables

88

Second, the specification for the conditional variance equation can be done in a
partitioned form as follows to test the effect of the size of the policy equilibrium interval

on the variability of regulatory policy choice or the degree of bureaucratic autonomy:
Variability of Policy Choices (0;? ) = exp(flo + )9. W... + fl2 Wm) (6)
where W... is a set of measures on the size of the policy equilibrium interval and issue

salience and W..2 includes a set of other control variables.

W... = Size of the PEI, Issue characteristics, Size *Issue characteristics
W..2 = Control variables

In order to examine the non-constant institutional effects that are theoretically
predicted to vary with the skewness of the agency preference distribution, I use Quantile
Regression (Koenker and Bassett 1978; Buchinsky 1994). Quantile Regression, via the
Minimum Distance (MD) algorithm, enables us to estimate institutional effects at
different quantiles of the entire distribution of the dependent variable (Y..) or regulatory
policy actions of agencies.9

Following Koenker and Bassett (1978), the 6th conditional quantile of the
dependent variable Y. given a vector of X is given by:

Q9(Y.|X.)=Xbo,wherei= l,...,n and0<6<l (7)

And, it is known that the 6th population quantile can be defined as a solution to

the following minimizing problem:

 

9 Koenker and Bassett (1978) contend that the MD estimators are useful especially when the Gaussian
assumption of constant variance of the disturbances (homoscedasticity) is not maintained. However, the
merit of Quantile Regression has been found in regard primarily to its flexible applicability rather than its
robustness itself. For instance, Buchinsky (1994) uses Quantile Regression to examine different (wage)
returns to workers’ education and experience at different quantiles of the distribution of wages. Koenker
and Hallock (2001) review applications of this model in various contexts.

89

1143;. Z 9|Y.-pl+ 2(1-6)|Y.-p| (8)
KZP K<p

From this, we can consider the 0th conditional quantile, Qg( Y.|X.-), as a solution to

the following minimizing problem: ’0

Arm 2 9|Y.-pI+ 2(1-9)IY.-pllX. (9)
KZp K<p

Since Q9(Y.IX1) = ng, the solution to the problem is ng. We can estimate by for
a given X at any 6th quantile, and unsurprisingly be can differ at different 0’s.
Figure 7 graphically illustrates how we can examine the non-constant institutional effects
by employing Quantile Regression. AL represents a regulatory agency that is skewed to
the minimal regulation, AM represents a regulatory agency that is unskewed, and AH
represents a regulatory agency that is skewed to stronger regulation. Since these
agencies’ regulatory actions approximate their officials’ preference distribution, these
agencies can be mapped onto the anti- and pro-regulation space as AL<AM<AH. We‘can
construct the distribution of agencies based on the heteroscedastic agency preference
distribution in this way. Then, we can assume that the lower tail of the distribution of
agencies consists of agencies that are biased to deregulation, the middle consists of
agencies that are unskewed, and the upper tail consists of agencies that are biased to
regulation. Therefore we can expect that the effects of the institutional preferences

(location) on a regulatory agency’s policy outputs (level) will be greater at the lower and

upper tails than at the middle.

 

'° Median Regression is a special case where 6 is set to 0.5, which yields the Least Absolute Deviation
(LAD) estimator.

90

Figure 7. Mapping Agencies onto Policy Continuum

 

 

 

 

 

 

 

Note: AL: Skewed to the minimal regulation
AM: Unskewed
AH: Skewed to the maximal regulation

91

In sum, these statistical approaches provide the opportunity to examine the
concurrence of bureaucratic responsiveness and bureaucratic autonomy and the
mediating role of the agency preference distribution. By using the ML Heteroscedastic
Normal Regression Model, we can test fundamental dynamics that underlie the two
processes (responsiveness and autonomy) as identified by the spatial model in one
integrative framework. By using Quantile Regression, we can account more accurately

for the mediating impact of bureaucratic preferences on those institutional effects.

92

CHAPTER 4
POLITICAL INSTITUTIONS AND OCCUPATIONAL SAFETY AND HEALTH

REGULATION

The substantive focus of this dissertation is on federal and state occupational safety and
health regulation. The Occupational Safety and Health Act (OSH Act) was enacted in
1970. The statutory goal of the OSHA Act was “to ensure safe and healthful working
conditions for working men and women.” The Occupational Safety and Health
Administration (OSHA) was created in the Department of Labor to enforce the law. The
agency was given the authority to promulgate and enforce national standards for
workplace conditions. State governments could avoid the promulgation of federal
workplace rules by obtaining OSHA’s approval for their own program on the condition
that the state programs be “as effective as” federal programs in protecting workers from
industrial risks and hazards. In the fiscal year 2004, OSHA has an authorized staff of
2,220, including 1,123 inspectors and appropriations of $457.5 million.

In this chapter I describe the regulatory structure in which state and federal
OSHA agencies operate and institutional environments in which political institutions---
the president and congressional actors---interact with each other to influence the agency.
In the previous chapter, I argued that OSHA inspectors’ regulatory policy choices would
be influenced by the interactions among political institutions. I also proposed that those
political institutions could be intertwined with each other in three different ways.
According to the majoritarian perspective, the important features of the inter-institutional
interactions are the executive veto and bicameralism. Due to these constitutional rules,

the set of the president, the House median, and the Senate median may collectively

93

influence OSHA. Second, the distributive politics perspective emphasizes the role of the
congressional committee in addition to the majoritarian actors. In the case of OSHA, the
role of House and Senate Labor Committees and the House and Senate Appropriations
Subcommittees can be closely examined. Third, according to the party government
perspective, the House and Senate median members and Committee median members
will behave in a close alignment to the majority party median. Since OSHA has been a
controversial issue that has divided among labor and business interests from the
beginning, we can examine whether the actual interactions among these political
institutions in fact reflect the partisan conflicts in the legislative policymaking process.
In the first part I describe the Occupational Safety and Health Act of 1970 and the
functions of OSHA as defined by the law. This overview will provide information about
OSHA officials’ authority and tasks in enforcing the occupational safety and health rules.
In the second part I describe how the political institutions interacted with each other to
influence agency policy and behavior throughout the legislative history of bills and
provisions that were related to occupational safety and health policy between 1968 and

2000.

4.1. The Occupational Safety and Health Act of 1970

The Occupational Safety and Health Act (OSH Act) of 1970 was the first federal law that
established comprehensive national standards for job safety and health in all industries.
The OSH Act authorized the creation of three agencies to set and enforce mandatory
health and safety standards, to conduct research on occupational hazards and their

control, and to review contested enforcement actions. The three agencies were the

94

Occupational Safety and Health Administration (OSHA), the National Institute of
Occupational Safety and Health (N IOSH), and the Occupational Safety and Health
Review Commission (OSHRC).

OSHA is a regulatory agency that sets and enforces regulations concerning the
control of health and safety hazards in workplaces. It began its operation as a part of the
Department of Labor on April 28, 1971. The agency is headed by the Assistant
Secretary of Labor for Occupational Safety and Health, who is appointed by the
president with the Senate’s consent. The Secretary of Labor has delegated authority to
the Assistant Secretary to administer the OSH Act. OSHA sets mandatory health and
safety standards, inspects workplaces to ensure compliance, and proposes penalties and
prescribes abatement plans for employers who are found violating the standards. OSHA
also provides for public, worker, and employer education and consultation, mostly
through grant activities. OSHA partially supports the operations of state agencies
operating state plans up to 50 percent and monitors their performance.

NIOSH is a research agency within the Center for Disease Control (CDC) of the
US. Public Health Service in the Department of Health and Human Services (HHS).
NIOSH is headed by a director appointed by the Secretary of HHS for a term of six years.
It was created from what had been the Bureau of Occupational Safety and Health and
started operations as NIOSH on June 30, 1971. It conducts research and related
activities on developing criteria or recommendations to be used by OSHA in setting
standards, on identifying and evaluating workplace hazards, and on measurement
techniques and control technologies, as well as providing professional education and

disseminating health and safety information.

95

The OSHRC has three members appointed by the president, with the advice and
consent of the Senate, for staggered terms of six years. Its duties are limited to
reviewing and resolving disputes concerning OSHA citations and penalties. In doing so,
the Commission can interpret the meaning of OSHA standards and thus determine the

nature and scope of many employers’ obligations concerning employee health and safety.

4.2. Functions of the Occupational Safety and Health Administration
Major functions of OSHA include standard-setting rulemaking, enforcement of the

standards, and assistance to employers’ compliance with the standards.

4.2.1. Standard-Setting Rulemaking

The OSH Act has provided OSHA with three different methods to establish health and
safety standards. First, OSHA could issue startup (interim) standards. Section 6(a) of
the Act provided the ground for interim standards: “the Secretary shall, as soon as
practicable during the period beginning with the effective date of this Act and ending
two years after such date, by rule promulgate as an occupational safety or health standard
any national consensus standard, and any established Federal standard, unless he
determines that the promulgation of such a standard would not result in improved safety
or health for specifically designed employees.” Second, OSHA could issue permanent
standards. Section 6(b) authorized the Secretary to issue new standards and to modify or
revoke existing ones through informal rulemaking through a multi-step process: the
receipt of a criteria document from NIOSHA, with reports from employers, labor unions,

or academic concerning a hazard, or with a petition for a standard from an interested

96

group. Lastly, OSHA was authorized to issue Emergency Temporary Standards (ETS)
under section 6(c) of the Act. An ETS could be issued if OSHA determined that
employees were exposed to a “grave danger” (6(c)(1)(A)) and that an emergency
standard was “necessary to protect employees from such danger”(6(c)(l)(B)).

OSHA currently has health standards for about 410 chemical substances. In most
cases, OSHA specifies the maximum levels for employee exposure, which is called the
Permissible Exposure Limits (PELs). Most of these standards were adopted in 1971 by
OSHA under section 6(a) as consensus standards or “startup standards.” These
consensus standards had been developed by industrial safety engineers and hygienists as
models of the best practice.

In addition to the initial lump-sum adoption of private standards, OSHA has
issued 34 health standards between 1971 and 2001 (Table 1). The number of new
standards remained almost constant over periods of 10 years: 10 in the 19705, 11 in the
19808, and 13 in the 19903. In addition to standards for hazardous chemicals, OSHA has
also set standards for other issues such as electrical, mechanical, fire protection,
construction, and maritime safety in workplaces. Between 1971 and 2001, OSHA issued

52 final safety standards (Table 2).

4.2.2. Enforcement

Inspections and enforcement are the heart of the regulatory scheme in the OSH Act. The
goals of enforcement are to correct identified hazardous conditions in inspected plants
and to provide an incentive for other plants to take appropriate actions to control hazards.

Procedures of OSHA inspections are outlined in several manuals. During inspections,

97

Table 1. OSHA Health Standards (1972- 2001)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Toxic and Hazardous Substance Final Standard
Asbestos 06/ l 972
13 Carcinogens 01/1974
Vinyl chloride 10/1974
Coke oven emissions 10/1976
Benzene 02/1978
DBCP 03/ l 978
Inorganic arsenic 05/1978
Cotton dust 06/1978
Acrylonitrile l 0/ 1978
Lead ll/1978
Cancer policy 01/1980
Access to employee exposure/medical records 05/1980
Occupational noise exposure 01/1981
Hazard communication 01/1983
Ethylene oxide 06/1984
Asbestos - revision 06/1986
Field Sanitation 05/1987
Benzene — revision 09/ 1 987
Formaldehyde l 2/ l 987
Access to employee exposure/medical records - revision 09/1988
Air contaminants PELs update 01/1989
Hazardous chemicals in laboratories 01/1990
Bloodbome pathogens 12/1991
4,4-Methylenedianiline 08/ 1 992
Cadmium 06/1993
Asbestos - court remand 06/1992
Formaldehyde - court remand 05/1992
Lead — construction 05/1993
Asbestos 08/1 994
1 ,3-Butadiene 1 1/1 996
Methylene Chloride 01/1997
Respiratory protection 01/1998
Ergonomics l 1/2000
Bloodbome pathogens — revision 01/2001
Ergonomics — revoked 06/2001

 

 

Source: US. Congress, Office of Technology Assessment. 1985. Preventing Illness and
Injury in the Workplace. p. 228; Subcommittee on Workforce Protections of the
Committee on Education and the Workforce, House of Representatives. The 107‘h
Congress. “The OSHA Rulemaking Process.” pp.106-128.

98

Table 2. OSHA Safety Standards (1972-2001)

 

Safety Standards

Final Standard

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cranes/derricks (load indicators) 07/1972
Roll-over protective structures (constructioy 04/1972
Power transmission and distribution 1 l/ 1972
Scaffolding, pump jack, and roof catch platform 12/1972
Lavatories for industrial employment 05/1973
Trucks, cranes, derricks, and indoor storage 06/1973
Temporary flooring—skeleton steel construction 07/1974
Mechanical power presses 12/1974
Telecommunications 03/1975
Roll-over protective structures for agricultural tractors. 04/1975
Industrial slings 06/1975
Guarding of farm field equipment 03/1975
Ground-fault protection 12/1976
Commercial diving operations 07/1977
Servicing multi-piece rim wheels 01/1980
Fire protection 09/1980
Guarding of low-pitched roof perimeters 11/1980
Design safety standards for electrical standards 01/1981
Latch-open devices Qn gasoline pumps) 09/1982
Marine terminals 07/1983
Servicing of single/multi-piece rim wheels 02/1984
Electrical safety in construction 07/1986
General environmental controls 09/1986
Marine terminals servicing single piece rim wheels 09/1987
Grain handling facilities 12/1987
Safety testing of certification of equipment 08/1988
Crane or derrick suspended personnel platforms 08/1988
Concrete and masonry construction 06/1988
Mechanical power presses — modified 03/1988
Powered platforms 07/1989
Underground construction 06/1989
Hazardous waste operations 03/1989
Excavations l O/ 1989
Control of hazardous energy sources 09/1989
Stairways and ladders 11/1990
Concrete and masonry lift-slab operations 10/1990
Electrical safety work practices 08/1990
Welding, cutting, and brazing 04/1990
Chemical process safety 02/1992
Confined spaces 01/1993
Fall protection 08/ 1994
Electrical power generation 01/1994
Retention of DOT markings, placards, and labels 07/1994

 

 

99

 

Table 2 (Cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

Safety Standards Final Standard
Personal protective equipment 04/1994
Logging operations 10/1994
Scafi‘olds 08/1996
PPE for shipyards 05/1996
Longshoring and marine terminals 07/1997
Powered industrial truck operator training 12/1998
Confined spaces — amended 12/1998
Dipping and coating - rewrite 07/1999
Steel erection 01/2001

 

 

Source: US. Congress, Office of Technology Assessment. 1985. Preventing Illness and
Injury in the Workplace. p. 229; Subcommittee on Workforce Protections of the
Committee on Education and the Workforce, House of Representatives. The 107th
Congress. Hearing on “The OSHA Rulemaking Process.” June 2001. pp.106-128.

 

100

OSHA inspectors look for any violations, following the procedures and interpretations
issued by OSHA in its Field Inspection Reference Manual (FIRM), Field Operations
Manual (FOM), Industrial Hygiene Manual, and program guidelines.

OSHA conducts several different kinds of inspections. The most basic
distinction is between inspections for safety hazards and those for health hazards. OSHA
Compliance Safety and Health Officers (CSHO) consist of safety inspectors and health
inspectors. Safety officers are specialists at industrial safety engineering and enforce
OSHA safety rules; health officers are certificated industrial hygienists and enforce
health rules concerning toxic and hazardous chemical substances.

OSHA classifies its inspections by priority (FIRM Ch. I. B(3)). The order of
priority is a) imminent danger, b) fatality and catastrophe investigations, c) complaints
and referrals investigations, and d) programmed inspections. Imminent danger
inspections are conducted when OSHA learns of a hazard that can be expected to cause
death or serious physical harm before it could be eliminated through normal enforcement
activity. Catastrophe and fatality investigations are spurred by reports of fatal
occupational injuries or of incidents that result in the hospitalization of five or more
employees. The third priority is employee complaints. Under section 8(t) of the OSH
Act, employees and their representatives who believe that an employer is violating a
health and safety standard may request an inspection. OSHA schedules inspections to
respond to what it determines are valid complaints. The lowest priority inspections are
programmed ones. These programmed inspections focus on industries with high injury

rates or those with known health hazards.

101

Under the OSH Act, employers are responsible for compliance with OSHA
standards. They have two major duties. First, unless they have obtained a variance,
employers are required by section 5(a)(2) of the OSH Act to comply with the terms of
OSHA standards and regulations (so called “specific duty clause”): “Each employer shall
comply with occupational safety and health standards promulgated under this Act.”
Second, employers must also comply with section 5(a)(1) of the Act (so called “general
duty clause”): “Each employer shall furnish to each of his employees employment and a
place of employment which are free from recognized hazards that are causing or are
likely to cause death or serious physical harm to his employees.” To prove a violation of
the general duty clause, OSHA must demonstrate that the employer failed to render the
workplace free of a “recognized” hazard that was causing or was likely to cause death or
serious physical harm. OSHA and the courts have held that a hazard is recognized if it is
of common knowledge in the industry in question and detectable by the senses or by
techniques generally known and accepted by the industry.

When there is a disagreement over OSHA enforcement, any affected parties can
take formal actions within first 15 days (OSH Act Sec. 10; FIRM Ch. IV). An employer
who disagrees with OSHA concerning a citation, a proposed penalty, or the date for
abatement of the hazard can file a “notice of contest.” Employees also have an
independent right to contest the reasonableness of the length of the proposed period of
time for abatement of a hazard. When contested, a hearing is held before an
Administrative Law Judge (ALJ) who is an employee of the OSHRC. The ALJ
examines the evidence and decides whether to affirm, vacate, or modify OSHA’s citation

and penalties. After this decision, any party can petition the OSHRC to review the

102

 

decision of the AL]. The Commission can grant such review either upon request or by
its own choice. Unless it is ordered to be reviewed within 30 days by a member of
OSHRC, the ALJ decision becomes a final order of the Commission. Any person
adversely affected or aggrieved by an order of the Commission can petition a US. Court

of Appeals for a judicial review within 60 days (OSH Act Sec. 11).

4.2.3. Compliance Assistance

Section 2(b)(2) of the OSH Act encourages “employers and employees in their efforts to
reduce the number of occupational safety and health hazards at their places of
employment, and to stimulate employers and employees to institute new and to perfect
existing programs for providing safe and healthful working conditions.” Moreover,
section 21(c) provides specific authorization for OSHA to conduct “education and
training of employers and employees in the recognition, avoidance and prevention of
unsafe or unhealthful working conditions” and to “consult with and advise employers
and employees, and organizations representing employers and employees as to effective
means of preventing occupational injuries and illnesses.”

To meet these statutory goals, OSHA currently operates various educational and
cooperative programs. For education and training, OSHA has developed its Outreach
Training Program, which trains individuals to teach others the basics of occupational
safety and health. Afier taking a one-week course, trainers teach 10-30 hour courses in
construction or general industry safety and health standards. OSHA Training Institute
(OTI) Education Centers, located throughout the country, offer the public 15 courses and

one seminar developed by the OTI staff on the topics of industrial hygiene,

103

 

recordkeeping, ergonomics, and construction-related issues. OSHA also presents
satellite broadcasts and simultaneous web-casts on new rules and guidance documents.
OSHA awards training grants under the Susan Harwood Training Grant Program to
nonprofit organizations to train workers and employers to recognize, avoid, and prevent
safety and health hazards in their workplaces.

As for cooperative programs, OSHA offers a variety of programs that enable the
agency to work cooperatively with trade or professional organizations, businesses, labor
organizations, educational institutions, and government agencies. OSHA’s Alliance
Program enables trade or professional organizations, businesses, labor organizations,
educational institutions, and government agencies that share an interest in workplace
safety and health to collaborate with OSHA to prevent injuries and illnesses in the
workplace. OSHA and the organizations sign formal agreement with goals that address
training and education, outreach and communication, and promotion of the national
dialogue on workplace safety and health. OSHA’s Onsite Consultation Program helps
small businesses-«particularly those in high-hazard industries or involved in hazardous
operations---to meet their obligations under the OSH Act and federal and state standards
by conducting onsite surveys. Consultants also help employers develop and implement
safety and health management systems that eliminate or control injuries and illnesses.
The program recognizes small employers that operate exemplary safety and health
management systems through the Safety and Health Achievement Recognition Program
(SHARP).

The OSHA Strategic Partnership Program (OSPP) tries to bring together

employers, employer groups, employees or their unions, and OSHA to address specific

104

safety and health issues. An agreement between partners may be local, regional, or
national in scope. A partnership agreement sets measurable goals and individual
responsibilities, specifies an action plan and a measurement system, and provides
procedures for verifying results. OSHA also has Voluntary Protection Programs (VPPs)
that include a rigorous on-site initial evaluation by expert OSHA teams to verify site
performance and regular evaluations afterwards. VPPs focus on comprehensive safety
and health management systems that go well beyond OSHA standards. In addition,
Compliance Assistance Specialists (CASS) in each OSHA area office respond to requests
for off-site safety and health assistance. Small businesses, trade associations, union
locals, and community and faith-based groups may call on CASs to speak at seminars,

workshops, and other events.

4.3. State Occupational Safety and Health Programs

Many who supported passage of the Occupational Safety and Health Act in 1970
believed that a federal program was necessary because the existing ‘patch-work’ system
of state programs was inadequate. But during congressional deliberations, a coalition of
business leaders, state government officials, a number of members of Congress, and the
Nixon administration pushed for the inclusion of ‘state plan’ programs in the OSH Act.
The OSH Act established a mechanism that enables states to regulate worker health and
safety subject to federal monitoring and approval. Section 18 of the Act declares that a
state program must, in general, “provide for the development and enforcement of safety

and health standards which are or will be at least as effective” as federal standards.

105

Under the OSH Act, states and territories can develop their own occupational
safety and health programs, which must be approved and monitored by the federal
OSHA (Sec. 18(c)). OSHA provides up to 50 percent of an approved state program’s
operating costs. The development of a state program is a stepwise process. After
application and initial approval on the development plan from OSHA, the state can begin
to enforce health and safety standards. States can adopt the existing federal standards.
States also can enforce their own standards, which OSHA has deemed to be “at least as
effective” as the corresponding federal ones. The first 3 years of a state program are
called the “developmental stage.” At the time of initial plan approval, both OSHA and
the state agency have concurrent jurisdiction—both have the authority to conduct
inspections and cite employers, and employers must therefore comply with both federal
and state standards. Any time after initial approval, however, as soon as the state is
“operational,” OSHA may suspend its concurrent enforcement jurisdiction through an
“operational status agreement.” After all developmental steps are completed OSHA can
issue a certification of the plan. If the state meets all of OSHA’s requirements, it
becomes eligible for “final approval” one year after certification. After the final
approval, the concurrent federal enforcement authority is relinquished under section
18(e) of the Act.

The requirement that states maintain a program “at least as effective” as federal
program means that if OSHA issues new or revised regulations, state agencies also must
follow the change, issuing an equivalent change, or making the case that there is no need
to alter the regulation. In all stages of its operation, OSHA monitors the quality of the

state program. Monitoring may involve “spot checks”—inspections by federal personnel

106

after a state inspection, or “accompanied” monitoring visits in which federal personnel
observe a state inspector during an inspection.

As of 2004, 21 states and 2 US. territories have “state plan” programs covering
both the private and public sectors (Table 3). Of the 23 state programs, 15 states,
including South Carolina, Utah, North Carolina, Iowa, Minnesota, Maryland, Tennessee,
Kentucky, Alaska, Nevada, Hawaii, Indiana, Wyoming, Arizona, and Virginia, have
obtained OSHA’s final approval status. The final approval status for the Virgin Islands
was disapproved. Other states and territories, including Oregon, Washington, California,
Michigan, Vermont, New Mexico, and Puerto Rico, have obtained OSHA’s certification
status. In addition to those comprehensive state programs, 3 states such as Connecticut,
New York, and New Jersey have obtained OSHA’s approval on partial programs
covering government employees only. Twenty-nine states and territories and the District

of Columbia that do not have their own programs are covered only by the federal OSHA

program.

107

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 3. State Occupational Safe 7 and Health Programs
State State. OSHA Initial OSHA. OSHA Final
Legrslatron Approval Certification Approval

S. Carolina 06/12/1973 1 1/30/ 1973 08/03/1976 12/15/1987
Oregon 07/01/ 1973 12/28/1972 09/15/1982

Utah 07/01/1973 01/04/1973 11/19/1976 07/16/1985
Washington 03/ 19/ 1973 01/26/ 1973 01/26/1980

N. Carolina 05/01/1973 01/26/1973 10/05/1976 12/10/1996
Iowa 07/01/1972 07/20/1976 09/14/1976 07/02/ 1985
California 09/1973 04/24/1973 08/12/1977

Minnesota 01/25/1973 05/29/1973 09/28/1976 07/30/1985
Maryland 07/01/1973 07/05/1973 02/15/1980 07/18/1985
Tennessee 07/01/1972 07/28/1973 05/03/ 1978 07/22/1985
Kentucky 03/27/1972 07/31/1973 02/08/1980 07/13/1985
Alaska 07/24/1 97 3 07/31/1973 09/09/1977 09/26/1984
Virgin Islands 02/26/ 1974 09/ 1 1/ 1973 09/22/1981 1 1/ 1 3/ 1995

Disapproved

Michigan 07/18/1974 09/24/1973 01/13/1981

Vermont 04/03/1974 10/01/1973 03/04/1977

Nevada 1 1/05/1973 12/28/1973 08/13/1981 04/18/2000
Hawaii 05/16/1972 12/28/1973 04/26/1978 04/30/1984
Indiana 05/01/1973 02/25/1974 10/16/1981 09/26/1986
Wyoming 01/01/1974 04/25/1974 12/30/ 1980 06/27/1985
Arizona 09/1974 10/29/ 1974 09/18/1981 06/20/ 1985
New Mexico 09/1975 12/04/1975 12/04/1984

Virginia 02/1973 09/23/1976 08/15/1984 1 1/30/1988
Puerto Rico 07/07/ 1975 09/07/1982

 

 

Source: 29 The Code of Federal Regulation (CF R) PART 1952, Approved State Plans
for Enforcement of State Standards, pp. 19-123.

108

4.4. Institutional Environments of OSHA

In this section I describe the history of the legislative decisions on bills and statutory
provisions related to OSHA between 1968 and 2000 to show how political institutions
interacted with each other to change occupational safety and health policy. I examine the
interactions among institutional actors in several legislative events: the enactment of the
Occupational Safety and Health Act (1968 -1970), the small business exemption issues
in

the Labor-HEW Appropriations bills (1972-1980), the issue of workers’ right to know in
the deliberation of the Risk Notification bill (1986-1988), and the overhaul and reform of
OSHA (1992-1996). Narratives of these legislative actions are aimed at substantively
revealing messages generated by the political institutions and received by OSHA in
regards with what kind of policy options were politically permissible. These descriptions
will also portray how the interactions among the president, the congressional committees,
and the parties have dynamically evolved surrounding OSHA issues over the long period

of time.

4.4.1. Enactment of the Occupational Safety and Health Act (1968-1970)

The enactment of the Occupational Safety and Health Act of 1970 took three years. All
institutional actors such as the president, the House, the Senate, and congressional
committees played important parts in the legislative process. Presidents Johnson and
Nixon provided clear policy agendas to be discussed in Congress, and congressional

actors such as the Labor Committees, floor majorities, and the parties engaged each other

109

to make their own views reflected in the legislative outcome. Unsurprisingly, OSHA
was created as a result of the interactions among those institutional actors.

In his message to Congress on January 23, 1968, President Johnson addressed
work safety problems. Afier declaring that “it is to the shame of a modern industrial
nation, which prides itself on the productivity of its workers” that each year 14,500
workers were killed and 2.2 million workers were injured, he promised to launch an
“attack” against the “source of the evil” (CQ Almanac 1968: 35-A). He acknowledged
that the federal government had offered no sufficient safeguards for workforces: safety
standards were narrow; research lagged behind; enforce programs were weak; safety
specialists fell far short of the need. He called government programs for job safety a
“patchwork of obsolete and ineffective laws.”

President Johnson’s proposal (HR 14816 and S 2864) was to empower the
Secretary of Labor to issue mandatory national standards for employee health and safety
and to close down plants if any imminent danger had been found. Federal inspectors
would be authorized to enter plants and order the immediate cessation of harmful
practices upon the Secretary’s determination. These decisions on the violations of the
standards could be reviewed by the US. courts. The preexisting state standards could be
maintained only if the Secretary approved them as being as effective as the federal rules.

Several hearings were held in the House and Senate Labor Committees in 1968.
In the first hearing held by the Senate Labor and Public Welfare Subcommittee on Labor
on February 15, 1968, Labor Secretary W. Willard Wirtz contended that “This country
faces a safety problem that has been with us for so long that as a nation we have become

impervious to its tragedy. The causality list is large and growing, yet the public voice

110

has been too little raised.” He emotionally continued, “Every minute we talk, 18 to 20
people will be hurt severely enough to have to leave their jobs, some never to work
again” (CQ Almanac 1968: 675). However, the Johnson bill met strong opposition by
powerful business groups including the Chamber of Commerce of the United States, the
National Association of Manufacturers (NAM), the American Iron and Steel Institute,
the American Medical Association, the Manufacturing Chemists’ Association, and
others. Throughout the spring and summer of 1968, business lobbyists made frequent
visits to key members of the House and Senate committees. The Chamber of Commerce
even assigned William J. Famrin, son of Senator Paul J. Famrin (R-Arizona), who was a
member of the Senate Labor Committee, to try to block the bill (CQ Almanac 1968:
677). Critics claimed that the Johnson bill would lead to arbitrary decisions by the
empowered Secretary and enormous economic costs and would bring the federal
government into an area in which state governments had operated their own programs.
Amid intensive business lobbying, the House Education and Labor Committee
replaced the original Johnson bill with a more lenient modification (HR 14816). This
bill, introduced by William D. Hathaway (D-Maine) and cosponsored by 12 Democratic
members of the Committee, reduced the power concentrated in the hands of the Labor
Secretary; the Secretary was allowed to promulgate preexisting “consensus” standards
developed by industrial standards-setting organizations and to appoint advisory
committees for drawing up new standards; penalties for violations became subject to
judicial review; a court injunction was required to close down a plant. The House Labor
Committee reported the amended bill on July 16. However, the House Rules Committee

never cleared the bill for floor consideration and the bill died. The Senate Labor

111

Committee held a series of hearings on the presidential bill (S 2864), but it did not take

further action.

Newly-elected President Nixon could not ignore the growing concerns about

workplace conditions which resulted in casualties. In his message to Congress on

occupational safety on August 6, 1969, President Nixon urged Congress to take action to

improve job safety. He remarked,

The side effects of [technological] progress present special dangers in the
workplaces of our country. For the working man and woman, the by-products of
change constitute an especially serious threat. Some efforts to protect the safety
and health of the American worker have been made in the past both by private
industry and by all levels of government. But new technologies have moved
even faster to create newer dangers. Today we are asking our workers to perform
far different tasks from those they performed five or fifteen or fifty years ago. It
is only right that the protection we give them is also up-to-date. (CQ Almanac

1 969: 68-A)

Although President Nixon acknowledged the urgent need for workplace

protection as did his predecessor, he had a different plan. In his message, Nixon

proposed the following specific recommendations:

Safety and health standards would be set by a new National Occupational
Safety and Health Board.

The Secretary of Labor would have the initial role in enforcing the standards
which the Board establishes.

The state governments would be encouraged to submit plans for expanding
and improving their own occupational safety and health programs.

The Secretary of Health, Education and Welfare would be given the specific
assignment of developing and carrying out a broad program of study,
experiment, demonstration, education, information, and technical assistance.
A National Advisory Committee on Occupational Safety and Health would be
established to advise the Secretary of Labor and the Secretary of Health,
Education and Welfare in the administration of the Act.

The Nixon bills (HR 13373 and S 2788) were to establish an independent board

to administer the federal regulation. The powers of the Secretary of Labor were limited

to enforcing standards set by the independent board. To counter the Republican version,

112

liberal Democrats inserted their own versions (HR 3809 and S 2193) into the legislative
process. Similar to the Johnson proposal, these bills authorized the Secretary of Labor to
set up and enforce safety and health standards while leaving the responsibility for
research on those standards to the Secretary of Health, Education and Welfare.

In the following hearings, business interests expressed opposition to both
versions. For instance, in the House Committee Hearing on October 16, 1969, J. Sharp
Queener from the Chamber of Commerce urged Congress to dilute the Nixon proposal
by making the federal program apply only where the states failed to act and by granting
the power to use injunctions for violations to the courts rather than the Secretary of
Labor. He emphasized that the national standards should only be guidelines, not
permanent rules. Wayne T. Brooks from the American Iron and Steel Institute
contended that the federal regulation “moves into the very essence of industrial and
business operation” (CQ Almanac 1969: 569). Endorsement of the administration bill
came from some industrial health organizations such as the American Academy of
Occupational Medicine and the American Industrial Hygiene Association and some
insurance company associations such as the American Insurance Association and the
American Mutual Insurance Alliance.

In contrast, organized labor interests were united in support of the Democratic
bills. In the House Committee Hearing on October 15, 1969, Andrew J. Biemiller,
director of legislation for the AF L-CIO, supported the Democratic bills that empowered
the Labor Secretary and criticized the Nixon bills on the ground that the proposed
independent board would be “another layer of bureaucracy, remote from the problems of

workers facing everyday work hazards” (CQ Almanac 1969: 569). Representatives of

113

other labor union organizations also appeared before the Senate Committee to support
the Democratic bills. Thomas E. Boyle, president of International Chemical Workers
Union, also urged congressional support of S 2193 and HR 3809.

In 1970 the House and Senate Labor Committees reported the Democratic
versions to their respective floors. On July 9, 1970, despite the strong opposition from
Republican members, the Democratic-dominant House Education and Labor Committee
reported a bill (HR 16785) that was similar to the previous year’s Democratic version.
Twelve Republican members of the Committee signed minority views arguing that the
Committee bill was unacceptable to them since it would create a monopoly function in
the Department of Labor. However, on November 24, 1970, the House floor adopted, on
a 220-173 roll-call vote, a substitute amendment introduced by William A. Steiger (R-
Wisconsin) and Robert L. F. Sikes (D-Florida). The coalition of Republicans and
conservative Southern Democrats successfully replaced key provisions of the Committee
bill with the Steiger-Sikes amendment that established an independent board to
promulgate safety and health standards and an appeals commission to enforce
regulations. While Northern Democrats overwhelmingly opposed the amendment (6-
135), Republicans (154-17) and Southern Democrats (60-21) supported it.

On October 6, 1970, the Senate Labor and Public Welfare Committee also
reported the Democratic bill (S 2193), introduced by Harrison A. Williams Jr. (D-New
Jersey), which gave the Secretary of Labor responsibility for both setting and enforcing
safety standards, while killing the Nixon bill (S 2788). On the Senate floor, Republican
Senators tried to revamp the Committee bill. Peter H. Dominick (R-Colorado)

introduced a substitute bill that allowed an independent board to set safety standards and

114

a special commission to enforce regulations and allowed only the courts to order a plant
with an imminent danger to be shutdown. William B. Saxbe (R-Ohio) submitted an
amendment that deleted the provision allowing the Labor Secretary to close down a plant
with an imminent danger to workers. Liberal Democrats managed to defeat these
Republican amendments. But the Senate approved, on a 43-38 roll-call vote, a
compromise amendment offered by Jacob K. Javits (R-New York) which would create a
three-member review commission. Liberal Democrats opposed the amendment by a
solid vote (0-33) but could not offset the support of Republicans (32-0) and Southern
Democrats (1 1-5).

Since there was a substantial gap between the House and the Senate bills,
bicameral compromise in the conference was critical. Although conferees reached an
agreement by mixing both versions, some key provisions of the Senate bill were
maintained. Most importantly, the Steiger-Spikes amendment that would create an
independent board to administer the program was dropped. Instead, the Secretary of
Labor was granted full responsibility to promulgate safety and health standards.
Conferees also retained other key provisions in S 2193, such as a rulemaking procedure
for setting interim and permanent standards, a three-member Occupational Safety and
Health Appeals Commission appointed by the president with the consent of the Senate,
an additional Assistant Secretary of Labor for Occupational Safety and Health, a
National Institute for Occupational Safety and Health, and pre-enforcement judicial
review of a standard by a local circuit court. Some House-passed provisions were also
retained in the conference report, including the requirement of a court order to end

imminent danger.

115

Although the conference version drew mixed reactions from labor and business
interests, final passage was assured when President Nixon and the Labor Department as
well as some business and labor groups endorsed the bill (Congress and the Nation Vol.
III: 713). President Nixon signed the bill on December 29, 1970, creating the

Occupational Safety and Health Act.

4.4.2. Labor-HEW Appropriations Bills (1972-1980)

Opponents of the newly established regulation began to criticize the agency right after
the creation of OSHA. The focus of the criticism was on the economic burdens and
technological demands that small businesses should bear under the new occupational
safety and health rules. The conservative coalition of Republicans and Southern
Democrats in Congress tried to amend the OSH Act by granting exemptions for those
small firms. However, neither the House Labor Committee nor its counterpart in the
Senate took any action to change the enabling law. Since the House and Senate Labor
Committees exercised their gatekeeping powers by rejecting every attempt to change
provisions of the OSH Act, the hostility of conservative legislators toward OSHA
programs was forced into the annual appropriations bill for the Department of Labor and
the Department of Health, Education and Welfare in the 19705.

On May 29, 1971, several months after the OSH Act became effective almost 250
pages of safety and health standards were published in the Federal Register. The newly-
appointed Assistant Secretary of Labor for Occupational Safety and Health, George
Guenther, announced five target industries and five target health hazards for initial

concentration due to their high accident rates. The target industries were marine cargo

116

 

handling, roofing and sheet metal, meat and meat products, mobile home manufacturing,
and lumber and wood products, and the five target hazards were asbestos, lead, silica,
cotton dust, and carbon monoxide. Most plants in these industries were relatively small
in size and lacked professional safety engineers or industrial hygienists.

Small businesses criticized the OSHA’s administration of regulation on the
ground that it was unfair to establishments lacking safety engineers and experts
specialized in highly technical regulations. In 1972 Congress responded to the growing
concerns with committee hearings by the House Select Small Business Subcommittee on
Environmental Problems, the Senate Labor and Public Welfare Subcommittee on Labor,
and the House Education and Labor Select Subcommittee on Labor. The biggest issue at
these hearings was whether small businesses should be exempted from the OSH Act. In
the Senate Labor and Public Welfare Subcommittee hearing on July 25, Richard B.
Berman, an attorney for the Chamber of Commerce, testified that:

Small employers cannot realistically be expected to be acquainted with existing

and continually changing requirements under the Act....Whatever is gained in

compliance by the Act’s punitive approach is lost many times over by the
growing resentment, lack of respect and feelings of disenfi'anchisement felt by
businessmen today. The Act should guarantee that a businessman will not be

fined after an inspector’s first visits if he did not know he was violating a

standard. (CQ Almanac 1972: 792)

Republican committee members joined the anti-OSHA protest. Senators Carl T.
Curtis (R-Nebraska) and Peter H. Dominick (R-Colorado) submitted a bill (S 3262) to
amend the OSH Act of 1970. The bill would exempt small business employers from
OSHA regulations and delay the effective date of OSHA coverage for small firms

employing between 25 and 100 workers. Senator Dominick criticized the OSH Act of

1970 on the ground that:

117

The administration of the Act, due to the police-oriented approach which
permeates the legislation, has gotten crosswise with Congress’ original purpose to
improve working conditions. . ..Employers are punished for noncompliance with
standards they know little about rather than assisted in complying. (CQ Almanac
1972:793)

Organized labor opposed the Curtis-Dominick bill. In his testimony before the
Senate hearing on July 27, Jacob Clayman from AF L-CIO criticized the proposed
exemption of small businesses on the ground that “the highest concentration of health
hazards, the fewest safeguards and the least awareness on the part of management of the
threat to life” could be found in small businesses (CQ Almanac 1972: 793). Leonard
Woodcock, president of the United Auto Workers (UAW), in the House Education and
Labor Subcommittee on Labor hearing on September 28, said that excluding small
businesses would “doom millions of workers to the continued presence and the insidious
escalation of hazardous materials and hazardous conditions of employment” (CQ
Almanac 1972: 794).

The House and Senate Labor Committees did not take further action regarding
the issues of small business exemption. In order to bypass the Committees’ gatekeeping,
opponents tried to constrain OSHA’s enforcement powers by using legislative riders on
the funds for the agency. The first attempt took place in 1972. On June 7, the House
Appropriations Committee reported out the Labor-HEW appropriations bill (HR 15417)
appropriating $28,239,346,500 for the two departments, including $69,207,000 for
OSHA. Although the portion for OSHA was less than 1 percent of the total Labor-HEW
appropriations package, the serious debates which took place on the House floor were a

harbinger for much more intense battles in later years. Representative David W. Dennis

(R-Indiana) submitted an amendment deleting $20 million for OSHA to “send a

118

message” to the agency that Congress was displeased with its implementation of the Act.
Paul Findley (R-Illinois) offered another amendment to prohibit use of funds in the bill
for inspecting firms employing 25 or fewer workers, and this amendment was adopted.

The Senate Appropriations Committee amended the House-passed Labor-HEW
appropriations bill. It recommended $12,498,000 more than the House appropriations,
including an increase of $10.8 million for OSHA to help the agency hire 400 additional
inspectors, develop safety and health standards, and assist small businesses.
Furthermore, the Senate Committee deleted the House amendment that exempted small
businesses with 25 or fewer workers. The OSHA exemption provision appeared again
on the Senate floor. Senator Curtis, who would later become the champion of the small
business exemption in the Senate, offered an amendment to set the exemption ceiling at
25 persons for OSHA coverage. Liberal Democrats managed to defeat this amendment
on a 41-44 roll-call vote. After this defeat, Curtis submitted another amendment that
lowered the ceiling to 15 persons, which was approved. However, on August 16, 1972,
President Nixon vetoed HR 15417, stating that the appropriations bill was an example of
“reckless federal spending” (CQ Almanac 1972: 873).

After failing to override the veto, the House Appropriations Committee reported
a revised Labor-HEW appropriations bill (HR 16654). On the House floor, the provision
of 15-person ceiling for small business exemption, introduced by CC. Fisher (D-Texas),
was approved on a 191-182 roll-call vote. On the Senate floor, debates on OSHA
provisions turned emotional. Six of seven roll-call votes were on amendments to the
OSHA exemption ceiling. The House-passed ceiling of 15 workers was deleted by an

amendment offered by Clifford P. Case (R-New Jersey) on a 47-33 roll-call vote. Curtis

119

countered with an amendment setting the ceiling at 7, which was defeated by a 38-43
roll-call vote. Curtis again lowered the ceiling to 4, which was defeated by a 39-39 tie
vote. Curtis’s final amendment setting the ceiling at 3 persons was approved by a 50-28
roll-call vote. Eventually, the House-Senate conferees agreed to the Senate ceiling at 3
and both houses approved it. However, on October 27, 1972, President Nixon again
pocket vetoed the second Labor-HEW appropriations bill (HR16654).

In 1976 critics of OSHA in Congress finally succeeded in chipping away at the
agency’s power. On the House floor, nearly 13 hours of debate on the Labor-HEW
appropriations bill (HR 14232) on June 23 and 24 were devoted to OSHA issues. Joe
Skubitz (R-Kansas) submitted an amendment that would exempt farmers employing
fewer than 10 workers from OSHA regulation. He referred to OSHA as a “monster” and
read from OSHA regulations which reminded farmers that “manure could be slippery”
(CQ Almanac 1976: 796). Although Democrats tried to stop the debate on the Skubitz
amendment and lower the exemption ceiling to 5 persons, the House floor rejected the
modified version and adopted the Skubitz amendment by 273-124. Paul Findley (R-
Illinois) rode on the anti-OSHA tide and offered an amendment to exempt small business
employing 10 or fewer workers from OSHA’s first citations. The House again adopted
Findley’s amendment by 231-161.

On the Senate floor the situation was more complicated. John A. Durkin (D-New
Hampshire) sought to sofien the House-passed provision language for the blanket
exemption for farms but mistakenly offered an amendment that would broaden the
exemption. His proposal would exempt all firms of any size from OSHA’s first instance

fines for non-serious violations while deleting the House provisions that would exempt

120

 

small farms from any OSHA regulation and small businesses from first-time citations.
Conservative Republicans and Southern Democrats wanted to save the blanket
exemption provision. Strom Thurmond (R-South Carolina) argued that “OSHA has
apparently decided that farmers are too dumb to understand the basics of farm safety”
(CQ Almanac 1976: 800). James Abourezk (D-South Dakota) offered a compromise
amendment to the Durkin proposal to exempt from all OSHA regulations farmers
employing an average of 5 or fewer workers a day in addition to the exemptions on first-
instance citations. But James B. Allen (D-Alabama), a conservative Southern Democrat,
wanted to further restrict OSHA enforcement and offered amendments that would
prohibit OSHA from issuing first-time citations to small businesses even for serious
violations. Liberal Senators opposed the Allen proposals and successfully killed them.
Fearing Allen’s parliamentary mastery and his promised filibuster, Senate Majority Whip
Robert C. Byrd (D-West Virginia) stepped in and filed a petition to invoke cloture. In
the end, Allen agreed to a modified version of Durkin amendment.

The conference report on HR 14232 included the House provision that would
exempt farms with 10 or fewer employers from all OSHA regulation and the Senate
provision that prohibited OSHA from issuing first-time citations to employers unless
more than 10 serious violations were found in the first inspection. Although President
Ford vetoed the $56.6 billion appropriation bill, including the OSHA-related provisions,
Congress overrode Ford’s veto.

From 1978 to 1980 conservative members of Congress tried to broaden the
eligibility for exemption from OSHA regulation based not only the size but also the past

safety records of establishments. In 1978 the House Small Business Committee tried to

121

 

amend the Small Business Administration (SBA) Act. An OSHA issue emerged during
the Senate floor consideration on the bill. Senator Dewey F. Bartlett (R-Oklahoma)
offered an amendment that would exempt small businesses with 10 or fewer employees
from OSHA regulations. On August 2 the exemption was approved by voice vote. In
the House-Senate conference, the SBA bill included the exemption from OSHA citations
on firms with 10 or fewer employees unless 10 or more violations were found in the first
inspection. And also the amendment would exempt small businesses with 10 or fewer
employees from OSHA requirements for record-keeping and report. On October 11,
1978, Congress cleared the bill. However, President Carter pocket vetoed the bill on the
ground that the bill included authorizations in excess of his requested budget.
Accordingly the OSHA exemption provisions also died.

The House Appropriations Committee had already attached provisions that
limited OSHA regulation to the 1978 Labor-HEW appropriations bill (HR 12929). The
House floor approved three committee recommendations related to OSHA. The Labor-
HEW appropriations bill prohibited OSHA from issuing fines for first-time non-serious
violations unless the establishment was cited for more than 10 violations in first
inspection. It also exempted farms with 10 or fewer workers from OSHA regulations.
OSHA was also prohibited from restricting work in an area because of potential dangers
posed by nearby recreational or hunting activities. On the Senate floor, Senator Bartlett
resubmitted his SBA proposal as an amendment to HR 12929. This time the Bartlett
amendment prohibited safety inspections of firms with 10 or fewer employees in
industries with occupational injury rates of seven or less per one hundred workers.

Under the amendment, most of the 10.5 million workers employed by the 3.9 million

122

 

small businesses would no longer be covered by OSHA regulations. In order to avoid
the possibly substantial damage to OSHA programs, Majority Leader Robert C. Byrd (D-
West Virginia) moved to table the Bartlett amendment and the Senate agreed to the

motion by 47-46.

4.4.3. The Risk Notification Bill (1986-1988)
Since 1968, more than 19 million workers were estimated by the Center for Disease
Control to have been exposed to toxic substances on the job, increasing their risk of
contracting cancer, respiratory disease, or other serious health problems (CQ Almanac
1988: 264). Under the OSH Act of 1970, the National Institute on Occupational Safety
and Health (N IOSH) was authorized to study occupational diseases. However, the OSH
Act did not establish any mechanism for notifying workers about the potential risk of the
hazardous substances. NIOSH had proposed a pilot program designed to notify workers
of potential job-related hazards in 1981. Although NIOSH had requested funding for
notifying workers of their potential risks on the job, the Reagan administration rejected
those requests in fiscal years 1983-1987. OSHA also was aware of the need for risk-
notification and began enforcing the Hazard Communication Standards in 1983 (29 CFR
1910.1200). The rule required that warning labels be put on chemical containers, work
sheets on hazardous chemicals be made available to all employees who work with those
chemicals, and workers be trained to understand and handle hazardous materials.

In 1986 Representative Joseph M. Gaydos (D-Pennsylvania) introduced a bill,
HR 1309, that would establish a national system for identifying and notifying employees

who had been exposed to dangerous substances in the workplace. The bill would

123

 

establish a Risk Assessment Board to review medical research and identify group of
workers at risk of occupational diseases. Once the board identified the group, the
Secretary of Health and Human Services was required to notify individual workers and
former workers. The notification would include the identification of the hazardous
substance and potential diseases, latency periods, and necessary medical testing and
monitoring. Employers were required to pay for the medical diagnoses. The bill set
procedures for transfer of workers to other jobs without loss of earnings or benefits. On
June 25 the House Education and Labor Committee approved the bill on a party-line vote
of 20-8, but the bill could not reach the floor that year.

The House Education and Labor Committee introduced a very similar version of
the Risk Notification Bill (HR 162) again in 1987. Business groups and the Reagan
administration were against the bill on the grounds that an OSHA rule (Hazard
Communication Standard) for warning workers about toxic hazards already existed and
that the proposed regulation would impose a heavy burden of legal fees on companies to
deal with legal challenges under the law. The National Federation of Independent
Business (N FIB) predicted that the bill would drive many small businesses into
bankruptcy and would duplicate the work of existing federal agencies (Congress and the
Nation, Vol. VII: 708). More than 250 trade associations and businesses formed the
“Coalition on Occupational Disease Notification” and joined the lobbying effort against
the measure (CQ Almanac 1988: 265). However, Democrats in the House Education and
Labor Subcommittee on Health and Safety used their large majority to defeat Republican
opposition; they approved the measure without change on April 23, 1987. The full

House Committee also approved HR 162 on May 19. On the House floor Republican

124

 

legislators, including the Republican ranking committee member James M. Jeffords (R-
Vermont) and Paul B. Henry (R-Michigan), offered several amendments that would
restrict coverage of the proposed law on small businesses, give the primary responsibility
to OSHA, and require a two-year study of the need for such risk-notification program.
Democrats defeated the J effords-Henry amendment by 191-234. However, the House
approved an amendment that would exempt small businesses with fewer than 50
employees and farms with fewer than 15 workers from the job-transfer requirements.
On October 15, 1987 the House passed HR 162 by 225-186.

In the Subcommittee on Labor of the Senate Labor and Human Resources
Committee, subcommittee chairman Howard M. Metzenbaum (D-Ohio) introduced a
companion bill (S 79). Metzenbaum tried to reach a compromise on the bill with his
chief Republican opponent, Dan Quayle (R-Indiana). This proposed compromise would
exempt companies with fewer than 10 employees from the requirements for the bill.
However, the changes were not enough to satisfy Quayle, and he offered several other
amendments, including one that would give the Secretary of Health and Human Service
more authority over the Risk-Assessment Board’s decisions. All these proposals were
defeated, and on September 23, 1987, the Senate Committee filed its report on this
measure, which was backed by the Democratic majority.

In March, 1988, S 79 was brought to the Senate floor. The first move on the bill
came when the Senate Majority Leader Robert C. Byrd (D-West Virginia) filed a cloture
motion limiting debates on the measure even before any discussion whatsoever began.
This motion was defeated by 33-59, far short of the 60 votes needed to invoke cloture.

Vexed by the Byrd motion and aided by strong lobbying efforts of business groups,

125

 

Republican Senators were able to wage a filibuster. In the weeks prior to the Senate
consideration, nearly 10,000 members of the National Federation of Independent
Business (N FIB) had contacted Senators to urge them to participate in a filibuster (CQ
Almanac 1988: 265). They argued that the bill could cost employers $6 billion a year in
lawyer’s fees, medical testing and monitoring, and other employee benefits. But chief
sponsor Metzenbaum countered that occupational diseases cost nearly $10 billion in
medical expenses and government expenses without the proposed measure.
Furthermore, he defended S 79 on the ground that the bill would complement current
efforts to control occupational diseases, and he noted that the bill included provisions
that prohibited employees from using the risk notification provision for lawsuits.

Two Republican Senators, Quayle and Orrin G. Hatch (R-Utah), led the
legislative opposition and criticized various provisions of the bill. They pointed out the
insufficient scientific data to warrant a conclusion about health risks, the existence of
more than a dozen other federal agencies to deal with worker health and safety, and the
cost of businesses’ increased liability. In order to attract broader support, Republicans
refused to vote on the modified versions offered by Democrats, such as one that would
exempt farms and small businesses from the bill. Quayle threatened that “We have at
least 40 votes we can hold” and admitted that the Republican strategy was “dragging feet
and killing time” (CQ Almanac 1988: 265). On March 24, one day after the Bird motion,
another motion to invoke cloture failed by 2-93 due to a bipartisan agreement. Two
more motions to invoke cloture on March 28 and 29 were also defeated by 41-44 and 42-

52. Eventually, floor managers pulled the bill and it died upon adjournment.

126

4.4.4. Reform, Overhaul, and OSHA (1992-1996)

In the 19905 there were several legislative attempts to reform occupational safety and

health regulation. The first wave of reform efforts began with the initiatives taken by the

House Education and Labor Committee and liberal Democrats who wanted to reinforce

OSHA regulation. In 1992 the House Education and Labor Committee approved a bill

(HR 3160) that would have made the first big change to the Occupational Safety and

Health Act in more than 20 years. HR 3160, drafted by committee chairman William D.

Ford (D-Michigan), would change the OSH Act as follows:

Increase criminal penalties from the six-month maximum to a 10-year
maximum for workplace safety violations that resulted in a worker’s
death

Apply the penalties to supervisors, while under existing law, only
employers could have been held liable.

0 Require businesses to establish workplace safety and health programs.
0 Require OSHA to issue a standard to control ergonomic and

environmental hazards such as repetitive strain injuries that could come
from using a computer or video display terminal.

Require OSHA to write regulations outlining how businesses should
respond to the problem.

Require businesses to establish joint committees for employers and
employees to work together to improve job-site conditions in companies
with 11 or more full-time workers.

Paul B. Henry (R-Michigan) and ranking Republican member Bill Goodling (R-

Pennsylvania) offered a less far-reaching substitute that would require the following:

0 Government should consider a rule’s cost-effectiveness and its possible

effect on employment when setting health and safety standards.

0 Employers can consult an employee participation committee on health

and safety issues for businesses with 50 or more workers.

After rejecting the Henry-Goodling amendment by a 14-24 party-line vote, the

committee agreed by voice vote to two amendments: one would allow OSHA citations

issued to employers to be dropped if the violations were caused by employees who

127

 

violated company work rules, and the other would require OSHA to target inspections of
work sites to those with a high potential for death, serious injury or exposure to toxic
materials. On May 28 the committee gave a voice-vote approval to HR 3160 and
reported it on July 9. The Bush administration opposed it. Labor Secretary Lynn Martin
threatened that she would recommend President Bush to veto the legislation. Eventually,
the bill died without further action.

On July 9, 1992 the House Education and Labor Committee reported HR 1063 to
protect construction workers by setting stricter safety guidelines for construction
companies. The bill, crafted by the House Education and Labor Subcommittee on Health
and Safety chairman Joseph M. Gaydos (D-Pennsylvania), required construction firms to
develop a written safety and health plan for individual projects. Companies had to hire
or designate a project constructor who would be responsible for the site, oversee the
plan, and conduct regular inspections. Builders were required to hire a safety
coordinator to implement the plan and investigate any serious injuries or deaths at the
site. The bill would establish an Office of Construction Safety, Health, and Education
within OSHA and the agency would be headed by a Deputy Assistant Secretary for
Construction. The office would investigate deaths and injuries and help develop
construction rules. Gaydos justified his proposal on the ground that about 2,500
construction workers were killed and more than 200,000 seriously injured in construction
accidents each year. The subcommittee approved HR 1063 on July 23, 1991 by a 5-3
party-line vote and the full committee approved the measure by 24-14 on September 24.

However, this bill also died without further action.

128

On October 3, 1994 legislation aimed at substantially overhauling the law
governing workplace safety was reported by the House Education and Labor Committee
GIR 1280). HR 1280 required that employers with 11 or more workers use
recommendations from OSHA to establish a written health and safety plan to reduce
hazardous working conditions. The bill also required OSHA to investigate potential
safety violations within 24 hours of an accident in which unsafe conditions might have
been a cause. In addition, the measure would extend the safety standards under existing
laws to the employees of state and local governments. Democratic supporters of the
measure maintained that an overhaul of the OSH Act was long overdue. But
Republicans countered that the bill would lead to excessive regulation, hampering
businesses, and slowing economic growth. On March 10 the House Education and Labor
Committee approved HR 1280 by a party-line vote of 26-1 7, but no further action was
taken.

The sweeping GOP victory in 1994 suddenly changed the political landscape.
The Democrats were now in the minority on Capital Hill. After the 1994 congressional
elections, a reform drive in the opposite direction was launched by the Republican Party.
The Republican reform efforts were aimed at diluting OSHA regulations. In general, the
new Republican leadership was committed to rolling back regulations it considered
overly burdensome to business.

President Clinton took a preemptive action before the beginning of the GOP
attack on OSHA. On February 21, 1995 President Clinton announced his
administration’s plan to reform OSHA. Clinton’s three regulatory reform initiatives on

OSHA included (Clinton 1995: S):

129

o The New OSHA—Partnership or Strong, Traditional Enforcement:
OSHA will change its fundamental operating paradigm from one of
command and control to one that provides employers a real choice
between a partnership and a traditional enforcement relationship.

0 Common Sense Regulation: OSHA will change its approach to
regulations by identifying clear and sensible priorities, focusing on key
building block rules, eliminating or fixing out of date and confusing
standards, and emphasizing interaction with business and labor in the
development of rules.

0 Results, Not Red Tape: OSHA will change the way it works on a day-to-
day basis by focusing on the most serious hazards and the most
dangerous workplaces and by minimizing red tape.

The “New OSHA” initiatives included various measures to encourage the
development of voluntary worksite safety and health programs. Firms with safety and
health programs would be given the lowest priority for enforcement inspections, the
highest priority for government assistance, and penalty reductions of up to 100 percent.
In contrast, for firms that would not implement such effective programs, OSHA would
focus its inspections on those firms at the highest level of priority and rigor. The
“Common Sense Regulation” initiatives were aimed at streamlining OSHA regulations
and standards. OSHA would identify new priorities to set standards for un-regulated
hazards, consolidate scattered and duplicative elements of current standards, and review,
rewrite, and revoke confusing and out-of-date standards. The “Results, Not Red Tape”
initiatives were designed to improve OSHA performance. The initiatives included
OSHA field office redesign to install technologies to identify the leading causes of death,
injury, and illness, a new partnership with state programs to encourage innovative ways
to prevent injuries and illnesses, new incentives such as penalty reductions for expedited

abatement of hazards, a new inspection targeting system to identify a target on the basis

of individual employers’ safety and health records rather than the industry-based records,

130

and the development of a comprehensive performance measurement system that would
shift the focus from tracking activities to monitoring achieved results.

Despite such a comprehensive plan promised by President Clinton to improve the
ways of managing OSHA regulations, the GOP began its legislative reform drive on
OSHA in 1996. On June 28 the Senate Labor and Human Resources Committee
reported a bill (S 1423) that would revamp the Occupational Safety and Health Act of

1970. S 1423 included the following changes:

- Authorizing employers to create their own safety plans and hire outside
inspectors to approve them.

0 Exempting companies that opted for this approach from regular OSHA
inspections and subjecting them to reduced penalties if a violation
occurred.

0 Repealing the requirement that OSHA make inspections whenever an
employee filed a complaint.

0 Eliminating penalties for minor violations, such as incorrect paperwork, in
many cases.

Senator Judd Gregg (R-New Hampshire), the chief sponsor of the bill, justified
the measure on the ground that it would free OSHA from making unnecessary
inspections and enforcing unimportant regulations and would allow the agency to focus
on companies that were exposing their workers to physical hazards. Committee
Democrats were united against the measure. They argued that the bill would deprive
OSHA of needed authority. Paul Simon (D-Illinois) offered an amendment that would
apply national safety and health laws to federal, state, and local governments. The
Congressional Workplace Compliance Act of 1995 (PL 104-l) had applied labor laws
including the OSH Act to Congress but not to the rest of the federal government or its

state counterparts. Even though Gregg opposed Simon’s amendment, the amendment

was approved by 9-7 afier two liberal Republicans, James M. Jeffords (R-Verrnont) and

131

Spencer Abraham (R-Michigan), broke the party-line and joined the committee’s seven
Democrats. However, Democratic Senators on the floor had already threatened a
filibuster on S 1423. President Clinton also had promised a veto of it (Congress and the
Nation, Vol. IX: 671). Eventually, the bill died without further action as the Senate
supporters were unable to obtain time for S 1423 on the packed floor calendar.

In the House the Economic and Educational Opportunities Subcommittee on
Workforce Protections approved a different set of changes (HR 3234) on April 17, 1996.
The House bill included the following requirements:

0 Applying cost-benefit analysis to new regulations issued by OSHA.

o Granting the Secretary of Labor the right to waive any fines against
businesses of 250 or fewer employees if a minor safety violation was
corrected quickly, or if the money for the fine was used to fix the
problem.

0 Prohibiting OSHA inspectors from issuing citations for paperwork
violations unless an employer had “willfully or repeatedly violated” a
regulation or if the violation had exposed a worker to a safety hazard.

During the markup, opponents of the bill argued that easing enforcement burdens
on businesses with 250 workers or fewer would exempt 99 percent of all workplaces
from mandatory safety and health inspections and greatly endanger workers. Such fierce

Democratic opposition and President Clinton’s veto threat hampered the measure in the

legislative track and the bill never even reached to the full committee.

4.5. Institutional Environments of OSHA in Perspective

Occupational safety and health regulation has been on the national-level policy agenda
since its inception. During this period of thirty years, the institutional actors in the
national policymaking arena never stopped discussing OSHA-related issues, and the

president, the House and Senate floor, and the House and Senate Committees closely

132

interacted with each other. Even though OSHA officials were given broad responsibility
to set and enforce standards for workplace safety and health, their choices could have
been affected by the contents of legislative discourse. Those institutional actors sent
policy messages to OSHA throughout the legislative process. Despite the fact that the
OSH Act remained intact and so did OSHA’s legal authority, the dynamic change in
policy message generated by the elected policymakers may have transmitted strong and
clear signals of their policy positions.

In this section I discuss how to interpret these complicated interactions among
political institutions from the three perspectives on pivotal policymakers mentioned in
chapter 3: the majoritarian, the distributive politics, and the party government
perspectives. When viewed from one of these theoretical lenses, we can focus on
particular modes of interaction among the institutions, which will shape the institutional
environments for OSHA inspectors. The main purpose here is to stress how different
aspects of the process can be highlighted from different perspectives. The full
assessment regarding which perspective best fits the reality of the politics of workplace

safety regulation will be conducted in chapters 5 and 6.

4.5.1. The Majoritarian Perspective

According to the majoritarian perspective, a major policy change can take place when
there is an agreement among three institutional actors: the president, the House, and the
Senate. Each of these three institutions is assumed to be governed by the majority rule.
The Presidency can be thought to be a single-member institution whose policy position

reflects the president’s preference. Legislative decisions made by the House and the

133

 

Senate also can be considered as a reflection of the policy preferences of the median
members. Since the preference distribution in the congressional committees is assumed
to mirror the overall preference distribution of the rest of the chamber, the role of
congressional committees is considered to be absorbed by that of the median members
on the floors. In support of this majoritarian view, we observe the following patterns in
the narratives presented earlier in this chapter.

First, the history of the policy debates highlights the importance of inter-
institutional agreement on the level of OSHA regulation. Presidents and the House and
Senate were all active participants in the policymaking process. Between 1968 and 1970,
the debates on the initial OSH Act focused on specific provisions of two competing
versions. The Nixon bill had proposed an independent board to assume rulemaking
authority while the Democratic bill proposed the creation of an executive agency in the
Department of Labor. Most of the legislative process was devoted to resolving the
differences between the Democratic-controlled Congress and the Republican president
(and its contingents in Congress). In the process, inter-chamber disagreements also
emerged. While the Senate wanted to maintain key provisions of the Democratic bill,
the House adopted a substantive amendment that made its final outcome closer to the
president’s version. Without the compromise made by the House-Senate conference, the
OSH Act of 1970 could not have been enacted.

Legislative actions on the small-business exemption riders to the Labor-HEW
Appropriations bills also point to the importance of inter-chamber agreement despite the
fact that both houses had been under the Democratic control. During the 1970s and the

19803, the House and Senate had set the exemption ceiling at different levels. For

134

instance, in 1972 the House-passed exemption ceiling for small businesses was 25 while
the Senate-passed ceiling was 15 workers. In 1974, while the House approved an
exemption ceiling of 25 workers, the Senate did not adopt any exemption provisions.
Due to the different decisions on the range of exemption, the House-Senate conference
had to find reasonable point that would be acceptable to both houses. The case of the
Risk Notification Bill in 1986-88 also clearly demonstrates the importance of inter-
charnber agreement. While the House took a smooth legislative course and passed the
measure (HR 162) in 1987, the Senate, after spending 2 years to get the companion bill
(S 79) on the floor, ended up failing to pass it in 1988. Due to the absence of policy
agreement between chambers, one of the major legislative changes to the OSH Act failed.
One possible criticism of this application of the majoritarian perspective is
whether the institutional decisions were governed by simple majority rule or by super-
majority rule (Krehbiel 1998; Brady and Volden 1998). In the entire history of the 82
recorded legislative votes on OSHA-related bills and provisions between 1968 and 2001,
the number of cloture on Senate filibuster attempted was only 4, of which 3 were on the
Risk Notification Bill in 1988. The Senate debates on OSHA issues were often heated
and occasionally lasted more than one day. However, there were few references to the
use of the 3/5 cloture-rule, and most of debates were settled under the simple majority
rule. There were no presidential vetoes on OSHA-related provisions during the period
under consideration. But two presidents, Carter and Clinton, had promised to veto
legislative proposals that would change the OSH Act in the very early stage of
congressional committee considerations and those measures never reached the floor in

either house. Excluding votes on the passage of the House-Senate conference report and

135

 

procedural motions, most of the 52 legislative decisions were made in the House and the
Senate with a simple-majority margin that was less than 2/3 in the House or 3/5 in the

Senate.

4.5.2. The Distributive Politics Perspective

The distributive politics perspective emphasizes the important role of the congressional
committees. Congressional committees are considered as having outlying policy
preferences that are noticeably distinguishable from the rest of the chamber.
Congressional committees thus may try to use their monopoly power over policy issues
in their jurisdiction to obtain legislative outcomes that can serve their own interests. In
the area of occupational safety and health, we can find the following as to the crucial role
played by congressional committees.

First, the House and Senate Labor Committees did clearly exert gatekeeping
power against every legislative attempt to change the Occupational Safety Health Act for
along period of time. During the 19703, the House and Senate Labor Committees
effectively blocked numerous legislative attempts to dilute provisions of the OSH Act of
1970. Even though the Labor Committees responded to criticism on the Act concerning
its economic burdens placed on businesses by holding hearings, none of legislative
proposals to change the Act was sent to the floor. Even during the period of the
“Contract with America” in the mid-19903, the House Labor Committee successfully
protected the OSH Act from the aggressive GOP reform drive to weaken occupational
safety and health regulation. Due to the committee’s effective gatekeeping, every single

provision of the Act remained intact despite the GOP’s anti-regulation campaign. The

136

 

GOP’s failure to amend the OSH Act was attributable at least in part to the fact that the
Labor Committee members consisted of relatively liberal members in both parties so that
the GOP majority in the committee could not overcome the pro-labor voices.

Second, the Labor Committees played a crucial role in setting the legislative
agenda. Especially in the course of the OSH Act enactment between 1968 and 1970, the
House and Senate Committees effectively changed the major points of legislative
discourse by providing their own bills for floor debates in the critical moment. When the
new Republican President, Richard Nixon, took over the White House and offered a
legislative proposal that was considered antithetical to provisions proposed by the
Johnson administration, the House and Senate Labor Committees reversed the course by
replacing the Nixon bill with their own version in 1969. The House and Senate
Committees revived key provisions of the Johnson bill such as the creation of executive
agency in the Department of Labor and the concentration of power to set and enforce
regulatory rules in the hand of the Labor Secretary. The Committee bill provided the
main framework for the Democratic bills, which would become, with some
modifications, the Occupational Safety and Health Act in the following year.

Third, note that the role of the House and Senate Appropriations Committees has
been overshadowed by legislative actions on the floors. Although the Appropriations
Committees decide the initial proposal for annual funding for OSHA, most of the
important discussions about the funding levels and about how to use the funds have
taken place on the House and the Senate floors. In particular, the Appropriations
Committees did not attach legislative provisions that would restrict OSHA regulation,

even though the issue of small business exemption was frequently raised by legislators

137

 

 

on the floor. This can be attributable to the fact that the House and Senate Labor
Committee maintained their firm position not to amend the OSH Act. Opponents of
OSHA regulation simply tried to use legislative riders on the annual Appropriations bill
to constrain the agency’s regulatory powers. As can be expected, the key participants of
the floor debates on OSHA-related provisions in the Labor-HEW were members of the
House and Senate Labor Committees. Therefore, the House and Senate Labor
Committees rather than the House and Senate Appropriations Committees were the most

dominant player from the distributive politics perspective.

4.5.3. The Party Government Perspective

According to the party government perspective, legislative policy outcomes are
determined in ways that can serve the legislative parties’ interests. The majority party
and its leadership tend to capitalize on their advantages in using various parliamentary
powers to control the legislative agenda and to manipulate the legislative procedures. As
a result, decisions to be made in committees or on the floor are orchestrated by the
majority party leadership so as to achieve partisan policy goals.

In the history of occupational safety and health legislations, legislative actions in
the congressional committee and the floor were closely connected by party leaders. The
majority party in the House and Senate Labor Committees ofien dominated committee
discussions and determined the major provisions of the bill to be sent to the floor. Thus,
. the crucial role of the congressional committees such as agenda setting and gatekeeping
can actually be attributed in many aspects to partisan politics. In the case of the

enactment of the Occupational Safety and Health Act in 1970, the Labor Committees

138

were clearly divided on party-lines in both chambers. Democratic members of the
committees tried to retain the key provisions of the Democratic versions, while all
Republican members strongly opposed the bill. The Democratic majorities in the
committees refused any major change and sent the companion bills (S 2193 and HR
16785) to the floor. All 12 Republican members of the House Labor Committee signed
the minority view that the Committee bill was unacceptable to them. In 1972, when
Republican committee members of the Senate Labor Committee submitted a bill (S
3262) to amend the OSH Act of 1970, the committee did not take any action beyond its
congressional hearing due to the opposition of its Democratic majority.

On the floor, consideration of OSHA-related bills often led to partisan
confrontations. In many floor debates on OSHA, the majority party tried to maintain the
majority view incorporated in the committee bill, while the minority party tried to amend
it. This competition between partisan bills and amendments for enactment has taken
place from the very beginning. In 1970, Republican legislators on the House and Senate
floors offered numerous amendments that would dilute the strong provisions of the
committee bills, and liberal Democrats tried to kill those amendments. Especially on the
Senate floor, the battle between the major parties over the key provisions of the OSH Act
bill was so fierce that it took 4 floor votes for the Senate to agree to a modified
Democratic version. This partisan strife continued in the following years when the
House and Senate floors considered the range of exemptions for small businesses and
farms as legislative riders to the Labor-HEW annual appropriations. Every year,
Republican legislators offered riders that would raise the exemption ceiling, and liberal

Democrats tried to defeat them so as to protect the unrestricted coverage of the OSHA

139

 

regulation. For instance, in 1972 it took 9 floor votes in the Senate and 4 floor votes in
the House for the floors to reach final decisions to set the exemption level at certain
levels.

Unlike the Labor Committees in which members’ policy positions were divided
along the clear party lines, an ideological cleavage was often superimposed on the
partisan cleavage on the House and Senate floors. Especially during the period of the
19703, the conservative coalition of Republicans and the Southern Democrats behaved
noticeably differently from the liberal coalition of the Northern Democrats and the
Northeastern Republicans. For instance, in 1970 the Steiger-Sikes amendment that
would substantially dilute the House Labor Committee’s stronger version of HR16785
passed the House by 220-173. However, not all those yeas came from the Republican
Party; and. not all those nays came from the Democratic Party. Although Republicans
were quite united (154-17) on the measure, Democrats were divided. While the Northern
Democrats overwhelmingly opposed the amendment by 6-135, the majority of the
Southern Democrats joined the Republican Party and supported it by 60-21. During the
period, the majority of the Southern Democrats seldom got along with the Democratic
Party majority.

These patterns of legislative behavior began to change in the 19803. For instance,
in 1987 the Democratic majorities in both the House Labor Committee and the House
floor consistently supported the Risk Notification Bill (HR 162). Despite the strong
campaign by more than 250 trade associations and their Republican legislators against
the measure, the Democratic majorities in the subcommittee and the full Labor

Committee approved the bill. On the House floor, the Democratic majority defeated the

140

 

J effords-Henry amendments that would restrict coverage of the proposed law on small
businesses. For this amendment, Republicans overwhelmingly cast yeas by 158-1 7 and
Democrats cast nays by 33-217. Both a majority of the Northern Democrats (4-166) and
a majority of the Southern Democrats (29-51) opposed the restriction to the bill. By the
mid-19903, the partisan cleavage over the OSHA issue became even clearer. Between
1992 and 2001, of a total of 20 recorded legislative votes on OSHA-related matters in the
House and Senate, all but 2 Senate votes were decided along clear-cut partisan lines. For
example, in 1997 an amendment to the Labor-HHS Appropriations (HR 2264) which
would reduce funding levels for OSHA by $11.25 million was defeated by 160-237. A
majority of Republicans supported it by 155-56, and all but 5 Southern Democrats
opposed it by 5-180. In 2001, the House passed a Senate resolution (S J Res 6) that
would revoke OSHA’s controversial Ergonomics standard, by 222-198. Not a single

Republican cast a nay vote (216-0) and only 5 Democrats cast a yea vote (5-198).

4.6. Discussion

OSHA was granted full responsibility to set and enforce occupational safety and health
standards. The agency had formidable powers to force employers to comply with those
rules. In addition to the formal authorities, the agency officials were given tasks that
would require considerable professional training and technological knowledge. The
nature of these tasks helps the agency officials to maintain a degree of discretion. The
agency operates the occupational safety and health programs in a highly decentralized
structure. More than 20 states have developed and operated their own programs, and the

actual enforcement of OSHA rules is administered by regional administrators.

141

Although these statutory structures imposed on OSHA may constitute elements
of institutional environments as to the legal authorities given to the agency, they are not
complete components of institutional forces surrounding the agency. Institutional actors
continuously tried to influence OSHA and its regulatory programs. Three alterative sets
of veto players played important roles in different ways. From the majoritarian
perspective, the policy agenda, the legislative outcomes, and the “message” can be
considered to be determined by interactions among three institutions: the president, the
House, and the Senate. Especially throughout the 19703 and the mid-19803, these three
institutions defined the scope and the level of possible regulatory enforcement activities.
Presidents were active agenda-setters, and the House and Senate floors kept OSHA
issues alive in the legislative arena. When viewed from the distributive politics
perspective, the most important role played by the House and Senate Labor Committees
was their effective gatekeeping. Due to the committees’ blocking of any attempt to
amend the OSH Act, it turned out to be almost impossible to permanently dilute key
provisions of the regulatory structure. Repeated occurrence of intensified discussions on
the OSHA-related riders to the Labor-HEW Appropriations bills on the House and
Senate floors ironically illustrates how strong the Labor Committees’ position was for
such a long period of time. The party government set of veto players also played a
prominent role since the 19803. The conservative coalition of Republicans and Southern
Democrats, which was often an obstacle to the Democratic Party’s legislative successes,
was severely weakened by the mid-19803. Heightened partisan unity might provide a
greater chance for the majority party to send a stream of consistent messages through

legislative actions to agency officials.

142

 

In sum, I draw the following conclusions. First, description of the interactions
among institutional actors shows us that the range of feasible administrative decisions
has been continuously redefined since OSHA was created. Institutional actors sent quite
clear messages to the agency with regards to the level of regulatory stringency, but the
content of messages have dynamically changed. Second, institutional actors were all
intertwined by the macro-rules such as the president’s legislative powers, parliamentary
rules in Congress, and political parties. It was apparent that legislative outcomes
reflected multi-institutional preferences and agreement. There was no single institution
that permanently dominated the legislative discourse to achieve its own policy goals.
Instead, all institutional actors must concede to change a current way of enforcing OSHA
rules. Third, substantively, these narratives reveal that the most important dimension
throughout the entire period from 1969 and 2000 was how to balance business and labor
interests in enforcing occupational safety and health rules. Although numerous interest
groups actively participated in the policymaking process and the workplace rules were
established to regulate such diverse industries, the major point of legislative debates was
to what extent workers should be protected and to what extent business interests should
be considered. It was quite clear that the relative position of those institutional actors on
that particular ideological dimension was the most important determinant of legislative

discourse and outcomes.

143

 

——————--
-:‘ a-___- m”--

CHAPTER 5

FEDERAL OCCUPATIONAL SAFETY AND HEALTH REGULATION
ENFORCEMENT

Political institutions interact with each other to pursue their policy goals through a
regulatory agency’s policy implementation. As the narratives in the previous chapter
reveal, important policy decisions were made as a result of the interactions among the
president, the House and Senate floor majorities, the House and Senate Labor
Committees, and the political parties. These pivotal policymakers took clear positions
on OSHA issues. Politicians exchanged their views on various matters such as how
strong the regulatory agency’s authority should be (1968-1970), to what extent
exemptions from OSHA regulations should be adopted (1972-1980), to what extent
workers’ rights to know should be protected (1986-1988), and how to renew
occupational safety and health programs (1992-1996). Despite the broad range of issues,
legislative debates on these matters were centered on a clear and simple question: to what
extent should businesses’ economic interests be sacrificed for the protection of workers’
safety and health in the workplace?

In this chapter I examine the hypothesized causal mechanisms involving
bureaucratic responsiveness and bureaucratic autonomy. OSHA officials were
presumably attentive to those policy debates. Agency officials could have developed a
“barometer” of politically acceptable levels of enforcement and may have adjusted the
stringency of regulatory activities according to the signals transmitted from these
national policymakers. Several bodies of past research have already found that policy

signals generated by policymakers do indeed affect OSHA officials’ behavior (Scholz

144

and Wei 1986; Scholz, Twombly and Headrick 1991; Headrick et al. 2002). However,
the important question to be answered here is not whether OSHA officials were affected
by political signals but how agency behavior was affected by the inter-institutional
dynamics. Thus, the focus of this empirical test is on (a) whether the directional change
of institutional preferences (i.e., from pro-business to pro-labor), as measured by the
location of the policy equilibrium interval, affects the level of OSHA’s regulatory
activities, and on (b) whether the degree of policy disagreement among institutional
actors, as measured by the size of the policy equilibrium interval, affects the variability
of OSHA’s regulatory actions. Other points of emphasis are (c) which of the three types
of regulatory politics-«the majoritarian, the distributive politics, and the party
govemment---has the most significant and consistent effects on the OSHA officials’
behavior, and (d) how the distribution of preferences of the agency’s officials mediates
the effects of the policy equilibrium interval.

In sections that follow, I discuss the longitudinal and cross-sectional patterns of
federal OSHA enforcement activities, model specification of federal OSHA enforcement

and measurement of variables, and empirical results.

5.1. Federal OSHA Enforcement Activities

Federal occupational safety and health regulation is enforced by the Occupational Safety
and Health Administration (OSHA). OSHA’s headquarters consists of eight functional
units directed by Directorates. The functional areas are Administrative Programs;

Construction; Cooperative and State Programs; Enforcement Programs; Evaluation and

145

 

Analysis; Information Technology; Science, Technology and Medicine; and Standards
and Guidance. Each functional unit is divided into smaller functional offices.

OSHA is also organized on a regional basis. OSHA’s jurisdiction is divided into
10 regions, each of which covers four or more states and territories. The Regional
Administrator is given full responsibility to manage, execute, and evaluate all programs
of OSHA in the region and reports to the Assistant Secretary through the Deputy
Assistant Secretary (FIRM Ch. I. A(1)). Each regional office is firrther divided into area
offices in each state. The Area Director is delegated authority to supervise the
Compliance Safety and Health Officers (CSHOs) in the area and to direct the issuing of
citations. Although the daily operations of OSHA enforcement programs in the field are
mostly supervised by the Area Director or the Assistant Area Director, the Area Director
consults with the Regional Administrator on judgments on high-profile cases which
involve imminent danger or willful violations that may require regional-level
coordination. The Regional Administrator is also responsible for managing personnel
and financial resources to ensure effective and efficient operation of OSHA programs in
the region.

The actual operation of OSHA inspection program rests heavily on the
professional judgments of the OSHA compliance officers. The main responsibility of
these compliance officers is to determine whether employers are furnishing safe places of
employment (general duty) and complying with safety and health standards and
regulations (specific duty). In doing so, compliance officers “must use professional
judgment to adequately document hazards in the case file” and are responsible for the

“technical adequacy of each case file” (FIRM Ch. I. A(4)(a)). Since compliance officers

146

are also required to testify in hearings on OSHA’s behalf in contests over their cases,
they are advised to keep an accurate record of conditions in the workplaces they have
visited.

Table 4 includes OSHA’s regional jurisdictions. Regional districts are
determined on a geographic and industrial basis. Each regional district governs four or
more states and territories as to the implementation of OSHA rules. Some regions cover
heavily industrialized states such as New York, New Jersey, Pennsylvania, California,
and Texas, while some regions cover only rural states such as Montana, North Dakota,
South Dakota, Wyoming, Colorado, and Utah. The federal OSHA and the state’s
authority have concurrent powers to enforce health and safety rules in a state where its
“state plan” for occupational safety and health program has been certified by the federal
OSHA. The federal government’s concurrent power is relinquished after the state
receives the “final approval” from OSHA, but the federal government does retain the
authority to the monitor operations of state programs.

The levels of federal OSHA enforcement activities vary across regions. Table 5
shows the total number of inspections by regions in 1992 and 2000. Region 2 which
includes two industrialized states, New York and New Jersey, had a large number of
inspection visits. OSHA made 3,963 and 3,016 visits in 1992 and 2000 respectively in
New York and 2,426 and 1,980 visits in 1992 and 2000 respectively in New Jersey. In
contrast, the number of inspection visits is very small in region 8 which includes rural
states. In 2000 OSHA made 1,207 visits in Colorado, 425 in Montana, 182 in North

Dakota, and 131 in South Dakota. The total number of inspection visits in region 8 is

147

Table 4. Federal OSHA Remnal Jurisdictions

 

 

 

 

 

 

 

 

 

 

 

 

 

Region States ,

1 Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island,
Vermont

2 New Jersey, New York, Puerto Rico, Virgin Islands

3 Delaware, Pennsylvania, West Virginia, Virginia, Maryland

4 Alabama, Florida, Georgia, Mississippi, Kentucky, Tennessee, North
Carolina, South Carolina

5 Illinois, Ohio, Wisconsin, Minnesota, Michigan, Indiana

6 Arkansas, Louisiana, Oklahoma, Texas, New Mexico

7 Kansas, Missouri, Nebraska, Iowa

8 Colorado, Montana, North Dakota, South Dakota, Utah, Wyoming

9 Arizona, California, Hawaii, Nevada, Guam, Samoa,

10 Idaho, Alaska, Oregon, Washington

 

 

Note: States and US. territories that operate their own programs are italicized.

148

 

Table 5. Federal OSH Inspections by Region and State (1992, 2000)

l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Region State 1992 2000 Note

1 Connecticut 1,201 776

Maine 663 746

Massachusetts 1 ,95 8 1 ,560

New Hampshire 386 315

Rhode Island 386 282

Vermont N/A N/A State plan
2 New Jersey 2,426 1,980

New York 3,963 3,016
3 Delaware 150 164

Pennsylvania 2,867 2,955

West Virginia 567 484

Virginia N/A N/A State plan

Maryland N/A N/A State plan
4 Alabama 1,198 821

Florida 2,238 1,643

Georgia 1 ,686 1 ,449

Mississippi 568 370

Kentucky N/A N/A State plan

Tennessee N/A N/A State plan

North Carolina N/A N/A State plan

South Carolina N/A N/A State plan
5 Illinois 2,732 2,445

Ohio 3,168 2,595

Wisconsin 1,849 1 ,258

Minnesota N/A N/A

Michigan N/A N/A

Indiana N/A N/A
6 Arkansas 703 517

Louisiana 862 454

Oklahoma 898 449

Texas 4,897 3,152

New Mexico N/A N/A State plan
7 Kansas 483 585

Missouri 1,582 1,386

Nebraska 259 290

Iowa N/A N/A State plan
8 Colorado 1,205 1,207

Montana 308 425

North Dakota 238 182

South Dakota 181 131

Utah N/A N/A State plan

Wyoming N/A N/A State plan

 

 

 

149

 

Table 5 (Cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Region State 1992 2000 Note

9 Arizona N/A N/A State plan
California N/A N/A State plan
Hawaii N/A N/A State plan
Nevada N/A N/A State plan

10 Idaho 440 469
Alaska N/A N/A State plan
Oregon N/A N/A State plan
Washington N/A N/A State plan

 

Note: US. Territories (Puerto Rico, Virgin Islands, Guam, and Samoa) are excluded.

150

 

 

smaller than the total number of inspection visits in New York in 2000. Even within a
regional district, OSHA inspections tend to concentrate on certain states. For instance,
most of OSHA inspection visits were made to Texas in region 6. In 1992 there were
4,897 inspection visits in Texas, while there were only 703 in Arkansas, 862 in
Louisiana, and 898 in Oklahoma.

Federal OSHA enforcement activities vary also across different points in time.
Figure 8 portrays the longitudinal trend of enforcement activities by violation types
between 1982 and 2000. The number of citations per inspection for non-serious
violations slightly increased in the late 19803 and then monotonically decreased
thereafter. In contrast, the number of citations for serious violations per inspection
substantially increased in the early 19903 and moderately decreased in the mid-19903.
The number of citations for willful and repeated violations remained relatively constant
during the period. These two series seem to be stable because the willful and repeated
violations are found only rarely, compared to non-serious and serious violations.

The amount of financial penalties imposed on violations also varies substantially.
Figure 9 portrays the change in penalties per inspection in 1984 constant dollars by
violation types between 1982 and 2000. Penalties on non-serious violations remained
stable at the lowest level. Penalties on repeated violations moderately increased in the
early 19903 and remained at that level thereafter. Penalties on serious and willful
violations fluctuated greatly during the period. The amount of financial penalties on
willful violations rapidly increased in the late 19803, plummeted in the early 19903,

substantially increased in the mid-19903, and then began to decrease thereafter. Penalties

151

 

 

Figure 8. Citations per Inspection by Violation Types (Federal, 1982-2000)

 

 

 

 

 

 

 

ID —1

Q ...t
c
i
g I, \\ - - I I ~~~~~~~
8N- /
.g
.8
o .1

o — nonsm:fl;fi2&.fiflfl:£t%fimzm3mamen-memameflflae

I I I I I I l I I l
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
Year
Non-Serious ----- Serious
----------- Willful —-—- - Repeated

 

 

 

152

 

Figure 9. Penalties per Inspection by Violation Types (Federal, 1982-2000)

 

 

 

 

 

 

 

 

 

 

(1984 constant dollars)
8
o '1
N
/ ‘\
r~~—-’ \
.5 8 _ [I \
§ ‘- 5‘ I \ a ’ ’ ’ -
m l '. l \ / I
.c— i i I
.- o 0: .I‘ I I
38 - :' = m.
a 1- : g. I}. “ l,‘
‘6 i .\ ". '0' \~
5 .-' , ....... .
m o '9 / s‘ “a" ‘~I .o' N‘
C 0 ~ 11' / i. a.
d.’ ‘9 ,I .V. ,.
I /
x I .‘, ..—
------—.—_:’A” ./ ~"0—.~.-— ..... ’
O - mg11.‘.::::—--—-'—’ fi'd/ \
V T I I I I I I f I
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
Year
Non-Serious ----- Serious
----------- Willful —-—'- Repeated

 

 

153

 

 

h“.

on serious violations skyrocketed in the early 19903, substantially decreased in the mid-
19903, and then slightly increased from then on.

These longitudinal patterns appear to match the tone and content of the legislative
discourse on occupational safety and health regulation during the period. The policy
debates on the Risk Notification Bill between 1986 and 1988, which generated strong
pro-labor signals, provided the beginning of the upward trend of citations of serious
violations, and penalties on serious and willful violations. The Democratic OSHA
overhaul drive between 1992 and 1994, which was aimed at stronger enforcement of
occupational safety and health rules, helped OSHA sustain the stringency of its
enforcement activities at the highest level. The following Republican OSHA reform
campaign from 1995 to 1996 under the “Contract with America” ended the upward
trends, and citations of serious violations and penalties for them decreased substantially
during the period.

These various aspects of enforcement have been examined in previous research
(Marvel 1982; Thompson and Sicchitano 1985; Scholz and Wei 1986; Scholz et al. 1991;
Scholz and Gray 1997; Headrick et al. 2002). Three measures that have been examined
by other researchers are the number of inspections, the number of citations, and the
amount of financial penalties. In this study I focus the analysis on the number of
inspections since the number of inspection visits to establishments can be considered as
the most comprehensive measure of regulatory stringency as well as regulatory
performance for several reasons. First, the number of inspections represents the
important aspects of decisions made by OSHA regarding the distribution of limited

resources. While OSHA has only an inspection staff of only 2,000 or less, the federal

154

OSHA is supposed to cover workers in all industries. OSHA has taken an approach that
aims at providing employers with the incentives for compliance with standards by
inspecting a very small portion of the total number of workplaces. The main purpose of
OSHA inspection activities is not to detect all violations of rules in all establishments.
Instead, the strategy is to provide a sense of reward to those who comply with rules by
some visible acts of punishing detected violations. Thus, the activities of visiting firms
themselves are expected to help maintain safety in workplaces to some degree (Scholz
and Gray 1997). Considering the limited resources that OSHA has and the approach the
agency has taken, the counts of inspections can inform us about differing levels of
commitment to stronger enforcement of regulatory rules.

Second, the number of inspections is a more transparent measure of the
enforcement agency’s regulatory decision than other measures. While whether to visit

an establishment is determined mostly by the agency’s decision, whether to cite

violations or whether to impose financial penalties is affected not only by the inspectors’

judgment but also by various factors of the establishment. For example, OSHA
inspectors can choose to visit an establishment, but they cannot determine the
establishment’s nature of work and its owner’s willingness to provide safe work
conditions. That is, the number of inspection reflects OSHA inspectors’ intention more
accurately than other measures such as the number of citations and the amount of
penalties (Mendeloff 1979). The latter two may be highly contaminated by factors that
cannot be controlled by OSHA inspectors.

Third, the number of inspections also provides us with information about the

level of regulatory stringency in two ways. One incidence of an inspection involves a

155

variety of activities that may be burdensome to employers. Inspectors should enter firms
without prior notice and make employer and managers provide their records of
workplace safety and health; the inspectors should walk around the establishment to
physically examine workplace conditions; the inspectors then hold a concluding
conference with management and workers’ representatives; once they return to their
office, the inspectors report results of their inspection visits to the centralized
information management system and mail citations of violations to the employer.
Second, one inspection-visit means several citations of violations and financial fines of
some thousand dollars as Figures 7 and 8 clearly present. Since there are some expected
numbers of citations and some expected amounts of financial penalties per inspection, a
greater number of inspections also represent a greater number of citations and a greater

amount of fines.

5.2. Modeling Federal OSHA Enforcement Activities

In chapter 3 I theoretically demonstrated that different aspects of inter-institutional
dynamics, such as the location and the size of the policy equilibrium interval, affect
bureaucratic actions in two ways: the level of policy production (the conditional mean)
and the variability of policy choices (the conditional variance). A change in the
conditional mean of the policy outputs, which corresponds to changes in institutional
preferences, suggests the presence of bureaucratic responsiveness to political institutions.
And the change in the conditional variance of bureaucratic policy choices, which
corresponds to the level of policy divergence among institutions, indicates the degree of

bureaucratic autonomy.

156

 

The Maximum Likelihood Heteroscedastic Normal Regression is used here to
examine those underlying processes involving political institutions and bureaucratic
actions as discussed in chapter 3. The model can be presented in the simplest form as

follows:

 

.- " 2 I —.,—1

Policy Output Level ( ,u" ) = a0 + ale + asz

Variability of Policy Choices (0“,, )— = exp(/30 + [91W :11 + #2 Wm)

Y,-, is the dependent variable, which is OSHA’s regulatory activities in a state i for
a year t; p“ is the conditional mean of the dependent variable, that is, the policy output
level; Xm is a set of variables that includes measures for the location of the policy
equilibrium interval on a one-dirnensional policy space and issue characteristics; X,~,2 is a
set of other control variables; a: is the conditional variance of the dependent variable,
that is, the variability of policy choices; Wm is a set of variables that includes measures
for the size of the PEI and issue characteristics; and Wm is a set of other control
variables. The level of regulatory policy outputs is a function of the location of the PEI
on a policy space and other control factors, which offers an empirical test for the
bureaucratic responsiveness thesis; the variability of bureaucratic policy choices is a

function of the size of the PEI and other control factors, which offers a test of the

bureaucratic autonomy thesis.

157

5.2.1. Dependent Variable (Y,,)

The dependent variable Y,, of this study is the number of inspections by federal OSHA in
a state i for a year I. Since there is a considerable gap among different states in terms of
the level of industrialization, direct comparison of counts of annual inspections across
states entails problems. In order to obtain state-by-state comparability, I normalize the
measure by using the size of non-farm industries in a state. Thus, the final measure for
the dependent variable is the number of inspections per 1,000 non-farm employees
conducted by federal OSHA in a state (i) for a year (t). The period of time is between
1982 and 2000 due to the limited availability of OSHA’s Intergrated Management

Information System (IMIS) data on federal and state inspections prior to 1982.

5.2.2. Model Specification for Bureaucratic Responsiveness ( [4,)

The Location of the Policy Equilibrium Interval

The responsiveness of OSHA enforcement officers to the policy preferences of
national policymakers can be tested by examining the effects of pro- or anti-regulation
movement of the location of the policy equilibrium interval. If the policy equilibrium
interval moves toward stronger OSHA regulation, politically responsive OSHA officials
are predicted to increase their regulatory activities, thereby yielding a greater conditional
mean of the dependent variable. 1 measure the location of the policy equilibrium interval
by using the 1St dimension Common Space (CS) Nominate scores for legislators and

presidents, which enable use to map policymakers’ preferences on the liberal-

158

 

conservative ideological space (Poole 1998)]. Since the 1St dimension CS Nominate
scores can captures about 90 percent of the roll call votes (Poole and Rosenthal 1997)
and the narratives in chapter 4 suggested that the ideological dimension of pro-labor
(left) and pro-business (right) was prominent, we can consider CS Nominate scores as
accurate indicators of the policy positions of those policymakers on OSHA issues. In
order to avoid confusion in a statistical analysis, the order of original scores was reversed
so that a greater value represents more liberal (pro-OSHA regulation) policy preferences,
ranging from -1 to 1.

The PEI can be located by measuring the lower and upper limits’ position in the
policy space. In a one-dimensional space, Lower Limit of the PEI is min(-) and Upper
Limit is max(-) where “-” includes the CS Nominate scores of veto players. Three PEI’s
are constructed from three alternative sets of veto players: the majoritarian set = {the
president, House chamber median, Senate chamber median}, the distributive politics set
= {the president, House chamber median, House committee median, Senate chamber
median, Senate committee median}, and the party government set = {the president,
House chamber majority party median, House committee majority party median, Senate
chamber majority party median, Senate committee majority party median}. I call these
the Majoritarian PEI, the Distributive PEI, and the Partisan PEI, respectively. Both the
Lower Limit and the Upper Limit of these PEI’s are expected to have positive effects on
the level (conditional mean) of OSHA inspections.

Table 6 includes the Lower and Upper Limits of the three alternative PEI’s in the

91”~106th U.S. Congresses. This table clearly shows that the political institution that

 

' The CS Nominate scores have been updated continuously to include all presidents and members of
Congress between 1937 and 2002. Data were accessed at http://voteview.uh.edu in December 2003.

159

Table 6. Policy Equilibrium Intervals (The 91"~106"' U.S. Congresses)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Majoritarian PEI Distributive PEI Partisan PEI
Congress Lower Upper Lower Upper Lower Upper

Limit Limit Limit Limit Limit Limit
91st P SM P SCM P HCMM
92'“I P SM P SCM P SCMM
93rd P SM P SCM P SCMM
941? P HM P SCM P SCMM
95th HM P HM P HMM P
96’" SM P SM P HMM P
97*“ P HM P HCM P HCMM
93‘“ P HM P HCM P HCMM
99fh P HM P HCM P HCMM
10W P HM P SCM P SCMM
101st P HM P SCM P SCMM
102'“I P SM P SCM P SCMM
103'“ SM P SM P HMM P
1041F HM P HCM P HCMM P
105“ HM P HCM P HCMM P
106m HM P HCM P HCMM P

 

 

Note: P: President

HM: House Median
SM: Senate Median
HCM: House Labor Committee Median
SCM: Senate Labor Committee Median

HMM: House Majority Party Median
SMM: Senate Majority Party Median

HCMM: House Labor Committee Majority Party Median
SCMM: Senate Labor Committee Majority Party Median

160

constitutes the limits of the policy equilibrium interval is not fixed not only across
different PEI’s but also in each PEI. For example, regarding the majoritarian PEI the
conservative president (P) and the liberal Senate median (SM) defined the set of policy
options in the 91”~93rd Congresses). At later times, however, the majoritarian PEI has
been defined by policy positions of the conservative president and the liberal House
median (HM) in the 97th~101th Congresses), and by those of the conservative House
median and the liberal president in the 104m~106th Congresses. For the distributive PEI,
the president and the House and Senate Labor Committee medians played crucial roles.
The conservative president and the liberal Senate Labor Committee median (SCM)
established the limits of the distributive PEI in the 91-11-94th and in the 100‘"~102“d
Congresses. At other times, the distributive PEI was defined by the conservative
president and the liberal House Labor Committee median (HCM) in the 97th~99th
Congresses, and by the conservative House Labor Committee median and the liberal
president in the 104th~106th Congresses. Regarding the partisan PEI, the president and
the majority party medians in the House and Senate floors and Labor Committees were
key participants. The partisan PEI was defined by the policy positions of the
conservative president and the liberal Senate Labor Committee Majority Party median
(SCMM) in the 92“"~94th and the 100"'~102“d Congresses. The partisan PEI has also
been defined by the conservative president and the liberal House Labor Committee
Majority Party median (HCMM) in the 97th~99th Congresses, and the conservative House
Labor Committee Majority Party median and the liberal president in the 104“‘~106th
Congresses. The House Majority Party median in the floor (HMM) played a pivotal role

in the 95th, 96th, and 103rd Congresses.

161

 

Issue Characteristics

Issue salience is measured by the number of times that the policy issues were
covered by the New York Times in the previous year. The point of this focus is on how
the character of OSHA issues mediates the effects of the lower and upper limits of the
PEI’s on bureaucratic actions. I expect that as the occupational safety and health issues
become more salient in the national policymaking arena, positive effects of the PEI’s on
the average level of enforcement activities will be reinforced. Since heightened public
attention to OSHA issues is expected to increase chances for the chamber and the
partisan leaders to engage in making policies of occupational safety and health problems,
the magnitude of the conditional effect will be greater with the majoritarian and partisan
PEI’s than with the distributive PEI. I include two interaction terms such as Lower
Limit *Issue Salience and Upper Limit *Issue Salience to examine these mediating effects.

Figure 10 presents the level of issue salience measured by the annual counts of
New York Times coverage of occupational safety and health issues. It is apparent that
this policy received increasingly intensive attentions from the mid-19703 to the mid-
19803 and lost much of them afterwards. As the narratives of chapter 4 showed, the
legislative discourse focused on the question of the economic impact of the OSHA
regulation on small businesses and this issue attracted attention from a broader set of
interest groups, professional associations, and the general public in the 19703. The early
19803 were marked by the “Reagan Revolution” and OSHA was cited as an example of
regulation that undermined economic growth by President Ronald Reagan during his
campaigns. For example, Vice President George Bush (senior) criticized that OSHA’s

“unrealistic, overzealous regulators” had threatened the very existence of some of small

162

Figure 10. Issue Salience: New York Times Coverage of Occupational Safety and
Health Issues (1969-2000)

 

 

 

olflllllllllllllll
(D-t

‘_

O

N—t

‘—

‘E

30.,

one

0

O-

V“

o-t
IIIIITIIIIIIIIIII
AAAAAAAAA-t-L-t-td-L-LN
sssaaass°°°§s°®§§
momammomggg N88

Year

163

 

 

businesses. OSHA experienced a reduction of its staff by 22 percent and a loss of budget
by 14 million dollars between 1981 and 1982. It would not be surprising that the Reagan
administration’s quick move to direct OSHA toward pro-business interests generated a
turmoil in public discourse and drew enormous attention to the agency, which was
attacked by both pro-business and pro-labor publics for its regulatory stringency as well
as its emphasis on “trivial” standards. Between 1986 and 1988 the Risk Notification
Bill, which would inform individual workers about their exposure to toxic substances on

the job, drew attentions from prominent policy actors inside and outside the legislative

 

arena. - k

When the general public pays much attention on OSHA and its regulation, we
can expect that the disputes over the relevance of occupational safety and health
regulation can easily go beyond the narrowly defined congressional committee
jurisdictions and that nationally visible political figures including congressional leaders
as well as the president are the ones who actively exert influence on policymaking
processes. The longitudinal trend portrayed in Figure 10 shows that when OSHA issues
attracted a substantial amount of attention during the 19703 and the 19803, the chamber
and the legislative parties played important roles in dealing with such matters as to what
extent exemptions from OSHA regulations should be adopted (1972-1980) and to what
extent workers’ rights to know should be protected (1986-1988). In contrast, when
OSHA issues lost must of attention in the 19903, the issue of OSHA reform (1992-1996)
was discussed only by the House and Senate Labor Committees.

Partisan polarization is considered to be constant at a high level. Legislative

voting records on OSHA bills and provisions from the 19803 to the present reveal that

164

the Democratic and Republican parties’ positions on workplace safety problems became
clearly divergent. Prior to that point, the two parties’ positions were not clearly
distinguishable due to the fuzziness of the ideological and partisan cleavages in the
legislative arena. Especially in the House and Senate floor, the Democratic Party could
not maintain party unity regarding legislative decisions on occupational safety and health
regulation. The majority of the conservative Democrats from Southern states generally
sided with the Republican Party’s efforts to dilute OSHA regulation. Similarly, the
liberal Republicans from the Northeastern states tended to break party lines and join the
Democratic majority. However, as the legislative discourse on the Risk Notification Bill
revealed, the gap between ideological and partisan cleavages disappeared during the
Reagan realignment. Southern Democratic seats in Congress were replaced by real
Republicans. With this collapse of the Southem-Republican conservative coalition, both
parties could maintain a high degree of internal homogeneity, which led a heightened
inter-party polarization in the area of OSHA policy. Thus, during the period under
consideration, the level of partisan polarization remained constantly high as the
ideological cleavage and the party-lines became closely aligned.

Control Variables

Many scholars have speculated that state politicians make every effort to
influence implementation of federal programs so as to maximize regional and local
benefits (Bardach 1977; Chubb 1985; Peterson 1995). Conservative leaders of state and
local governments are reluctant to leave federal officials free to regulate and enforce
central standards and rules in their territories. These prominent local actors have no

reason to welcome federally established regulatory programs and their regulators. In

165

their view, these regulatory programs provide no tangible benefits and may increase their
constituents’ costs of compliance.

However, the empirical evidence is mixed regarding this general expectation
about the effectiveness of informal influence of subnational governments on the federal
bureaucracy. For instance, Thompson and Sicchitano (1985) report that federal OSHA
enforcement activities do not reflect state governments’ preferences; in fact, federal
regulators even increased enforcement levels in Democratic states. On the other hand,
Scholz and Wei (1986) argue that federal OSHA enforcement activities varied
systematically with state-level political institutions. For example, federal OSHA
officials tended to raise the level of regulatory stringency in the states with liberal and
Democratic governments. Wood (1988) also finds similar patterns: federal EPA
enforcement activities tended to increase in more liberal state environments.

In order to control for the possible influence of state institutional preferences on
federal OSHA activities, I include State Government Ideology. This factor is measured
by using state government liberalism scores developed by Berry, Ringquist, Fording, and
Hanson (1998). l rescaled the original scores by a factor of 0.01 so that they can range
from 0 to 1 to minimize statistical confusion concerning the magnitude of coefficients on
this measure. I expect that a more liberal state government will increase OSHA’s
regulatory outputs.

State socio-economic factors also have been thought to affect outcomes of these
federal programs. According to previous OSHA studies, labor and business group
activities that constitute crucial elements of local policy networks may influence the local

operations of federal regulatory programs. Depending on their relative strength in the

166

 

regions and localities, policy implementers may face favorable or unfavorable local
reactions. Despite these plausible expectations, evidence from quantitative tests is
mixed. For example, according to Marvel (1982), because federal agency officials are
less susceptible to local group demands and at the same time unionized industries may be
safer than non-unionized establishments, the negative relationship between labor group
strength and the stringency of OSHA enforcement can be observed. In contrast, Scholz
et al. (1991) and Headrick et al. (2002) report a positive relationship between the local
labor group strength and the level of federal OSHA inspections. According to them,
since local labor organizations can participate in OSHA’s enforcement process through
filing complaints and by reporting imminent dangers to the agency, even the federal
agency officials cannot ignore local interests’ demands for stronger enforcement. In
order to control for the effects of organized labor, I include Non-farm Workers per 100
thousand as a proxy measure of the strength of labor following practices of previous
research.

State economic conditions have also been considered to affect OSHA officials’
regulatory behavior. Since a state’s unfavorable economic conditions may force the
government to take measures on behalf of economic development rather than impose
regulatory constraints on businesses, OSHA regulation may be opposed in states with
bad economic conditions (Chubb 1985; Thompson and Sicchitano 1985; Scholz and Wei
1986; Scholz et a1. 1991; Peterson 1995). I include Per Capita Income to control for
states’ economic conditions.

In order to minimize the possibility of spurious relations caused by a linear time

trend, I include a Time index variable. The lagged dependent variable, Inspect[t-I], is

167

 

also included in order to control for the effects of unspecified year-to-year factors that
may cause the problem of serial correlation. Especially when the maximum likelihood
estimation is used with a panel data set, inclusion of the lagged dependent variable is of
great help in obtaining consistent and asymptotically normal test statistics (Wooldridge
2002: 405-413). I control further for unobserved regional fixed-effects by using dummy
variables for the regional districts of OSHA. The base category is OSHA regional

district 1.

 

5.2.3. Model Specification for Bureaucratic Autonomy (0'3)

The Size of the Policy Equilibrium Interval

The main determinant of the variability of bureaucratic actions is the level of
preference divergence among the veto players. As policy disagreement among those
veto players increases, so does the possibility of autonomous bureaucratic choices. I
measure preference divergence by using Size of the PEI. In a one-dirnensional issue
space, the size of the PEI is simply the distance between a veto player whose policy
position constitutes the lower limit of the policy equilibrium interval and the other veto
player whose position is the upper limit of the policy equilibrium interval. I construct
this variable by the absolute magnitude of Lower Limit and Upper Limit (UL) of the PEI,
that is, |min(-) - max(-)l where “0” includes the CS Nominate scores of the three sets of
veto players.

Three alternative measures for Size can be created, involving three different sets
of veto players: the majoritarian set = {the president, House chamber median, Senate

chamber median}, the distributive politics set = {the president, House chamber median,

168

House committee median, Senate chamber median, Senate committee median}, and the
party government set = {the president, House chamber majority party median, House
committee majority party median, Senate chamber majority party median, Senate
committee majority party median}. Since greater preference divergence among these
veto players means a broader range of politically permissible policy options, I expect a
positive relationship between the size of the PEI and the conditional variance of agency
choices.

Figure 11 portrays changes in the policy equilibrium intervals from the 91St to the
106th U.S. Congresses. The upper panel of Figure 11 shows the majoritarian PEI. The
location of the PEI shifted in the liberal direction between the 94th and the 95th and
between the 102nd and the 103rd Congresses; it moved in the conservative direction
between the 96th and the 97th Congresses. Regarding the size of the majoritarian PEI,
discernible changes took place in several occasions. For example, when conservative
president Ford was replaced by liberal president Carter between the 94th and the 95th
Congress, the size maj oritarian PEI substantially decreased as a result of the emergence
of an unified Democratic government that led the limits of the PEI to be defined by the
liberal House median (HM) and the liberal president (P). The size of the majoritarian
PEI increased between the 96th and the 97th Congresses as conservative president Reagan
constituted the Lower Limit and the liberal House median constituted the Upper Limit of
the majoritarian PEI. In the 103rd Congress, one could observe a small majoritarian PEI
due to the unified Democratic government of a short period of time.

The middle panel of Figure 11 portrays changes in the distributive PEI. Although

the change in the location of the distributive PEI is very similar to that of the

169

 

Figure 11. Policy Equilibrium Intervals (The 91“~106"' U.S. Congresses)

Majoritarian PEI

 

5;? l I 1 L
iii I I I l I?

‘7 4 -
3" <9 3'5 9.“ 913 a“) 3‘ 915 3°» (39 a" (it (65 a“ .513 (at
U.S. Congress

 

 

 

Distributive PEI

 

 

 

F— b-
2‘" “
' .
so— ~

3- P

I

I'r)—t i—

I .5.

IO

N— h—

‘T r-

 

 

 

%\I g(LI 8'51 %&I %%I %%I QI %%I ggl \%%T \%\I \Q‘L‘ \QEI \le \%%I \§%I
U .S. C ongress

Partisan PEI

 

I
fl

I
I

l
I-—-l
I—I
l--I

I

l

- Liberal

-.75 -.5 -.25 0 .25 .5 .75 1
l

 

 

 

 

 

 

L
I

l
I

 

 

3" 31' 035' a“ 033' if" <2.“ 3" if? a“ (8" (if (65' a“? (63' @Pl
U.S. Congress

 

170

majoritarian PEI, the size of the distributive PEI tends to be larger than that of the
majoritarian PEI. Consistent to the distributive politics perspective, the House and
Senate Labor Committee medians had noticeably more liberal policy ideology than the
rest of the chambers, thereby providing OSHA with a wider range of possible policy
alternatives and a greater degree of autonomy. Especially when conservative presidents
(Nixon, Ford, Reagan, and Bush) were the Lower Limit, the liberal House and Senate
Committee medians (HCM and SCM) increased the distributive PEI by moving the
Upper Limit substantially in the liberal direction in the 91“~94th and the 97“‘~102"d
Congresses.

The lower panel of Figure 11 portrays changes in the partisan PEI. The location
and the size of the partisan PEI differ from those of the majoritarian and the distributive
PEI’s. The partisan PEI changes according to ups and downs of the lower limit that
moves dramatically in either liberal or conservative direction. In general, inter-
institutional divergences and convergences in policy preferences were emphasized by the
partisan PEI more than by the maj oritarian and the distributive PEI’s: the partisan PEI
became smaller than the maj oritarian and the distributive PEI’s when the government
was unified, while the PEI became larger than the other two PEI’s when the government
was divided. For example, when conservative presidents were the Lower Limit of the
partisan PEI, the House and Senate Committee Majority Party medians (HCMM and
SCMM) were substantially more liberal than floor medians (HM and SM) and committee
medians (HCM and SCM). In contrast, when liberal presidents (Carter and Clinton)
were the Upper Limit, the Lower Limit was also constituted by the liberal Majority Party

medians in the floor (HMM) and the committee (HCMM).

171

 

Issue Characteristics

To examine the conditional effect of issue characteristics, I include Issue Salience
and interaction terms of issue salience and the size of the PEI (Size *Salience). I
hypothesize that heightened public attention will reinforce the effect of the size of the
majoritarian and partisan PEI’s more than the distributive PEI on the conditional
variance. Since I expect the size of those PEI’s to have positive effects on the
conditional variance, these effects will increase in the same direction with greater issue
salience of workplace safety and health. As for partisan polarization, I consider it

constant, as discussed previously.

 

Control Variables

In order to examine the level of preference divergence among state institutions, I
include Divided Control of branches of state governments. Although state government
officials such as governors and legislators do not have formal oversight authority, they
may influence operations of the federal bureaucracy through local policy networks
(Scholz at al. 1991). Assuming the effectiveness of subnational policy signals in federal
program implementation, inconsistent signals from different elected officials may
contribute to an increase in the variability of OSHA actions in the field. Thus, despite
the crudity of the measure, I expect that the conditional variance of federal OSHA
enforcement activities will be greater in a state with a divided government. I also control

for regional fixed-effects by using dummy variables for the regional districts of OSHA.

172

 

5.3. Empirical Results

The effects of inter-institutional dynamics on bureaucratic responsiveness and
bureaucratic autonomy as hypothesized from the majoritarian, the distributive politics,
and the party government perspectives are examined in this section. The key questions
are whether OSHA field officers responded to changes in veto player’s preferences,
whether the agency officials could make autonomous choices in the context of inter-
institutional policy conflict, whether issue characteristics affected the relative importance
of the alternative sets of veto players, and how the preferences of OSHA field officials
mediate the effects of institutional preferences on agency actions. The effects of other
subnational control variables will be discussed at the end of this chapter.

The main hypotheses (H1, H2, and H5) are tested by using the ML
Heteroscedastic Normal Regression as discussed earlier. For each alternative set of veto
players, I will discuss empirical results of the ML Heteroscedastic Normal Regression
with regards to bureaucratic responsiveness (HI), bureaucratic autonomy (H2), and issue
characteristic (H5). In addition, I use Quantile regression analysis for an empirical test
of the mediating effects of diverse distributions of the agency officials’ preferences (H3
and H4)---non-constant institutional effects along with varying degrees of skewness of

the agency preference distribution-«as discussed in chapter 3.

5.3.1. Majoritarian Politics
The maj oritarian perspective predicts that representative members of the three
majoritarian institutions-«the president, the House median member, and the Senate

median member---will collectively exert the most influence over OSHA enforcement

173

 

activities. As shown in chapter 4, policy agreements not only between the executive and
legislative branch but also between the upper and lower chambers were as critical for
successful legislative policymaking as for OSHA regulation. Since these institutional
actors sent out policy signals to OSHA, the prediction is that OSHA field officers will be
affected by the policy preferences of these three institutional actors.

Bureaucratic Responsiveness

H1: As the policy equilibrium interval moves in a pro-regulation (anti-regulation)

direction, the level of an agency’s regulatory policy outputs will increase

(decrease).

If the president, the House, and the Senate all favor stronger regulatory measures
for workplace safety, OSHA’s enforcement activities will increase. This theoretical
expectation is consistent with the ‘political control’ or bureaucratic responsiveness
perspective. This preferential change of the maj oritarian veto players is represented by
the horizontal movement of the maj oritarian PEI in a one-dimensional policy space. Pro-
regulation shifts of location measures of the PEI are thus expected to have positive
effects on the conditional mean of OSHA enforcement activities. If this expectation is
correct, coefficients on two measures of the location of the PEI, Lower Limit and Upper
Limit, should be positive in the conditional mean equation.

Results of statistical analysis of the effects of the majoritarian veto-player set on
federal OSHA enforcement activities are presented in Table 7. Statistical results,
however, do not support this proposition of bureaucratic responsiveness. Coefficients on
Lower Limit and Upper Limit of the majoritarian PEI turn out to be significant but with
unexpected signs in the conditional mean equation. All other things being equal, pro-

regulation movement of the majoritarian PEI tends to decrease OSHA enforcement

174

Table 7. ML Heteroscedastic Normal Regression Analysis of the Effects of the

Majoritarian Politics on Federal OSH Enforcement

 

 

 

Variable Model 1 Model 11
Mean Equation
Location of the Majoritarian PEI
Lower Limit -0.517***
(0.145)
Upper Limit -0.619***
(0.1 53)
Issue Characteristics
Lower Limit“ Issue Salience 0.010***
(0.003)
Upper Limit*lssue Salience 0014*”
(0.003)
Issue Salience 0.005* -0.002*
(0.002) (0.001)
Controls
State Government Ideology -0.004 0.000
(0.022) (0.022)
Per Capita Income -0.010*** -0.009**
(0.003) (0.003)
Non-farm Workers -0.002 -0.002
(0.002) (0.002)
Time -0.003 -0.009"'
(0.003) (0.004)
Inspect[t-l] 0695*" 0709*"
(0.031) (0.030)
Constant 6.477 17.546*
(6.144) (6.965)
Variance Equation
Size of the Majoritarian PEI 1.958 3.796
(4.806) (4.897)
Size*Issue Salience -0.131 -0.l74*
(0.087) (0.088)
Issue Salience 0.094 0.1 19*
(0.057) (0.057)
Divided State Government -0.352* -0.315*
(0.137) (0.137)
Time -0.047 -0.059*
(0.029) (0.028)
Constant 88.398 1 10.469
(58.573) (57.312)
Wald Chi-square 2934.16* ** 3038.56***
Log-likelihood 897.00 900.47
N 504 504

 

Note: l"p<0.05, "p<0.01, "*p<0.001 (two-tailed tests). Standard errors are in parentheses. F ixed-effect

dummy variables for regional districts are not reported in this table.

175

 

 

activities. Federal OSHA field officials might not have been closely attuned to policy
signals emanating from the majoritarian veto players. Instead, the federal bureaucrats
might have focused on policy signals generated by the actors involved in the distributive
or partisan politics, including such actors as the congressional committees and the
political parties.

Bureaucratic A utonomy

H2: As the policy equilibrium interval gets larger (smaller), the variability of an
agency’s regulatory choices will increase (decrease).

H2 expects that as policy conflicts among the president, the House floor median,
and the Senate floor median increase, the degree of bureaucratic autonomy measured by
the variability of OSHA regulation actions will increase. The level of preference
divergence among the majoritarian veto players is represented by the size of the
majoritarian PEI, which is the distance between two veto players which constitute the
lower and upper limits of the PEI. A larger majoritarian PEI should lead to a greater
variability of OSHA enforcement actions. That is, the coefficient on Size of the PEI
should be positive in the conditional variance equation.

Statistical results show that coefficients on the size of the majoritarian PEI in
columns 2 and 3 in Table 7 are positive, as expected, but they are not significant at any
conventional significance level. The degree of policy disagreement among the
majoritarian veto players-«the president, the House floor median, and the Senate floor
median-«does not have the expected influence on the degree of bureaucratic autonomy
of OSHA either. This lack of impact of the size of the majoritarian PEI on the
conditional variance of OSHA field activities is not surprising. Given that the fact that

OSHA field officials did not respond to changes in the majoritarian PEI’s location, one

176

 

cannot expect the maj oritarian PEI’s size to have a theoretically consistent effect on
bureaucratic autonomy.

Issue Characteristics

H5-l: As issue salience increases and polarization decreases, the majoritarian PEI

will be more likely to affect the level and variability of an agency’s regulatory

actions.

In terms of the conditional effect of issue characteristics, the effects of the
majoritarian PEI on the level and variability of OSHA enforcement activities are
expected to be greater when OSHA issues are highly salient and do not involve polarized
partisan interests. Since the level of partisan polarization was high during the entire
period under consideration, one may not predict a consistent mediating effect of issue
salience here. Despite this difficulty, I expect that increasing public attention to OSHA
issues should increase the effects of the majoritarian PEI on OSHA enforcement
activities. Statistically, this means that the coefficients on the interaction terms, Lower
Limit*Issue Salience and Upper Limit *Issue Salience in the conditional mean equation
and Size *Issue Salience in the conditional variance equation, should be positive in
direction and relatively great in magnitude.

Statistical results provide mixed support for these expectations. Although the
comparison of the magnitude of the effects across different PEI’s cannot be made at this
point, we can check if the coefficients have the expected directional sign. The
conditional effects of issue salience on the effect of the location of the majoritarian PEI
turn out to be positive and consistent with our theoretical expectation in the conditional

mean equation: heightened issue salience tends to increase the impact of lower and upper

limits on the level of OSHA enforcement activities. In column 2, for example, about 45

177

 

or more New York Times coverage turns the joint effect of the lower limit of the
majoritarian PEI (-0.619+0.014*Issue Salience) to a positive one and this effect is jointly
significant (78:21.28, p<0.001).

However, the effect of issue salience on the effect of the size of the maj oritarian
PEI does not support the expectation. In column 2, for example, about 22 or more New
York Times coverage turns the joint effects of the size of the majoritarian PEI (3.796—
O.174*Issue Salience, 78:51.57, p<0.001). Therefore, issue characteristics do not have

conditional effects that are consistent with our expectations.

 

5.3.2. Distributive Politics

The distributive politics perspective highlights the role of congressional committees in
addition to the majoritarian veto players. In the case of OSHA policy, the members of
the House and Senate Labor Committees were generally more in favor of the protection
of workers from workplace hazards than their parent chambers. As we saw in chapter 4,
the House and Senate Labor Committees played crucial roles in protecting OSHA
programs by rejecting legislative attempts to dilute the Occupational Safety and Health
Act and by proposing reform measures for stronger OSHA enforcement. Therefore, we
might expect that the distributive politics veto-player set---the president, the House and
Senate floor medians, and the House and Senate Labor Committee medians---would
affect the OSHA inspectors’ behavior in a way that should be discernible from how the

majoritarian veto-player set did.

178

 

 

Bureaucratic Responsiveness

According to the bureaucratic responsiveness (political control) thesis, the
movement of the distributive politics PEI in a pro-regulation direction should be
expected to increase the level of OSHA enforcement activities (HI). This movement of
the PEI on a policy space represents the situation where the president, the House, the
Senate, and the House and Senate Labor Committees tend to support OSHA regulation.
In fact, the effects of the distributive PEI can be distinguished from the majoritarian PEI
only when policy positions taken by the Labor Committees’ medians constitute one or
both of the limits so that the location (and of course the size) of the PEI can differ from
the maj oritarian PEI. This can happen only when the Labor Committee members have
more pro-regulation preferences than the rest of the chamber. And in fact, CS Nominate
scores of the Labor Committee members from the 97th to the 106th Congresses were more
liberal than others at the conventional significance level. In the House the mean CS
score of the Labor Committee members and the rest of the chamber were 0.053 and
0.003 respectively, whose difference is significant at the 0.01 level (t=2.59, df=4,385).
In the Senate the mean CS score of the Labor Committee members and the rest of the
chamber were 0.057 and -0.022 respectively and the difference is also significant at the
0.01 level (t=2.884, df=1,009). This explains why members of the House and Senate
Labor Committees blocked numerous legislative proposals that would weaken
occupational safety and health regulation.

Nonetheless, results of statistical analyses included in Table 8 do not support H1.
Pro-regulation shifts of the distributive politics PEI did not lead to an increase in OSHA

enforcement levels. Coefficients on the Lower Limit and the Upper Limit of the

179

Table 8. ML Heteroscedastic Normal Regression Analysis of the Effects of the
Distributive Politics on Federal OSH Enforcement

 

 

 

 

 

Variable Model I Model 11
Mean Equation
Location of the Distributive PEI
Lower Limit -0.500**
(0.182)
Upper Limit -0.165
(0.281)
Issue Characteristics
Lower Limit*Issue Salience 0.009“
(0.004)
Upper Limit“ Issue Salience -0.004
(0.003)
Issue Salience 0004* 0.001
(0.002) (0.001)
Controls
State Government Ideology -0.011 -0.007
(0.022) (0.023)
Per Capita Income -0.009** -0.010***
(0.003) (0.003)
Non-farm Workers -0.002 -0.002
(0.002) (0.002)
Time -0.002 0.003
(0.003) (0.003)
Inspect[t-l] 0689*" 0667*"
(0.031) (0.031)
Constant 3.149 -5.818
(5.515) (6.668)
Variance Equation
Size of the Distributive PEI -4.040 -1 .420
(2.192) (2.197)
Size*Issue Salience 0.017 -0.024
(0.037) (0.037)
Issue Salience -0.007 0.034
(0.032) (0.033)
Divided State government -0.281* -0.304*
(0.137) (0.138)
Time -0.102** -0.051
(0.029) (0.030)
Constant 201 .041 * "' 97.673
(59.117) (61.671)
Wald Chi-square 2827.58*** 2683.48***
Log-likelihood 898.21 898.87
N 504 504

 

Note: *p<0.05, "p<0.01, ***p<0.001 (two-tailed tests). Standard errors are in parentheses. Fixed-effect
dummy variables for regional districts are not reported in this table.

180

distributive politics PEI turn out inconsistent with our expectations. The coefficient on
Lower Limit in column 2 is significant but negative and the coefficient on Upper Limit in
column 3 is insignificant at any conventional significance level. It appears that the
federal OSHA field officials did not increase their enforcement level when the policy
preferences of the distributive politics set of veto players changed in favor of stronger
regulatory activities between 1982 and 2000. Despite the pivotal role played by the
Labor Committees in maintaining a degree of statutory stability in opposition to anti-
OSHA policy change, these Committee members’ preferences did not significantly affect
the level of enforcement activities of the OSHA field officials.

Bureaucratic Autonomy

The bureaucratic autonomy hypothesis expects that an intense policy conflict
among the distributive politics set of veto players will increase the degree of bureaucratic
autonomy (H2). With a larger policy equilibrium interval, OSHA field officials can
choose policy options from a wider range of possible alternatives. Given that the size of
the distributive PEI tends to be larger than the maj oritarian PEI, OSHA officials may
have had a greater degree of autonomy under influence of the distributive politics model
than under influence of the maj oritarian politics model.

However, empirical tests of OSHA enforcement do not provide support for this
hypothesis. The effects of Size of the distributive politics PEI on the variability of
OSHA inspectors’ actions turn out to be insignificant in the conditional variance
equation. The variability of OSHA field officials’ policy choices were not systematically
affected by the extent of policy disagreement among the House and Senate Labor

Committees, the president, and the House and Senate floor medians.

181

Issue Characteristics

H5-2: As issue salience and polarization decrease, the distributive politics PEI

will be more likely to affect the level and variability of an agency’s regulatory

actions.

When OSHA issues do not draw intense public attention and they do not divide
between the Democratic and Republican parties, it is expected that the chamber or
leaders of the parties will be less likely to intervene in the legislative activities in
congressional committees. These circumstances may create a vacuum which allows the
House and Senate Committees to exert their own judgment on OSHA issues. However,
since the level of partisan polarization is considered to be constantly high throughout this
period, there is no way to test issue salience’s impact on the effects of the distributive
PEI. Nonetheless I predict that a lower level of salience of OSHA issues will increase
the importance of the policy messages produced by the House and Senate Labor
Committees. Technically these conditional effects of issue characteristics can be
examined by their significance and magnitude. Lower levels of salience are expected to
increase the effect of the distributive PEI on OSHA actions in both significance and
magnitude. Because the former criterion is difficult to apply with the statistical
specification used in this study, I again focus on the magnitude of the conditional effects.

The conditional effects can be estimated by coefficients on those interaction
terms such as Lower Limit *Issue Salience, Upper Limit *Issue Salience, and Size *Issue
Salience. I expect that estimated coefficients for these terms will be negative in direction
with increasing issue salience. In other words, the expected positive effects of the

location and size of the distributive PEI on the conditional mean and conditional variance

of OSHA actions will be reduced when there is a greater chance that more powerful

182

policymakers do not allow congressional committees to exert their delegated authority on
their own.

The statistical results, however, do not support this expectation. In column 2, for
instance, Lower Limit*1ssue Salience, turns out to have an unexpectedly positive sign.
About 56 or more New York Times coverage turns the joint effect of the lower limit of
the majoritarian PEI (-0.500+0.009*Issue Salience) to a positive one and this joint effect
is significant (x2=9.64, p<0.01) in the conditional mean equation. Likewise, in the
conditional variance equation, the interaction term Size *Issue Salience has a positive
coefficient, making the joint effect (-4.040+0.017*Issue Salience, 18:28.27, p<0.001)
positive when the number of New York Times coverage is greater than 238. From these
results we can conclude that issue characteristics did not have expected conditional

effects on the relative importance of the distributive PEI.

5.3.3. Party Government Politics

According to the party government perspective, political parties can mobilize legislative
support for their own policy goals. Legislative politics can be characterized by inter-
party conflicts and intra-party cohesiveness. Party members are considered to participate
in coordinated efforts to achieve partisan goals through the legislative and policymaking
processes. Policy conflict between parties can be so intense that the possibility of
bipartisan cooperation is very low. From this perspective, the role of party leadership is
crucial. In particular, the majority party’s leaders can use their legislative powers to
reduce the impact of the minority party’s opposition. Thus, legislative policy outcomes

are, for the most part, in accordance with the majority party’s interests.

183

As shown in the previous chapter, the House and Senate Labor Committees often
made decisions following the Democratic majority’s goal. Legislative votes on OSHA-
related provisions and bills show that the conservative coalition of Republicans and
Southern Democrats was weakened and that floor decisions were made along clear party-
lines since the 19803. Thus, this perspective leads us to expect that the party government
veto-player set---the president, the majority party medians in the House and Senate

Labor Committees, and the majority party medians in the House and Senate floors---will

.J ‘33 I. my"... -—
'17

influence OSHA’s enforcement activities.

‘7' 'fisa "

Bureaucratic Responsiveness

 

Statistical results of the effects of the party government set of veto players on
OSHA enforcement activities are presented in Table 9. According to the bureaucratic
responsiveness thesis, we expect that the partisan PEI’s pro-regulation movement will
increase OSHA’s regulatory policy outputs. Empirical results provide some support for
this proposition. In the conditional mean equation, the coefficient on Upper Limit of the
party government PEI in column 3 turns out positive and statistically significant at the
0.05 level in a two-tailed test (a=2.529, s.e.=1.158). With all else being equal, a one-unit
shift of the upper limit of the party government PEI in a pro-regulation direction tends to
increase OSHA inspections per 1,000 non-farm workers by 2.529. However, the data do
not support our expectation about the effect of Lower Limit. The coefficient in column 2
turns out negative. OSHA field officials may have considered the policy preference of
the upper limit more seriously so that they increase regulatory stringency with a
perceived pro-regulation movement of the upper limit. Whereas movement of the lower

limit in the pro-regulation direction may demand some increase in OSHA citations

184

Table 9. ML Heteroscedastic Normal Regression Analysis of the Effects of the
Party Government Politics on Federal OSH Enforcement

 

 

 

 

Variable Model I Model 11
Mean Equation
Location of the Partisan PEI
Lower Limit -0.174**
(0.066)
Upper Limit 2529*
(1.158)
Issue Characteristics
Lower Limit“ Issue Salience 0.004***
(0.001)
Upper Limit*lssue Salience -0.065**
(0.023)
Issue Salience 0.0015* 0.026"
(0.0007) (0.009)
Controls
State Government Ideology 0.012 0.016
(0.024) (0.024)
Per Capita Income -0.008** -0.008**
(0.003) (0.003)
Non-farm Workers -0.003 -0.003
(0.002) (0.002)
Time -0.007 0.001
(0.003) (0.004)
Inspect[t-l] 0.701 *** 0680*"
(0.031) (0.031)
Constant 13.377“ -2.103
(6.779) (7.364)
Variance Equation
Size of the Partisan PEI 2824* 3.254*
(1.414) (1.500)
Size"‘Issue Salience -0.093*** -0.094**
(0.026) (0.028)
Issue Salience 0087*" 0094*"
(0.025) (0.027)
Divided State Government -0.195 -0.169
(0.142) (0.142)
Time -0.210*** -0.133**
(0.039) (0.041)
Constant 410.592*** 256.059"
(77.171) (81.069)
Wald Chi-square 2713.70*** 2592.67***
Log-likelihood 901 .72 902.24
N 504 504

 

Note: ‘p<0.05, "p<0.01, ***p<0.001 (two-tailed tests). Standard errors are in parentheses. F ixed-effect
dummy variables for regional districts are not reported in this table.

185

involving low-cost punishments, this movement may not cause behavioral compliance of
field officials.

Bureaucratic Autonomy

The bureaucratic autonomy hypothesis proposes that the degree of bureaucratic
autonomy as measured by the variability of agency officials’ actions will increase when
policy conflict among the party government veto players such as the president and the
majority party medians in the Labor Committees and in the floors. If the behavior of the
congressional Labor Committees and if voting behavior on the floors are governed by
partisan leaders and their goals, the level of inter-party preference divergence should
define the range of political feasible set of policy options for OSHA officials. And this
underlying process is likely to be perceived by OSHA officials, given the fact that the
partisan PEI has sent the strongest signals to field operations of the rank and file.

There is strong evidence for this expected effect of the size of the partisan PEI on
OSHA’s bureaucratic autonomy. The coefficient on Size of the partisan PEI turns out to
be positive and significant at the 0.05 level in two-tailed tests (fl=2.824, s.e.=1.414 and
,B= 3.254, s.e.=1.500 in columns 2 and 3, respectively). The variability of OSHA field
officials’ actions increased as the president and the majority party median members took
more divergent policy positions. OSHA field officials could capitalize on the policy
conflict between parties in various stages of national policymaking to choose options
from a wider set of alternatives, thereby increasing the conditional variance of their
regulatory actions.

Issue Characteristics

H5-3: As issue salience and polarization increase, the party government PEI will
be more likely to affect the level and variability of an agency’s regulatory actions.

186

 

Regarding the conditional effect of the issue characteristics, heightened issue
salience is expected to increase the effect of the party government PEI on the level and
variability of OSHA’s regulatory actions. Since inter-party policy conflict has been
intense since the early 19803, growing public attention to OSHA issues may motivate the
majority party leadership to take clearer policy position and to make concerted efforts to
achieve their policy goals. The data, however, provide some mixed evidence for this
expectation with the party government PEI. The coefficient on Lower Limit *Issue

Salience in column 2 is positive so that the joint effect of the lower limit of the partisan

 

PEI (-0.174+0.004*Issue Salience, x2=7.99, p<0.05) becomes positive when New York L
Times coverage is greater than 43.5. A heightened level of issue salience tends to

reinforce the positive effect of the lower limit of the party government PEI on the level

of OSHA enforcement activities. But there is no systematically convincing evidence that

the conditional effect of issue salience on the upper limit or the size of the party

government PEI is consistent with our expectation. For instance, in the conditional

variance equation in column 2, issue salience has a negative conditional effect on the size

of the partisan PEI, making the joint effect (2824—0093 *Issue Salience, xz=51.6,

p<0.01) negative when New York Times coverage is greater than 31.

5.3.4. Mediating Effects of the Agency Preference Distribution

In the previous three sections I examined the effects of the location and size of policy
equilibrium intervals on bureaucratic responsiveness and bureaucratic autonomy in
federal OSHA enforcement. The main finding was that the partisan PEI had more

consistent effects than the majoritarian or the distributive PEI’s on the level (or

187

responsiveness) and variability (or autonomy) of federal OSHA actions. However, as
discussed in chapter 3, we have a theoretical expectation that these effects of the PEI will
vary systematically with the extent to which an agency’s policy preferences are skewed
toward either deregulation or regulation in a one-dimensional policy space. In terms of
these mediating effects of agency preference distribution, I proposed the following two
hypotheses.

H3: The effect of the location of the policy equilibrium interval on the production

level of a regulatory agency will increase (decrease) as the skewness of agency

preference distribution increases (decreases).

H4: The effect of the size of the policy equilibrium interval on the variability of a

regulatory agency’s policy choices will decrease (increase) as the skewness of

agency preference distribution increases (decreases).

The main argument of H3 is that if an agency’s preference distribution is
asymmetric in favor of either the lower or higher level of regulatory enforcement, that
agency will be more responsive than an agency with symmetric preference distribution to
changes in the preferences of elected policymakers in national institutions. In other
words, we can theoretically predict that an agency’s policy bias will reinforce the effects
of PEI’s location on the agency’s enforcement levels. Thus, we may observe a positive
correlation between the effects of PEI’s location and the skewness of the agency’s
preference distribution. H4 argues that if an agency’s preference distribution is
substantially skewed to either side of the policy continuum, the agency’s policy choices
will be limited to a very narrow range of potentially feasible options. For this agency,

even a large policy equilibrium interval that would provide a large number of feasible

options might not increase the variability of bureaucratic choices. That is, we can expect

188

 

 

 

a negative correlation between the effects of PEI’s size and the skewness of agency
preference distribution.

In this section, I offer an empirical test for H3. We can consider a population of
OSHA field offices that have heteroscedatic preference distributions in the anti- and pro-
regulation space (see Kelman 1981: 177-220). Some of OSHA field offices may be
skewed toward the anti-regulation direction in a policy space so that they will constitute
a cohort of agencies that tend to enforce OSHA rules at a lower level. Some of them
may be skewed to the pro-regulation and will be members of a group of agencies that
enforce rules at a higher level. Yet others may not be skewed and enforce rules at a
medium level. The first group of agencies will constitute the lower tail of the
population; the second group of agencies will be around the upper tail of the population;
and the third group will be in the middle of the population. Then, we can expect that the
effects of PEI’s location on the level of OSHA enforcement activities will be greater in
the lower and upper tails than in the middle of the population as discussed in chapter 3. I
test this expectation by using quantile regression and compare coefficients on the
measures of the location of the PEI at 0.25, 0.50, and 0.75 quantiles. The 0.25 quantile
represents agencies in the lower tail; the 0.50 quantile represents agencies in the middle;
and the 0.75 quantile represents agencies in the upper tail. Coefficients are expected to
be greater at the 0.25 and 0.75 quantiles than at the 0.50 quantile.

Table 10 includes results of this analysis. Since the upper limit of the partisan
PEI turns out to have the most consistent effect on the level of federal OSHA
enforcement activities in states, I focus on its coefficient. All coefficients at the 0.25,

0.5 , and 0.7 5 quantiles turn out significant, thereby suggesting that the upper limit tends

189

Table 10. Quantile Regression Analysis of the Mediating Effects of the Agency
Preference Distribution on Federal OSH Enforcement

 

 

 

 

Variable Quantiles
0=0.25 6:0.5 6=0.75
Upper Limit 5.350*** 2.9171‘ 4.135"
(1.634) (1.649) (1.518)
Upper Limit*Issue -0.140*** -0.071* -0.086*"‘
Salience (0.034) (0.035) (0.030)
0.057*** 0.029* 0.035”
Issue Salience (0.014) (0.014) (0.012)
0.013 0.022 -0.023
State Government Ideology (0.037) (0.038) (0.03 5)
-0.003 -0.007 -0.01 1*
Per Capita Income (0.005) (0.004) (0.005)
-0.001 -0.002 -0.007*
Non-farm Workers (0.003) (0.003) (0.003)
0.006 0.002 0.001
Time (0.006) (0.005) (0.005)
0679*" 0719*" 0696*"
Inspect[t-l] (0.041) (0.050) (0.045)
-14.941 -4.434 -4.073
Constant (1 1.234) (10.872) (10.626)
Pseudo R2 0.545 0.578 0.614
N 504 504 504

 

Note: lP<0-05 (one-tailed test), *p<0.05(two-tailed test), **p<0.01(two-tailed test), ***p<0.001(two-tailed
test). Standard errors are in parentheses. F ixed-effect dummy variables for regional districts are not
reported in this table. Pseudo R-square is for the fill] model that includes all those fixed-effect dummy
variables.

190

to increase enforcement levels. But a closer examination points to the mediating role of
agency’s preferential bias. Coefficients estimated at either lower (0=0.25) or upper tail
(6=0.75) are greater than the coefficient estimated in the middle (6=0.50). The former
(5.350 and 4.135) are significant at the 0.01 and 0.001 levels in two-tailed tests while the
latter (2.917) is significant at the 0.05 level in a right-tailed test. This result suggests that
agency’s policy bias toward either deregulation or regulation tends to reinforce the
positive effect of the partisan PEI’s location on the agency’s enforcement levels.

Although this result provides evidence that the policy equilibrium’s effect does
not remain constant with different distributions of agency preferences, it still is not a
complete analysis. As discussed chapter 3, we should consider not only how much the
agency is biased but also in what direction it is biased. The direction of an agency’s bias
will determine the relative importance of the upper or lower limit of the policy
equilibrium interval. Since I focus on the upper limit’s effect in this case, one can expect
that OSHA officials in the upper tail, which represent skewness toward stronger
regulation, will respond to the upper limit’s movement more sensitively than officials in
the lower tail will do. This expectation, however, is not supported by the empirical
results. There is no evidence that the coefficient on the upper limit at the 0.75 quantile is
greater or more significant than the coefficient at the 0.25 quantile.

This evidence for the instability of bureaucratic choices in the boundary of the
policy equilibrium interval also provides indirect support for H4. The greater
responsiveness of OSHA field offices in either the lower or the upper tails suggests the
empirical validity of the logic of policy (dis)equilibrium and the importance of the

agency preference distribution. The agency’s policy bias indicates the potential policy

191

disagreement among veto players and agency officials, which is predicted to decrease the
effects of inter-institutional conflicts on bureaucratic autonomy (H4). Thus, the
empirical evidence provided in Table 11 also supports the expected negative relationship

between an agency’s policy bias and the impact of the size of policy equilibrium interval.

5.4. Discussion

Empirical analyses of federal OSHA field officials’ regulatory behavior between 1982
and 2000 show that party government politics prevailed in enforcement of occupational
safety and health rules. Although the empirical evidence was not perfect, the location
and the size of the partisan PEI had effects on OSHA’s responsiveness and autonomy in
a more consistent manner than the majoritarian or the distributive PEI. When the
president and the majority party median members in Congress took policy positions that
were in favor of the protection of workers from industrial hazards (especially the
institution constituting the upper limit), OSHA field officials increased their enforcement
activities. At the same time, as policy conflict among the party government veto players
became intensified, OSHA inspectors could make choices from a wider range of
politically-feasible options, thereby reflecting their own preferences to a greater degree.
These results provide empirical support for the theoretical propositions regarding
bureaucratic responsiveness and bureaucratic autonomy. OSHA field officials responded
to institutional preferences by adjusting the level of their enforcement activities and they
also capitalized on policy disagreements among the pivotal policymakers to make

autonomous bureaucratic choices.

192

 

 

The absence of consistent empirical support for the effects of the majoritarian and
the distributive politics implies that OSHA officials were most attuned to policy signals
that came from inter-party battles over OSHA regulation. Consistent with the narratives
in chapter 4, the policy dimension of “pro-labor” and “pro-business” was reinforced by
the increasing importance of the “Democrat” and “Republican” cleavage. These party-
related differences on OSHA issues may have been so clear that even field officials could
view their daily operations along this dimension of OSHA policy. In the area of OSHA
regulation, Democrats were strongly united in support of stronger measures for worker
protection in workplaces, and Republicans were united in their efforts to dilute OSHA
regulatory rules and protect business interests. Thus, it is unsurprising that OSHA
enforcement activities were most systematically affected by the interactions among the
party government veto players.

The relative importance of the three alternative sets of veto players was not
determined systematically by issue characteristics. Although there was some evidence
that heightened salience of OSHA issues reinforced the effects of some measures of the
location of the policy equilibrium interval on the level of OSHA enforcement activities,
empirical results failed to provide evidence for the conditional effect on the size of the
policy equilibrium interval. However, the relative success of the partisan PEI in
explaining the level and variability of OSHA enforcement actions is consistent to the fact
that OSHA issues divided the competing partisan interests on the topics of small business
exemption and regulatory reform.

The explanatory power of other subnational factors turns out to be extremely

weak compared to that of time-variant factors in this study. Neither state government

193

ideology nor divided state government variables had significant and expected effects on
OSHA enforcement activities. The states’ socio-economic factors such as wealth and
industrialization are not significant determinants of OSHA field officials’ behavioral
patterns. However, it would be erroneous to conclude that all those cross-sectional
variations were explained solely by top-down institutional influences that vary only in
the time dimension. A chi-square test shows that there were significant regional fixed
effects: regional dummy variables in the conditional mean and the conditional variance
equations with the party government set of veto players are not jointly insignificant
(x2=63.36 and x2=70.08 in columns 2 and 3 respectively, and p<0.001 in Table 9).

The top-down influence of national political institutions on agency actions is
conditional on how agency officials’ preferences are distributed. OSHA field offices that
were skewed toward either deregulation or regulation in the policy space were more
sensitive to changes in the preferences of national policymakers than those which were
non-skewed. The instability of bureaucratic choices is due to the fact that bureaucratic
choices around the limits are more likely to fall outside the region of politically-
feasibility when the policy equilibrium interval changes in its location and size. This
process reveals the importance of preference congruence among the bureaucratic agents
and political principals. A skewed agency preference distribution means that there is a
great possibility of preference divergence among veto players and agency officials. As
the tension between these two groups is raised, the possibility of veto players’
punishment also increases and agency officials are likely to adjust to those institutional
forces. Second, a substantial discrepancy in preferences among agency officials and veto

players also suggests that there is little room for agency officials to take advantage of to

194

enhance their autonomy. Thus, the potential policy disagreement among the principals
and bureaucratic agents may lead to an increase in the latter’s behavioral compliance and

a decrease in the latter’s autonomy.

195

CHAPTER 6
STATE OCCUPATIONAL SAFETY AND HEALTH REGULATION

ENFORCEMENT

In this chapter I examine the effects of inter-institutional dynamics on bureaucratic
responsiveness and bureaucratic autonomy involving states’ occupational safety and
health regulation activities. When examining the state agencies’ enforcement activities,
the relationships among political institutions and agencies are more complicated. In
addition to the general question of how inter-institutional dynamics affect an agency, we
must now address the question of whether the preferences of national institutions affect
subnational agencies that implement federally-established workplace rules.

Previous research offers three different views of the state agencies’
implementation of federal regulatory programs (Marvel 1982; Chubb 1985; Thompson
and Scicchitano 1985; Scholz and Wei 1986; Wood 1992; Wood and Waterman 1994).
First, a top-down view argues that state regulatory agencies will be affected primarily by
the national institutions since the latter use a system of legal and budgetary incentives
and sanctions to hold the former responsible for pursuing national policy goals. Second,
a bottom-up view contends that state regulatory agencies are more vulnerable to
subnational demands than top-down influences since state bureaucracies are under
formal control of state political institutions and state bureaucrats tend to share economic
and cultural experiences with local policy actors. Third, a mixed view argues that state
regulatory agencies are affected by both top-down and bottom-up factors and they tend

to consider demands from both the top and the bottom in implementing federal programs.

196

 

 

Regarding OSHA regulation in particular, there have been disagreements over
which level factors are most influential. Marvel (1982) argues that state agencies’
enforcement of federally-established OSHA rules was likely to be ineffective because
state occupational safety and health agencies were out of control of the federal OSHA.
Similarly, Thompson and Scicchitano (1985) contend that a state could have decided to
operate its own programs in order to limit the federal government’s pressure for stronger
enforcement. Their results suggest that a state with a conservative state government and
which is geographically distant from Washington, DC. is more likely to exit from the
federal OSHA program and create its own programs. Scholz and Wei (1986) offer the
most systematic test of state occupational safety and health regulation in an inter-
governmental context. They show that state occupational safety and health agencies turn
out to respond to both the national political institutions such as the president and
Congress and the state political institutions such as governor and state legislatures.

Here I examine state occupational safety and health enforcement by focusing on
the following questions. First, is the top-down influence of inter-institutional dynamics
on state agencies consistent with the hypothesized processes of bureaucratic
responsiveness and bureaucratic autonomy? Second, do the effects of the preference
configuration of national institutions on state agencies vary with the agencies’ preference
distributions in theoretically expected ways? Third, are the national institutional
influences on state agencies significant, or do subnational political forces emerge as

strong determinants of the state agencies’ enforcement activities?

197

 

 

6.1. State Occupational Safety and Health Enforcement Activities

The Occupational Safety and Health Act provided states with opportunities to develop
and operate their own programs to protect workers on the job. Section 18 of the OSH
Act declares that a state program must, in general, “provide for the development and
enforcement of safety and health standards which are or will be at least as effective”
as federal standards. In all stages of its operation, OSHA monitors the quality of the
state program through spot checks and monitoring visits. Currently, 21 states and 2 U.S.

territories have comprehensive “state plan” programs covering both the private and

 

public sectors, including Alaska, Arizona, California, Hawaii, Indiana, Iowa, Kentucky, L
Maryland, Michigan, Minnesota, Nevada, New Mexico, North Carolina, Oregon, Puerto
Rico, South Carolina, Tennessee, Utah, Vermont, Virgin Islands, Virginia, Washington,
and Wyoming.

The states’ occupational safety and health regulation enforcement activities vary
substantially. Table 11 presents the total number of inspections by states in 1992 and
2000. New Mexico Environment Department made 485 inspection visits in 1992 and
367 in 2000. Wyoming Department of Employment conducted 386 inspections in 1992
and 494 in 2000. These numbers can be contrasted with those in some industrialized
states. California Department of Industrial Relations conducted 13,355 inspections in
1992 and 9,532 in 2000. The number of inspections in California is 20 times greater
than New Mexico and Wyoming.

Figure 12 portrays longitudinal patterns of enforcement activities per inspection
of state programs between 1982 and 2000. These data provide us with the number of

citations of non-serious, serious, willful, and repeated violations per one inspection visit

198

 

Table 11. State OSH Inspections (1992, 2000)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

State 1992 2000 Status (as of 2000)
Alaska 1 ,107 487 Final Approval
Arizona 2,246 813 Final Approval
California 13 ,35 5 9,532 Certification
Hawaii 1 ,771 390 Final Approval
Indiana 4,340 2,324 Final Approval
Iowa 802 520 Final Approval
Kentucky 1,126 1,096 Final Approval
Maryland 1 ,900 1 ,717 Final Approval
Michigan 8,763 6,757 Certification
Minnesota 2,910 1,955 Final Approval
Nevada 1 ,83 8 2,3 81 Final Approval
New Mexico 485 367 Certification
North Carolina 1,921 3,772 Final Approval
Oregon 5,588 5,171 Certification
South Carolina 2,272 1,900 Certification
Tennessee 2,407 1,793 Final Approval
Utah 705 1,026 Final Approval
Vermont 61 8 379 Certification
Virginia 2,294 2,716 Final Approval
Washington 7,656 6,855 Certification
Wyoming 386 494 Final Approval

 

Note: U.S. Territories (Puerto Rico and Virgin Islands) are excluded.

199

 

 

Figure 12. Citations per Inspection by Violation Types (State, 1982-2000)

 

 

 

 

 

 

 

 

m —
q. -1
C
.3
E.
21
(0 4
g N \/_/\f\ D
o ,f r
v- - ’/ L
o _. "In our. sandman-vamea'erwa'rrJ-m'om :23: :31: sun-sun; an: WJWJMI'M‘V ‘-
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 E
Year
Non-Serious ----- Serious
----------- Willful —-—-- Repeated

 

 

 

200

 

to an establishment. Compared with federal OSHA activities, the state agencies’
enforcement activities remained relatively stable during the period. There was some
moderate downward trend in the number of non-serious citations between 1982 and
1984, but the level began to monotonically increase until the early 19903 although the
magnitude of the change was only moderate. The number of non-serious citations
dropped a little in 1992 and remained at that level thereafter. The number of serious
citations monotonically increased from 1983 to 1994; between 1994 and 1996 there was
a slight decrease in the number but regained the previous level after 1996. The number

of citations of willful and repeated violations remained at the lowest level throughout the

 

entire period.

These patterns of state inspections are different from those of federal OSHA
inspections in several important respects. First, compared to federal inspections, state
inspections tend to be more lenient. While the number of non-serious citations per

inspection tends to be greater in state enforcement than in federal enforcement, the

 

number of serious citations per inspection is much smaller in state enforcement than in
federal enforcement. Second, while one can expect much a greater number of serious
citations than non-serious citations when a federal OSHA inspector visits a firm, one can
expect very similar number of serious and non-serious citations when a state inspector
visits a firm. Third, while the data series of federal serious citations arose rapidly right
after the legislative initiation of Risk Notification bill in 1986, which generated strong
pro-OSHA signals, the data series of state serious citations rose when the Democratic

reform campaign began under the Clinton administration in the early 19903. State

201

inspection agencies seem to take more time than their federal counterparts to respond to
national policy signals.

Figure 13 presents the amount of penalties per inspection by violation types
between 1982 and 2000. The expected amount of penalties for serious violations per
inspection began to increase from the mid-19803 and rapidly increased between 1991
and 1994. There was a substantial decrease between 1994 and 1996. The level remained
stable at the level thereafter until 1999 and there was a substantial increase in 2000.
Penalties for willful violations increased in the late 19803 and remained at that level with
some moderate changes thereafter. Trends in the other two series-«penalties on non-
serious and repeated violations-«remained stable throughout the period under

consideration.

202

 

 

Figure 13. Penalties per Inspection by Violation Types (State, 1982-2000)

Penalties($) per Inspection

 

 

 

 

 

 

 

 

 

 

 

(1984 constant dollars)
8
g 1
O
o _
:‘2
O
O -
9.
’/\\ I,
8 - r” \\\ ______ J/
to /
/
Ix
’fip?‘:......oO-o... ,o" “u ...........
o_ ‘_------_--"“ .12; .—__=—.t———- “ ~ — -‘ —— *“‘
I I I I I I I I I I
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000
Year
Non-Serious ----- Serious
----------- Willful —-—-- Repeated

 

 

 

203

 

6.2. Modeling State Occupational Safety and Health Enforcement Activities

In order to examine the effects of institutional preferences and inter-institutional
preference divergence on state occupational safety and health regulation enforcement, I
use the ML Heteroscedastic Normal Regression Model as specified in the previous
chapter. The model of state enforcement activities for the most part resembles that of
federal enforcement activities. The dependent variable (Y,-,) is the number of inspections
per 1,000 employees in a state i and for a year t between 1982 and 2000. The level of
state enforcement activities is explained by measures of institutional preferences or the
location of the PEI, issue characteristics factors, and other control variables. For this
model of state agencies’ activities, dummy variables for states are included to control for

state fixed effects. In sum, the conditional mean equation is specified as follows:

,u” = a0 + alLLH + azUL,.1+ a3LL,.1*Salience,.1 + a4UL,.1*Salience,.1
+ a5Salience,-1 + agState Government Ideology,” + a7Nonfarm Workers,“

21
+ agPer Capita Income,” + agTime, + 01,011,.) + z d j State,-
j=2
The variability of the state agencies’ policy choices is modeled by using
measures of inter-institutional preference divergence or the size of the PEI, issue

characteristics, and other variables. The conditional variance equation of bureaucratic

autonomy is specified as follows:

0'5 = exp(,Bo+,61Size of the PEI,.1+,82Size *Issue Salience,-1+,83lssue Salienc,-;

21
+fl4Divided State Government,,,.1+ B5 Time, + Z 6!. Statej)
j=2
Using this model, the bureaucratic responsiveness and bureaucratic autonomy of
the state occupational safety and health agencies can be examined in the context of the

federal system where those state agencies have responsibilities for implementing the

federally-established OSHA program.

204

 

 

6.3. Empirical Results

The interactions among the national institutions can be considered to affect state
agencies’ behavior especially when those state agencies implement federal programs.
Although states are allowed to develop their own programs to enforce occupational
safety and health standards, these state programs are being continuously monitored by

the federal OSHA. OSHA maintains the authority to terminate its approval of a state

 

program if the latter is considered ineffective in achieving statutory goals. For instance,
in 1995 the Virgin Islands’ application for OSHA’s final approval on its program was
rejected on the grounds of the agency’s insufficient resources and poor regulatory
performance. The federal OSHA continuously keeps watch on state occupational safety
and health programs through its spot checks and monitoring visits. Furthermore, federal
OSHA provides state agencies with their operational costs up to 50%. Due to this
hierarchical structure, state agencies might be expected to take into account policy
signals generated by the national institutions. State enforcement officers would thus be
expected to adjust their enforcement levels according to changes in the policy
preferences of the national institutions; at the same time they should be expected to
capitalize on inter-institutional policy disagreements to make autonomous policy

choices.

6.3.1. Majoritarian Politics
According to the majoritarian politics perspective, the movement of the policy

preferences of the president, the House median, and the Senate median in a pro-

 

205

regulation direction will increase the level of state occupational safety and health
enforcement activities while preferential divergence among those institutional actors will
increase the variability of bureaucratic choices.

Bureaucratic Responsiveness

H1: As the policy equilibrium interval moves in a pro-regulation (anti-regulation)

direction, the level of an agency’s regulatory policy outputs will increase

(decrease).

According to H1, we can expect that state agency officials will respond to

 

institutional preferences by adjusting the level of their enforcement activities. The
effects of the interactions among the majoritarian veto players on state occupational
safety and health enforcement activities are presented in Table 12. Empirical test of this
hypothesis reveals strong support for the majoritarian veto players. The effects of two
measures of the majoritarian PEI’s location, Lower Limit and Upper Limit, turn out to be

significant in the expected direction. The coefficient on the lower limit of the

 

majoritarian PEI in column 2 is 0.875, which is significant at the 0.001 level, and the
coefficient on the upper limit of the PEI in column 3 is 1.022, which is significant at the
0.001 level. All other things being equal, a one-unit movement of the maj oritarian PEI

in favor of stronger regulation tends to increase state agencies’ inspections about by 1

per 1,000 state workers. These results provide solid empirical ground for concluding that
state occupational safety and health agencies were held responsible by the national
institutions. We can thus conclude that the subnational agencies’ behavior was
constrained by the hierarchical structure imposed on the relations between the national

and state layers of the U.S. federal system. The federal OSHA’s monitoring efforts on

206

Table 12. ML Heteroscedastic Normal Regression Analysis of the Effects of the

Majoritarian Politics on State OSH Enforcement

 

 

 

Variable Model 1 Model 11
Mean Equation =
Location of the Majoritarian PEI
Lower Limit 0.875***
(0.202)
Upper Limit 1022*"
(0.241)
Issue Characteristics
Lower Limit*Issue Salience -0.017***
(0.004)
Upper Limit“ Issue Salience -0.019***
(0.004)
Issue Salience -0.006** 0005*”
(0.002) (0.001)
Controls
State Government Ideology 0.038 0.053
(0.046) (0.048)
Per Capita Income 0.006 0.006
(0.015) (0.016)
Non-farm Workers -0.011 -0.014
(0.028) (0.028)
Time -0.006 -0.008
(0.012) (0.013)
Inspect[t-l] 0.660* * * 0.659* * "'
(0.038) (0.03 8)
Constant 13.058 16.655
(23.891) (25.519)
Variance Equation
Size of the Majoritarian PEI 22.24?" 21.198***
(5.371) (5.427)
Size*lssue Salience -0.422*** -0.402***
(0.104) (0.105)
Issue Salience 0.260*** 0.248* **
(0.067) (0.068)
Divided State Government 0.234 0.254
(0.227) (0.229)
Time -0.256*** -0.255***
(0.035) (0.035)
Constant 494590” * * 492.461 "' * *
(71.290) (71.977)
Wald Chi-square 6363.29*** 6319.17***
Log-likelihood 361 .68 361.56
N 350 350

 

Note: ‘p<0.05, "p<0.01, ""p<0.001 (two-tailed tests). Standard errors are in parentheses. F ixed-effect
dummy variables for states are not reported in this table.

207

behalf of its political principals to maintain a degree of control over state agencies were
effective.

These results also point to the behavioral difference between federal OSHA and
state agency officials. In the case of federal OSHA, there was no evidence that federal
enforcement officers responded to the maj oritarian veto players in the expected way.
Unlike their federal counterpart, state agencies consistently responded to the preferential
changes of the maj oritarian veto players such as the president, the House, and the Senate
medians. Although any conclusive remarks on this difference in bureaucratic
responsiveness across different layers of the federal system should wait for further
analysis of the effects of the distributive and partisan policy equilibrium interval, state
officials appear to be responsive to the policy signals generated from the interactions
among different branches and chambers.

Bureaucratic A utonomy

H2: As the policy equilibrium interval gets larger (smaller), the variability of an
agency’s regulatory choices will increase (decrease).

According to this hypothesis, state occupational safety and health agencies will
have more room for choices when there is considerable policy disagreement among the
majoritarian institutions. Inconsistent policy signals transmitted through the federal
OSHA to state agencies will increase the level of bureaucratic autonomy by positively
affecting the variability of agency choices.

Statistical results provide consistent empirical support for this hypothesis.
Coefficients on Size of the majoritarian PEI in columns 2 and 3 turn out positive and
significant (,B=22.242, s.e.=5 .371 and ,8=21.198, s.e.=5.427, respectively). The estimated

magnitude of the effect is substantial. A slight change in the size of the majoritarian PEI,

208

 

 

about 3.1% of the mathematical maximum size, will lead to a change of 2 standard
deviations in bureaucratic choices.

Although a numerical comparison of coefficients across the federal and state
programs is not appropriate, these results suggest that policy conflict among the national
policymakers has a stronger effect on the state agencies’ autonomy than on the federal
OSHA officials’ autonomy. In the case of federal enforcement, the effects of the size of
the majoritarian PEI on the variability of agency actions were statistically insignificant.
Thus, we can conclude that the location and the size of the maj oritarian PEI have more
consistent effects on state policy implementers than on federal officials.

One possible explanation for this difference is that state occupational safety and
health agencies may have limited capacities to collect detailed information about
legislative discourse in every stage of the national policymaking process. Unlike the
federal OSHA, which can keep track of subtle changes in the policymakers’ positions
inside the legislative arena, state agencies can be informed of decisions made by
institutions only through indirect sources such as the federal OSHA and state delegates to
Congress. This kind of explanation, however, cannot be conclusive until we examine the
influence of the distributive and party government veto players on state occupational
safety and health agencies.

Issue Characteristics

H5-1: As issue salience increases and polarization decreases, the majoritarian PEI

will be more likely to affect the level and variability of an agency’s regulatory

actions.

The effects of issue characteristics in conjunction with maj oritarian institutional

actors can be examined by looking at the interaction terms. Since an increase in the

209

 

 

salience of occupational safety and health issues is expected to increase the likelihood of
the majoritarian veto players’ intervention in this policy area and to reinforce these

maj oritarian institutions’ effects on state agencies’ responsiveness and autonomy,
positive conditional effects are expected. This hypothesis, however, lacks any empirical
support. None of the estimates for coefficients on interaction terms, Lower Limit*Issue
Salience, Upper Limit *Issue Salience, and Size *Issue Salience, support the expectation
that these will be positive and significant. In fact, these coefficients actually have l"
negative signs. For example, in column 3 the joint effect of the upper limit (1.022- 1

0.019*Issue Salience, x2=18.25, p<0.001) becomes negative when New York Times

 

coverage is greater than 54 and that of the size of the PEI (21 .198—0.402*Issue Salience, P
78:15.26, p<0.001) becomes negative when New York Times coverage is greater than 53.

In other words, heightened attention of the national media did not increase the effects of

the interactions among the majoritarian veto players on the level and variability of state

agency actions.

6.3.2. Distributive Politics

From the distributive politics perspective, one must focus on the role of congressional
committees in the policymaking process. Congressional committees may use the
parliamentary powers delegated by the chamber to pursue their own policy goals, which

are distinct from the rest of the chamber. Rather than acting as a loyal agent of the

 

chamber, the congressional committee plays an important policymaking role as a de

facto independent actor.

210

The question here is whether the policy preferences of the House and Senate
Labor Committees influence the behavior of state agency officials. A traditional view of
the influence of congressional committees in the federal system would argue that the
policy triangle or policy subgovemment remains strong at different levels of the federal
government (Nice and Fredericksen 1995; Sabatier and Jenkins-Smith 1993). Interest
groups make a lobbying effort not only at the national level but also at state and local
levels. These groups provide political support and resources for committee members and
agency officials in Washington and states and localities. Agencies in a certain policy
area can develop a shared framework of policy problems through professional training
programs and inter- governmental cooperation to implement federal programs. Members
of committees in Congress and state legislatures thus may have similar concerns and
seek benefits from the same interest groups and the same bureaucratic professions in
exchange for their legislative decisions. This system of subgovemment in “picket fence”
federalism may create channels through which the House and Senate Labor Committees
can send signals to state occupational and health agencies (Nice and Fredericksen 1995:
11-15).

Bureaucratic Responsiveness

The bureaucratic responsiveness hypothesis (H1) predicts that state occupational
safety and health agencies will respond to the distributive set of veto players’ policy
preferences by adjusting their enforcement levels. As the distributive PEI moves in a
pro-regulation direction, state enforcement levels will increase accordingly.

Results of the empirical tests of this hypothesis are included in the conditional

mean equation in Table 13. There is strong evidence for the hypothesis. Coefficients on

211

 

 

Table 13. ML Heteroscedastic Normal Regression Analysis of the Effects of the
Distributive Politics on State OSH Enforcement

Variable Model 1 Model 11

Mean Equation

Location of the Distributive PEI

 

Lower Limit 0.968***
(0.253)
Upper Limit 1.392*
(0.588)
Issue Characteristics
Lower Limit" Issue Salience -0.019***
(0.005)
Upper Limit*lssue Salience -0.025"'*
(0.008)
Issue salience -0.007** 0010*"
(0.002) (0.003)
Controls
State Government Ideology 0.034 0.068
(0.048) (0.054)
Per Capita Income 0.010 0.006
(0.015) (0.016)
Non-farm workers -0.015 -0.013
(0.028) (0.030)
Time -0.010 -0.012
(0.012) (0.013)
Inspect[t-l] 0.648*** 0653*"
(0.040) (0.040)
Constant 21.218 22.803
(23.936) (26.004)
Variance Equation
Size of the Distributive PEI 2.783 2.926
(2.332) (2.362)
Size*lssue Salience -0.033 -0.038
(0.041) (0.041)
Issue Salience 0.016 0.022
(0.035) (0.036)
Divided State Government 0.305 0.257
(0.226) (0.226)
Time -0.293*** -0.280***
(0.032) (0.033)
Constant 58l.467*** 553.748***
(64.209) (66. 1 82)
Wald Chi-square 6074.41 *** 5539.94***
Log-likelihood 354.15 353 .65
N 350 350

 

Note: *p<0.05, "p<0.01, "*p<0.001 (two-tailed tests). Standard errors are in parentheses. Fixed-effect
dummy variables for states are not reported in this table.

212

Lower Limit and Upper Limit of the distributive PEI turn out to be positive and
significant (a=0.968, s.e.=0.253 and a=l .392, s.e.=0.588 in coltunns 2 and 3,
respectively). A pro-regulation movement of the lower and upper limits of the
distributive PEI significantly increased state agencies’ level of enforcement of
occupational safety and health rules.

These results are different fiom what we found with federal OSHA enforcement
activities: while there was no evidence that federal OSHA officials responded to the
distributive PEI, the state agency officials did. The effectiveness of the location of the
PEI in defining politically-feasible agency choices means that the size of the PEI should
be expected to determine the amount of variation of bureaucratic choices. Therefore, this
estimated difference between federal and state agencies in their responsiveness to the
national politicians could be confirmed more convincingly if the variability of state
agencies’ policy choices is also affected by preference divergence among the president,
the House and Senate medians, and the House and Senate Labor Committee medians.
We now examine the evidence for this hypothesis.

Bureaucratic A utonomy

Results of the effects of the size of the distributive PEI on the variability of state
agency actions are shown in the conditional variance equation in Table 13. Surprisingly
there is no evidence that a higher degree of policy disagreement among the distributive
veto-players significantly affects the variability of policy choices by state agencies.
Coefficients on Size of the distributive PEI in columns 2 and 3 are positive but
insignificant at any conventional significance levels. This result disconfirrns the

bureaucratic autonomy hypothesis. The variability of regulatory policy choices made by

213

 

 

“n hula-t - I.-

state agencies was not determined by the level of preference divergence among the
president, the House, the Senate, and the House and Senate Labor Committees.

This also raises a question about the validity of our inference about the
bureaucratic responsiveness hypothesis. If the preferences of distributive veto players
effectively induce bureaucratic compliance by state occupational safety and health
agencies, they also should be able to define the extent to which state agency officials can
make autonomous policy decisions, thereby significantly affecting the variability of state
agency actions. This lack of empirical support for the effects of the size of the
distributive PEI thus suggests that we cannot conclude that distributive politics at the
national level consistently affected state officials’ behavior in regard to both aspects of
the level and variability of regulatory actions. Accordingly, from the perspective of the
distributive politics theory, state agency officials’ behavioral patterns appear to be
similar to federal OSHA implementers’ behavior: there is only partial evidence that field
officials in these two layers of the federal system behave differently when assessed from
the distributive politics perspective.

Issue Characteristics

H5-2: As issue salience and polarization decrease, the distributive PEI will be
more likely to affect the level and variability of an agency’s regulatory actions.

Regarding the mediating effects of issue characteristics, I expect that the
distributive PEI’s effects on the level and variability of bureaucratic choices will be
stronger when both salience and polarization of OSHA issues are low. In this situation,
the chamber or partisan leaders are less likely to intervene into policymaking in this
policy area. Empirical results for this conditional effect of issue salience on the location

of the distributive PEI support the hypothesis. Coefficients on the interaction terms in

214

 

o 13.-1;

 

the conditional mean equation, Lower Limit *Issue Salience and Upper Limit *Issue
Salience, are statistically significant and have the expected signs. For example, in
column 3, the joint effect of the upper limit is 1.392—0.025*Issue Salience that is
significant (18:1 1.77, p<0.01). As issue salience of workplace safety and health issues
increases, the positive effect of the distributive PEI on the level of state enforcement
activities tends to decrease. However, there is no evidence that the effect of the size of
the distributive PEI on the variability of state agency actions systematically varies with
different levels of issue salience. For example, in column 3 the joint effect (2.926—
0.038*Issue Salience) turns out insignificant at any conventional levels (x2=2.51).
Overall, these results provide only partial evidence that issue salience determines the

relative impact of the distributive veto players on state agencies’ regulatory behavior.

6.3.3. Party Government Politics
The party government perspective argues that bureaucratic actions are affected by the
interactions among the partisan veto players such as the president, and the majority party
medians in the House and Senate floors and Labor Committees. The narratives of
legislative discourse described in chapter 4 and statistical analysis of the federal OSHA’s
regulatory behavior presented in chapter 5 showed that the primary ideological cleavage-
--that is “pro-worker” vs. “pro-business”---was closely aligned with party lines during
the period under consideration and that federal OSHA field officers responded most
consistently to the partisan veto players.

How does this partisan politics affect subnational agencies? Four decades ago,

Grodzins predicted that “Any tightening of the party apparatus would have the effect of

215

 

 

strengthening the central government [in the federal system]” (1960). Vertical
integration of political parties has taken place in both the Democratic Party and the
Republican Party (Gray, Hansen, and Jacob 1999). The national party organizations
such as the DNC and the RNC provide their state and local parties with resources and
services for maintenance and victory in elections. Subnational party platforms have been
increasingly nationalized under these influences of the national party organizations. That
is, political parties became “disciplined.” Thus, partisan initiatives pursued at the
national policymaking arena can flow down to lower levels of the federal system.
Considering this recent development of party system, it may not be surprising if we see
policy signals generated by inter-party interactions spread down to subnational agents via
party channels in the states.

Bureaucratic Responsiveness

If state occupational safety and health agencies complied with the preferences of
the party government set of veto players in the national institutions, they must have
adjusted their enforcement levels according to changes of the partisan PEI in its location
in a policy space. When these pivotal policymakers favor stronger regulatory activities,
state agencies would respond to those preferences by increasing enforcement activities.

Results of empirical tests for these expected effects of the partisan veto players
on state occupational safety and health agencies are presented in Table 14. There is
strong empirical evidence for the impact of the location of the partisan PEI on the level
of state enforcement activities: coefficients on Lower Limit and Upper Limit are positive
and significant as expected in columns 2 and 3 (a=0.411, s.e.=0.101 and a=10.975, s.e.=

2.095, respectively). State agencies showed bureaucratic compliance with the policy

216

Table 14. ML Heteroscedastic Normal Regression Analysis of the Effects of the
Party Government Politics on State OSH Enforcement

 

 

 

Variable Model I Model 11
Mean Equation
Location of the Partisan PEI
Lower Limit 0.41 1* **
(0.101)
Upper Limit 10.975***
(2.095)
Issue Characteristics
Lower Limit*lssue Salience -0.007***
(0.002)
Upper Limit*Issue Salience -0.209"'**
(0.038)
Issue Salience -0.001 0088*"
(0.001) (0.016)
Controls
State Government Ideology 0.053 0.046
(0.051) (0.048)
Per Capita Income 0.013 0.009
(0.016) (0.016)
Non-farm workers -0.014 -0.018
(0.029) (0.028)
Time -0.015 -0.008
(0.014) (0.013)
Inspect[t-l] 0647*" 0647*"
(0.040) (0.039)
Constant 29.473 10.820
(26.71 1) (26.467)
Variance Equation
Size of the Partisan PEI 2.672 2.187
(2.196) (2.156)
Size*Issue Salience -0.049 -0.036
(0.043) (0.042)
Issue Salience 0.036 0.026
(0.040) (0.039)
Divided State Government 0.379 0.434
(0.225) (0.229)
Time -0.300*** -0.269***
(0.041) (0.041)
Constant 593.472*** 532.662***
(81.502) 82.590
Wald Chi-square 6040.32*** 6319.14***
Log-likelihood 353.00 357.78
N 350 350

 

Note: *p<0.05, "p<0.01, *"”"p<0.001 (two-tailed tests). Standard errors are in parentheses. F ixed-effect
dummy variables for states are not reported in this table.

217

 

 

preferences of the party government veto players. In particular, a one-unit movement of
the lower and upper limits of the partisan PEI toward stronger regulation in a policy
space tends to increase state enforcement levels by 0.4 and 10 per 1,000 workers,
respectively.

In fact, the partisan PEI’s effects on state agencies’ enforcement levels are even
more consistent than the effects of the partisan PEI on federal OSHA’s enforcement
levels. In the previous chapter, I found that the partisan PEI (especially its upper limit) is
the only one, among the three alternative PEI’s, that performed as a predictor of the level
of federal OSHA enforcement activities. In the case of state occupational safety and
health regulation, the two measures of the location of the partisan PEI have the expected
positive effects. However, again, any conclusive interpretation should wait until we find
that the variability of the state enforcement actions is also determined by the partisan
PEI.

Bureaucratic A utonomy

The bureaucratic autonomy hypothesis predicts that disagreement among the
party government veto players will increase the variability of state agencies’ actions. If
policy preference among the president and the majority party medians in Labor
Committees and the floors increases, state agencies are expected to choose more variable
policy options. However, results of the empirical tests presented in Table 14 do not
support this expectation. Although I expect positive coefficients on Size of the partisan
PEI in the conditional variance equation, those coefficients are positive but do not

achieve statistical significance at the 0.05 level.

218

 

 

This result suggests that the effect of partisan PEI on state agencies’ behavior is
at best partial. Although the partisan PEI successfully explains the level of the states’
workplace regulation activities, it does not determine the variability of the state agencies’
activities. Thus, as with the case of the distributive PEI, I conclude that the interactions
among the party government veto players have partial effects on state occupational safety
and health enforcement only for responsiveness but not for autonomy.

These results clearly show that there is difference between federal and state
occupational safety and health agencies. While federal OSHA was affected most by the
partisan PEI but not by the majoritarian PEI, state agencies were affected not by the PEI
but by the majoritarian PEI. These different behavioral patterns in the different layers of
the federal system may be attributable to the fact that the state agencies are under the
influence of dual-agent relations. That is, while federal OSHA can be seen as an agent of
the national principals, state agencies can be considered as agents of the agent (federal
OSHA) of the national principals. This structural difference may cause a disparity in
regard to the quality of policy signals that generated by different sets of veto players.
Federal OSHA may have an informational advantage compared with its counterparts in
the states and it may have access to more accurate information about current policy
messages exchanged between partisan leaders who play a pivotal role inside the
legislative arena through undercover contacts. State agencies may not have such a direct
access to legislative processes and some may depend more heavily on institution-level
decisions such as what are the policy positions taken by the president, the House, and the

Senate. State agencies may thus receive policy signals contained in outcomes in an

219

 

 

intermittent way; federal OSHA, in contrast, may be able to capture subtle changes in the
legislative arena in a continuous manner.
Issue Characteristics

H5-3: As issue salience and polarization increase, the partisan PEI will be more
likely to affect the level and variability of an agency’s regulatory actions.

Since partisan polarization is considered to be constant during the period under

consideration, heightened public attention to occupational safety issues will motivate

partisan leaders to take more active roles and make more visible efforts to accomplish

partisan goals in the area of workplace safety. An increased level of activities by those 4 :

 

party government veto players is then expected to reinforce the partisan PEI’s positive LL}.
effects on the level and variability of state enforcement activities. However, state

enforcement data do not provide any support for this hypothesis. Coefficients on

interaction terms of the lower and upper limits of the partisan PEI and issue salience do

not have the expected positive signs. For example, in column 2 the joint effect of the

lower limit is 0.411—0.007*Issue Salience, which is significant (x2=17.71, p<0.001).

 

And coefficients on the interaction term of the size of the PEI have unexpected negative
sign and the joint effect (2.672—0.049*Issue Salience) is not significant at any

conventional level (x2=1.54).

6.3.4. Mediating Effects of the Agency Preference Distribution

In chapter 3 I proposed that the effects of the policy equilibrium interval-«the PEI-«on
agency actions can vary systematically with diverse bureaucratic preference distribution.
While the effects of the PEI’s location on bureaucratic responsiveness will be larger

when the agency’s preferences are skewed to either side of a policy space, the effects of

220

 

the PEl’s size on bureaucratic autonomy will be smaller when the agency’s preferences
are skewed. More specifically, I deduced the following hypotheses:

H3: The effect of the location of the policy equilibrium interval on the production

level of a regulatory agency will increase (decrease) as the skewness of agency

preference distribution increases (decreases).

H4: The effect of the size of the policy equilibrium interval on the variability of a

regulatory agency’s policy choices will decrease (increase) as the skewness of

agency preference distribution increases (decreases).

In chapter 5 we found that the bureaucratic responsiveness of federal OSHA was
greater when it was biased to either lenient or strong enforcement of workplace rules
(H3), which led to the logical conclusion that the agency’s policy bias would lead to a
lower level of bureaucratic autonomy (H4). In this section 1 offer a similar empirical test
for the mediating effects of agency preference distribution with state inspection data. I
again expect that the effects of the PEI ’3 location on the level of state enforcement
activities will be greater at lower or upper tails than at a middle portion of a distribution
of agencies. The former group of agencies near the boundary of the PEI has a greater
probability that it should adjust to new institutional forces as the PEI moves.

Table 15 includes quantile regression estimation of the mediating effects of
agency preference distribution. Since earlier analysis suggests that the maj oritarian PEI
has the most consistent effects on the actions of state occupational safety and health
agencies, I focus on the particular PEI’s effects at 0.25, 0.50, and 0.75 quantiles. First,
the responsiveness of state agencies to the preferences of the maj oritarian veto players
was greater in the lower (0=0.25) and upper (6=0.75) tails. Coefficients on Upper Limit

are greater in magnitude in the tails (1.485 and 1.433, respectively) than in the middle

(0.689). Furthermore, only the former turn out significant at the 0.05 level in two-tailed

221

 

Eta-T

 

Table 15. Quantile Regression Analysis of the Mediating Effects of Agency
Preference Distribution on State OSH Enforcement

 

 

 

 

Variable Quantiles
0:0.25 $0.25 0:05 6:05 6:075 0:075
Lower Limit 0.8881 0.332 0.9291'
(0.538) (0.505) (0.563)
Upper Limit 1.485* 0.689 1.433"
(0.579) (0.545) (0.631)
Lower Limit*Issue -0.015 -0.005 -0.016
Salience (0.011) (0.010) (0.011)

Upper Limit*lssue -0.023* -0.011 -0.023
Salience (0.01 1) (0.010) (0.012)
Issue Salience -0.006 0.004* -0.001 0.002 -0.005 0.006”

(0.005) (0.002) (0.005) (0.002) (0.005) (0.002)
State Government -0.050 0.016 -0.158 -0.161 -0.192 -0.177
Ideology (0.174) (0.177) (0.133) (0.137) (0.151) (0.153)
Per Capita Income -0.003 0.000 -0.033 -0.027 0.005 0.004
(0.043) (0.045) (0.040) (0.042) (0.060) (0.062)
Non-farm Workers 0.074 0.076 0.222 0.222 0.432 0.396
(0.157) (0.172) (0.173) (0.185) (0.230) (0.242)
Time -0.014 -0.022 -0.012 -0.016 -0.055 -0.057
(0.028) (0.029) (0.028) (0.029) (0.047) (0.051)
Inspect[t-l] 0651*" 0.661 *** 0710*" 0726*" 0.624'" 0623""
(0.065) (0.061) (0.061) (0.063) (0.086) (0.089)
Constant 27.892 43.845 24.470 33.61 1 1 1 1.210 1 13.668
(55.301) (57.174) (54.732) (57.049) (93.667) (100.082)
Pseudo R2 0.648 0.650 0.707 0.708 0.748 0.750
N 350 350 350 350 350 350

 

 

Note: tp<0.05 (one-tailed test), *p<0.05(two-tailed test), **p<0.01(two-tailed test), *“p<0.001(two-tailed
test). Standard errors are in parentheses. Fixed-effect dummy variables for states are not reported in this
table. Pseudo R-square is for the full model that includes all those fixed-effect dummy variables.

 

222

tests while the latter fails to achieve statistical significance at the 0.05 level even in a

ri t-tailed test. Coefficients on Lower Limit also are greater and significant at the 0.05
level in right-tailed tests when estimated in the lower and upper tails, while the
coefficient is not significant when estimated in the middle. These results support the
hypothesis that agency’s policy bias will increase the effects of the PEI’s location (H3).

From this evidence, we can expect that heightened preference divergence

 

P
between veto players and state agencies will decrease the variability of bureaucratic
choices (autonomy) (H4). If a state agency consists of officials whose preferences are
very different from the preferences of majoritarian veto players, the agency cannot take g

advantage of its own autonomy since policy conflict between the veto players does not

offer an opportunity to choose options in its own interest.

6.4. Discussion

The regulatory behavior of state occupational safety and health agencies was affected by

 

the interactions among the national political actors. Statistical analysis provides
evidence that state agencies responded to the policy preferences of the national
policymakers and at the same time they maintained autonomy as long as those
policymakers were in conflict. Surprisingly, state agencies responded significantly to all
three modes of inter-institutional dynamics examined in this study----the majoritarian,
the distributive, and the partisan politics. Measures of each of these PEI locations had
significant effects on the state agencies’ enforcement levels. On the other hand, only the

majoritarian PEI had effects on the variability of state agency actions: the state agencies’

223

autonomy increased as preference divergence among the majoritarian veto players
increased.

These results suggest that the interactions among the national policymakers
played important roles in defining feasible policy options for state enforcement agencies.
Especially since state occupational safety and health agencies implement a federally
established OSHA program in a hierarchical arrangement, policy preferences of the
national policymakers are likely to be considered seriously by state implementers. h;

Federal OSHA’s monitoring activities such as spot checks and monitoring visits seem to

 

play crucial roles in maintaining state occupational safety and health programs in the g -
way that national politicians think desirable. Policy messages generated by national i
institutions may have been transmitted to state implementers. ‘

Analysis of state enforcement also reveals differences between federal OSHA and
state agencies. While the partisan set of veto players and their preferences had the most
significant effects on the regulatory behavior of federal OSHA officials, the majoritarian

set of veto players had the most influence on the state implementers’ behavior. This

 

difference may be attributable to the structure of the federal system. The two layers of
the federal system situate federal agencies and state agencies in different enviromnent.
Federal agencies can be considered as agents of national principals, and they may
maintain intimate and continuous interactions with their national masters. On the other
hand, state agencies can be viewed as agents of the agents of national principals (Chubb
1985; Wood 1992). Federal agencies, which serve the national principals, control state
agencies. This structural difference may cause informational disparity between federal

and state agencies. Federal OSHA officials may be informed more than state agency

224

officials about the happenings inside the national policymaking arena. Presumably
federal OSHA officials may have access to information about the legislative debates and
subtle political changes. For instance, partisan politics in the legislative arena may be
well known to federal OSHA officials, but the federal bureaucrats may not transmit
detailed information about subtle political changes to state agencies. Unlike such well-
informed federal officials, state agencies’ officials will be most sensitive to inter-branch
or inter-chamber processes rather than to other intra-institutional processes such as the
congressional committees and the majority party medians.

These influences of the national policymakers on state agencies were contingent
on the state agencies’ preference distribution in a way that was similar to how
institutional effects on OSHA were conditional on OSHA’s preference distribution.
Even in a complicated situation in which state agencies are monitored by the federal
OSHA which is monitored by the national political institutions, the fundamental features
of principals-agent relations remain basically the same. The effects of the preference
configuration of national veto players on state agencies rested in part on how state
agencies’ policy preferences on workplace safety were distributed and to what extent the
agency preferences were different fiom the national veto players.

Empirical support for the effectiveness of the top-down influences can be clearly
contrasted with the lack of empirical support for the states’ institutional effects on state
occupational safety and health enforcement. Throughout analyses in this chapter, state
government ideology, a measure of the states’ institutional preferences, and divided state
government, a proxy for preference divergence among state institutions, did not have

significant effects. Furthermore, other socio-economic factors at the state level did not

225

 

 

 

have significant effects either. Although these results provide evidence for the top-down
perspective, one should note that measures of state institutional preferences were
underdeveloped in comparison with measures of national institutional preferences. Thus,
it appears to be difficult to conclude that the interactions among state-level institutional

actors did not matter in the state agencies’ enforcement.

 

 

226

CHAPTER 7

CONCLUSION

The proper role of bureaucracy and its unelected officials in democratic governance has
been controversial. Underneath these disputes lies the dichotomous way of thinking that
can be succinctly expressed as “democratic control vs. bureaucratic autonomy.” Many

scholars take positions regarding this question in support of either the idea that

 

gr
representative institutions should maintain a degree of political control over the f
unelected officials in administrative apparatus or the opposite idea that bureaucratic
organizations should be empowered and granted greater autonomy to better serve f

societal interests. The thrust of this dichotomous thinking is that bureaucratic autonomy
is incompatible with the principles of democracy.

The validity of this dichotomous thinking has been challenged in this study. This
study shows that government agencies are subject to political controls to some extent but

also that the agencies can strategically maneuver among competing and divided political

 

institutions to make autonomous policy choices. Political influence on agency actions
could be identified since it provides agency officials with a set of feasible agency
choices. Agency officials have to adjust their actions in response to changes in the
political institutions’ policy preferences. However, bureaucratic autonomy is not
completely eliminated in any case. Even when the agency should change its actions in
conformance to substantial movement of the key institutions’ positions on a policy issue,
the agency officials still can autonomously choose from a wide range of diverse policy
options. In other words, the presence of political responsiveness by the agency does not

necessarily deny the existence of some degree of bureaucratic autonomy. Furthermore,

227

even when the agency is able to maintain its current way of doing its work due to
political stability, the agency may suffer from a narrowly defined set of feasible policy
options. That is, the lack of political responsiveness is not necessarily a testimony to
great bureaucratic autonomy. In sum, when viewed from the logic of the policy
equilibrium interval, political influence and bureaucratic autonomy are simply different
aspects of the same relationship between the political institutions and the bureaucracy.
An extensive analysis of behavior of federal and state occupational safety and
health agencies provides some empirical evidence that supports the conceptualization
developed in this study. First, the bureaucratic responsiveness hypothesis (HI) predicts
that the location of the policy equilibrium intervals-«Lower and Upper Limits-«will
have effects on the level of OSHA enforcement activities. In chapter 5 empirical
analyses of federal OSHA activities showed that a pro-regulation movement of the
Upper Limit of the partisan PEI tended to increase the federal agency’s regulatory
outputs. In chapter 6 pro-regulation movements of the Lower and Upper Limits of the
majoritarian, distributive, and partisan PEI’s tended to increase state OSHA agencies’
regulatory outputs. Second, the bureaucratic autonomy hypothesis (H2) predicts that the
size of the policy equilibrium interval will systematically affects the amount of
bureaucratic autonomy or the variability of agency policy choices. In chapter 5 empirical
analyses showed that an increase of the size of the partisan PEI tended to increase the
conditional variance of the federal OSHA’s actions. In chapter 6 an increase of the size
of the majoritarian PEI tended to increase the conditional variance of the state agencies’

actions.

228

 

 

These institutional effects on bureaucratic behavior are expected to be mediated
by the agency’s preference distribution. In general, when bureaucrats’ preferences are
skewed in either the anti- or pro-regulation direction, there is a greater possibility that the
agency’s officials have to adjust their policy choices in response to changes in political
principals’ preferences (or a horizontal shift of the policy equilibrium interval in the
policy space). Closely related, when bureaucrats’ preferences are skewed in either the
anti- or pro-regulation direction, the officials do not have great room for choice even
under the condition that a set of politically-feasible options is widely defined by the
political principals. Thus, the shape of the bureaucratic preference distribution-«the
agency’s policy bias-«can mediate the institutional effects on bureaucratic behavior in
predictable ways: the skeweness of the agency’s preference distribution will reinforce the
institutional effects on bureaucratic responsiveness (H3) while the skewness of the
agency preference distribution will have a negative impact on the institutional effects on
bureaucratic autonomy (H4). In other words, the effects of institutional dynamics on
bureaucratic responsiveness and autonomy vary systematically with different
distributional shapes of the agency’s preferences.

Regarding empirical tests for these expectations about the mediating effects of
the agency preference distribution, analyses of OSHA enforcement activities provide
some evidence. In chapter 5 empirical analyses showed that the effects of the location of
the policy equilibrium interval (the Upper Limit of the partisan PEI) on federal OSHA’s
responsiveness were greater and more significant in states where federal OSHA tended to
under- or over-produce regulatory outputs, which was considered to reflect the agency’s

policy bias, than in other states. In chapter 6 the effects of the policy equilibrium

229

 

 

interval’s location (the Lower and Upper Limits of the majoritarian PEI) on state
OSHA’s responsiveness were greater and more significant in states where their agencies
under- or over-produced outputs than other states. These results clearly suggested that
the institutional effects on bureaucratic responsiveness varied with different degrees of
agency policy bias, supporting H3. From this conclusion, we also may predict that the
effects of the inter-principal disagreement (the size of the policy equilibrium interval) on
bureaucratic autonomy will be non-constant due to the mediating effects of the agency
preference distribution as hypothesized by H4.

What factors can account for the relative importance of different sets of veto
players in affecting bureaucratic behavior? This question remains unanswered in this
study. Building mainly on the congressional literature that has suggested that different
sets of veto players would play a pivotal role depending on situations, this study
predicted that the characteristics of the issue at hand, such as salience and partisan
polarization, would motivate different sets of veto players to take a part in directing
agency behavior (H5). However, this expectation about the conditional impact of issue
characteristics was not supported by any systematic evidence. In chapter 5, issue
salience tended to reinforce the effects of the Lower and Upper Limits of the maj oritarian
PEI and the Lower Limit of the partisan PEI on federal OSHA’s enforcement levels. But
the issue factor’s effects on other measures of the location and the size of the policy
equilibrium intervals in chapters 5 and 6 were not consistent with the theoretical
expectations.

This study provides several agendas for the future research. First, the effect of

institutional preferences on bureaucratic responsiveness and bureaucratic autonomy can

230

be further investigated in the context of the federal system. Results of current research
highlight the impact of the interactions among the national institutions on subnational
agencies. However, the relationship between the national and subnational layers of the
federal government has not been taken into account in rigorous ways. In the future, one
could examine the effects of national and subnational institutional actors on bureaucratic
actions in conjunction with the distribution of authority between the two layers of the
government. The influence of subnational actors may vary in different federal regimes.
For instance, the subnational institutions may play a more critical role in defining
feasible bureaucratic actions and affecting the variability of bureaucratic choices in a
“devolution” regime where the most part of the national authority is delegated to the
subnational institutions. On the other hand, the subnational institutions’ influence will
be only marginal in “centralized” regime where the national government maintains strict
hierarchical controls over subnational institutions. The bounds of feasible bureaucratic
actions are likely to be defrned by the national institutions but not by the subnational
institutions.

Second, also involving the interactions among the institutional actors in the
subnational layer, the research reported here attempted to include some measures of
subnational institutions’ preferences such as state government liberalism and divided
state government. However, these measures need to be elaborated. In the first place,
these state-level measures are not comparable with the national-level measures of
institutional preferences. This lack of comparability between the federal and state-level
measures makes it difficult to examine the effect of subnational institutions in a

systematic way. This underdevelopment of state-level preference measures keeps state-

231

focused research on inter-institutional interactions from advancing. More elaborated
measures of state institutional actors’ preferences can help us examine the nature of
inter-institutional interactions and their effects on bureaucratic actions in the same way
that the national institutions are examined.

Third, the conditional effect of policy characteristics can be better understood by
cross-policy-area study. Current research has limitations in examining various generic
characteristics of policy such as issue salience and partisan polarization. The effects of
these policy characteristics on the relative importance of different modes of political
interactions (i.e., majoritarian, distributive, and partisan politics) and their influences on
bureaucratic actions can be tested more clearly by a comparative policy study across
areas where salience and polarization differ. This kind of research may be able to show
that the logic of the policy equilibrium interval can be generalized under various
contextual conditions where interactions among different sets of actors are pivotal in
influencing bureaucratic actions.

What theoretical implications does this studyhave? When political controls (or
bureaucratic responsiveness) and bureaucratic autonomy are viewed as products of the
joint actions of multiple principals, many conceptual and empirical ambiguities in past
research can be resolved. First, bureaucracies seem not to respond to all discrete signals
generated by individual political institutions. Instead, they may respond to changes in
the overall preference configuration, which can be determined by interactions of multiple
political principals. Thus, bureaucrats do not always take the wishes of a certain political
institution (i.e., the president or Congress) into account when they implement policy, but

they may consider the preferences of any political institution that plays a pivotal role

232

(i.e., establishes the limits of the policy equilibrium interval) in the course of inter-
principal interactions. In other words, the relative importance of an individual institution
is not determined statically by how formidable each institution’s weapons are. Rather, it
is determined dynamically by the nature of the interactions among the multiple
institutions.

Second, bureaucratic autonomy can be conceived as a matter of degree. The
question should be to what extent bureaucratic agencies are autonomous rather than
whether or not bureaucratic agencies are autonomous. As long as the set of equilibrium
policies consists of a large number of potential policy options and as long as agency
officials can choose the options that are closest to their ideal points, the policy
equilibrium interval can be seen as the area of bureaucratic autonomy. From the
properties of the policy equilibrium interval, we should expect that the degree of
bureaucratic autonomy should change as the extent of disagreement among the political
principals changes.

Third, bureaucratic responsiveness and bureaucratic autonomy are distinct but
integrated processes. Since bureaucratic autonomy has bounds and these bounds are
determined by the interactions among the political principals, bureaucratic
responsiveness is, in fact, a foundation for the concept of bureaucratic autonomy. If one
denies the impact of institutional preferences on bureaucratic choices, we cannot expect
the area of bureaucratic autonomy to be defined. The variability of politically-feasible
bureaucratic choices can be examined only when feasible options for the agency are
constrained by the principals to a certain range. For this reason, the concept of a ‘run-

away-bureaucracy’ is misleading and may be an exaggeration resulting from the myopia

233

of past research. An autonomous bureaucracy can still be held accountable to
representative institutions in a democracy.

Fourth, this study makes it clear that bureaucratic autonomy can be
conceptualized without assuming that bureaucrats can pursue their policy goals by
“cheating.” Previous theoretical models tend to overemphasize the “agent problems” to
depict autonomous bureaucratic actions. From this viewpoint,,one should conclude that
bureaucrats can make autonomous decisions because bureaucrats’ hidden actions and
hidden information may nullify the effects of hierarchical controls by the political
principals. In contrast, this study shows that bureaucrats can make autonomous
decisions due to the lack of agreement among the political principals even under the
condition of complete information in which the political overseers can monitor
bureaucrats’ behavior. Although bureaucrats are not assumed to be “cheaters,” and
hierarchical relations between the principals and the agents are considered to be present,
the bureaucratic agents still can pursue policy goals that are in accordance with their own
interests to some degree due to the macro rules that require multilateral agreement
among the principals in directing bureaucratic agents.

Finally, bureaucratic preferences of a regulatory agency play an important role as
a factor that mediates political influence on bureaucratic outcomes. Political influence
on bureaucratic responsiveness and bureaucratic autonomy can vary with different
distributions of bureaucratic preferences. When the agency officials’ preferences are
skewed so that there is a greater possibility of goal-conflict between the principals and
the bureaucratic agents, the bureaucrats’ behavioral responsiveness will increase and

their autonomy will be less affected by inter-principal preference divergence. That is, the

234

relationship between political institutions and bureaucratic outcomes is not constant and

can be accurately predicted only when bureaucratic preferences are taken into account.

235

BIBLIOGRAPHY

236

BIBLIOGRAPHY

Aberbach, Joel D. 1990. Keeping a Watchfid Eye: The Politics of Congressional
Oversight. Washington, DC: Brookings Institution.

Aldrich, John H., and David W. Rohde. 1999. “The Consequences of Party
Organizations in the House: Theory and Evidence on Conditional Party
Government.” Presented at the Conference on the Congress and the President in a
Partisan Era, Texas A&M University.

Allison, Graham. 1971. Essence of Decision: Explaining the Cuban Missile Crisis.
Boston: Little, Brown.

Alvarez, R. Michael and John Brehm. 1995. “American Ambivalence Towards Abortion
Policy: A Heteroskedastic Probit Method for Assessing Conflicting Values.”
American Journal of Political Science 39: 1055-1082.

Austen-Smith, David, and Jeffrey S. Banks. 2000. Positive Political Theory 1: Collective
Preference. Ann Arbor: The University of Michigan Press.

Austen-Smith, David, and William Riker. 1987. “Asymmetric Information and the
Coherence of Legislation.” American Political Science Review 81: 897-918.

Banks, Jeffrey S., and Barry R. Weingast. 1992. “The Political Control of Bureaucracies
under Asymmetric Information.” American Journal of Political Science 36: 509-
524.

Bardach, Eugene, and Robert A. Kagan. 1982. Going by the Book: The Problem of
Regulatory Unreasonableness. Philadelphia: Temple University Press.

Bardach, Eugene. 1977. The Implementation Game: What Happens After a Bill Becomes
Law. Cambridge: MIT Press.

Barnard, Chester 1. 1938. The Functions of the Executive. Cambridge: Harvard
University Press.

Baumgartner, Frank, and Bryan Jones. 1993. Agendas and Instability in American
Politics. Chicago: University of Chicago Press.

Bawn, Kathleen. 1995. “Political Control Versus Expertise: Congressional Choices about
Administrative Procedures.” American Political Science Review 89: 62-73.

Bawn, Kathleen. 1997. “Choosing Strategies to Control the Bureaucracy: Statutory
Constraints, Oversight, and the Committee System.” Journal of Law, Economics,
and Organization 13: 101-126.

237

Behn, Robert D. 2001. Rethinking Democratic Accountability. Washington, DC:
Brookings Institution Press.

Bendor, Jonathan, A. Glazer, and Thomas J. Hammond. 2001. “Theories of Delegation.”
Annual Review of Political Science 4: 235-269.

Bendor, Jonathan, and Thomas H. Hammond. 1992. “Rethinking Allison’s Models.”
American Political Science Review 86: 301-322.

Bendor, Jonathan, Serge Taylor, and Roland Van Gaalen. 1987. “Politicians,
Bureaucrats, and Asymmetric Information.” American Journal of Political
Science 31: 796-828.

Bendor, Jonathan. 1988. “Formal Models of Bureaucracy.” British Journal of Political
Science 18: 353-395.

Bernstein, Marver H. 1955. Regulating Business by Independent Commission. Princeton:
Princeton University Press.

Berry, William, Evan Ringquist, Richard F ording, and Russell Hanson. 1998.
“Measuring Citizen and Government Ideology in the American States, 1960-
1993.” American Journal of Political Science 42: 327-348.

Binder, Sarah A. 2003. Stalemate: Causes and Consequences of Legislative Gridlock.
Washington, DC: Brookings Institutions Press.

Bond, Jon R., and Richard F leisher. 1990. The President in the Legislative Arena.
Chicago: University of Chicago Press.

Brady, David W., and Craig Volden. 1998. Revolving Gridlock: Politics and Policy from
Carter to Clinton. Boulder: Westview Press.

Brehm, John, and Scott Gates. 1993. “Donut Shops and Speed Traps: Evaluating Models
of Supervision and Police Behavior.” American Journal of Political Science 37:
555-581.

Brehm, John, and Scott Gates. 1997. Working, Shirking, and Sabotage: Bureaucratic
Response to a Democratic Public. Ann Arbor: University of Michigan Press.

Buchinsky, Moshe. 1994. “Changes in the U.S. Wage Structure 1963-1987: Application
of Quantile Regression.” Econometrica 62: 405-458.

Calvert, Randall L., Mathew D. McCubbins, and Barry R. Weingast. 1989. “A Theory of
Political Control and Agency Discretion.” American Journal of Political Science
33: 588-611.

Cameron, Charles. 2000. Veto Bargaining: Presidents and the Politics of Negative
Power. New York: Cambridge University Press.

238

Carpenter, Daniel P. 1996. “Adaptive Signal Processing, Hierarchy, and Budgetary
Control in Federal Regulation.” American Political Science Review 90: 283-302.

Carpenter, Daniel P. 2001. The Forging of Bureaucratic Autonomy: Reputations,
Networks, and Policy Innovation in Executive Agencies, 1862-1928. Princeton:
Princeton University Press.

Carpenter, Daniel P. 2002. “Groups, the Media, Agency Waiting Costs, and FDA Drug
Approval.” American Journal of Political Science 46: 490-505.

Chiou, Fang-Yi, and Lawrence S. Rothenberg. 2003. “When Pivotal Politics Meets
Partisan Politics.” American Journal of Political Science 47: 503-522.

Chubb, John E. 1985. “The Political Economy of Federalism.” American Political
Science Review 79: 944-1015.

Clinton, Bill. 1995. “The New OSHA: Reinventing Worker Safety and Health.”
http://wwwoshagov.

Code of Federal Regulations (CFR), Title 29, Parts 1902 and 1952-1956.

Cohen, Michael D., James G. March, and Johan P. Olson. 1972. “A Garbage Can Model
of Organizational Choice.” Administrative Science Quarterly 17:1-25.

Congress and the Nation, Vol. II. 1965-1968. Washington, DC: Congressional
Quarterly Service.

Congress and the Nation, Vol. 111. 1969-1972. Washington, DC: Congressional
Quarterly Service.

Congress and the Nation, Vol. IV. 1973-1976. Washington, DC: Congressional
Quarterly Service.

Congress and the Nation, Vol. IX. 1993-1996. Washington, DC: Congressional
Quarterly Service.

Congress and the Nation, Vol. V. 1977-1980. Washington, DC: Congressional
Quarterly Service. .

Congress and the Nation, Vol. VII. 1985-1988. Washington, DC: Congressional
Quarterly Service.

Congress and the Nation, Vol. VIII. 1989-1992. Washington, DC: Congressional
Quarterly Service.

Cox, Gary W., and Mathew D. McCubbins. 1993. Legislative Leviathan: Party
Government in the House. Berkeley: University of California Press.

CQ Almanac. 1968-2001. Washington, DC: Congressional Quarterly Service.

239

Davis, Otto A., Morris H. DeGroot, and Melvin J. Hinich. 1972. “Social Preference
Orderings and Majority Rule.” Econometrica 40: 147-157.

Dodd, Lawrence C., and Richard L. Schott. 1986. Congress and the Administrative State.
New York: Macmillan.

Downs, Anthony. 1967. Inside Bureaucracy. Boston: Little, Brown.

Dunleavy, Patrick. 1991. Democracy, Bureaucracy and Public Choice: Economic
Explanations in Political Science. New York: Prentice Hall.

Edwards, George C ., III, and Stephen J. Wayne. 1999. Presidential Leadership: Politics
and Policy Making. New York: Worth Publishers, Inc.

Eisner, Marc Allen, and Kenneth J. Meier. 1990. “Presidential Control versus
Bureaucratic Power: Explaining the Reagan Revolution in Antitrust.” American
Journal of Political Science 34: 269-287.

Eisner, Marc Allen, Jeff Worsham, and Evan J. Ringquist. 2000. Contemporary
Regulatory Policy. Boulder: Lynne Rienner Publishers, Inc.

Eisner, Marc Allen. 2000. Regulatory Politics in Transition. Baltimore: Johns Hopkins
University Press.

Epstein, David, and Sharyn O’Halloran. 1996. “Divided Government and the Design of
Administrative Procedures: A F ormal Model and Empirical Test.” Journal of
Politics 58: 373-397.

Epstein, David, and Sharyn O’Halloran. 1999. Delegating Powers: A Transaction Cost
Politics Approach to Policy Making Under Separate Powers. New York:
Cambridge University Press.

Fenno, Richard, Jr. 1973. Congressmen in Committees. Boston: Little, Brown.

Fenno, Richard, Jr. 1978. Home Style: House Members and Their Districts. Boston:
Little, Brown.

Fiorina, Morris P. 1982. “Legislative Choice of Regulatory Forms: Legal Process or
Administrative Process?” Public Choice 39: 33-66.

Fiorina, Morris P. 1989. Congress: Keystone of the Washington Establishment. New
Haven: Yale University Press.

Franklin, Charles H. 1991. “Eschewing Obfuscation? Campaigns and the Perception of
Senate Incumbents.” American Political Science Review 85: 1193-1214.

Friedman, James W. 1990. Game Theory with Applications to Economics. New York:
Oxford University Press.

240

Gerth, H.H, and C. Wright Mills. 1946. From Max Weber: Essays in Sociology. New
York: Oxford University Press.

Gillian, Thomas W., and Keith Krehbiel. 1989. “Asymmetric Information and
Legislative Rules with a Heterogeneous Committee.” American Journal of
Political Science 33: 459-490.

Gormley, William T. 1986. “Regulatory Issue Networks in a Federal System.” Polity 18:
595-620.

Gormley, William T. 1997. “Regulatory Enforcement: Accommodation and Conflict in
Four States.” Public Administration Review 57: 285-293.

Gray, Virginia, Russell L. Hanson, and Herbert Jacob. 1999. Politics in the American
States: A Comparative Analysis. Washington DC: CQ Press.

Greene, William H. 2000. Econometric Analysis. Upper Saddle River, N.J.: Prentice-
Hall, Inc.

Grodzins, Morton. 1960. “The Federal System.” In Report of the President’s
Commission on National Goals: Goals for Americans. Englewood Cliffs:
Prentice-Hall.

Gulick, Luther. 1937. “Notes on the Theory of Organization.” In Papers on the Science
of Administration. eds. Luther Gulick and Lyndall Urwick. New York: Columbia
University Press.

Hall, Richard L. 1993. Participation in Congress. New Haven: Yale University Press.

Hammond, Thomas H. 1986. “Agenda Control, Organizational Structure, and
Bureaucratic Politics.” American Journal of Political Science 30: 380-420.

Hammond, Thomas H. 1998. “The Epistemology of Power and Autonomy in the
Policyrnaking Process: If An Agency 13 — Or is not — Autonomous, How Would
We Know?” Presented at the Annual Meeting of the Public Choice Society, New

Orleans, Louisiana.

Hammond, Thomas H. 2003. “Veto Points, Policy Preferences, and Bureaucratic
Autonomy in Democratic Systems.” In Politics, Policy, and Organizations:
Frontiers in the Scientific Study of Bureaucracy. eds. George A. Krause and
Kenneth J. Meier. Ann Arbor: University of Michigan Press.

Hammond, Thomas H., and Chris K. Butler. 2003. “Some Complex Answers to the
Simple Question ‘Do Institutions Matter?’ - Policy Choice and Policy Change in
Presidential and Parliamentary Systems.” Journal of Theoretical Politics 15: 145-
200.

241

Hammond, Thomas H., and Gary J. Miller. 1985. “A Social Choice Perspective on
Expertise and Authority in Bureaucracy.” American Journal of Political Science
29: 1-28.

Hammond, Thomas H., and Gary J. Miller. 1987. “The Core of the Constitution.”
American Political Science Review 81: 1155-1174.

Hammond, Thomas H., and Jack H. Knott. 1996. “Who Controls the Bureaucracy?:
Presidential Power, Congressional Dominance, Legal Constraints, and
Bureaucratic Autonomy in a Model of Multi-Institutional Policy-Making.”
Journal of Law, Economics, & Organization 12: 119-166.

Hammond, Thomas H., and Jack H. Knott. 1999. “Political Institutions, Public
Management, and Policy Choice.” Journal of Public Administrative Research
and Theory 9: 33-85.

Hammond, Thomas H., and Jeffrey S. Hill. 1993. “Deference or Preference? Explaining
Senate Confirmation of Presidential Nominees to Administrative Agencies.”
Journal of Theoretical Politics 5: 23-59.

Headrick, Barbara, George Serra, and Jim Twombly. 2002. “Enforcement and Oversight:
Using Congressional Oversight to Shape OSHA Bureaucratic Behavior.”
American Politics Research 30: 608-629.

Heimann, C. F. Larry. 1992. “Understanding the Challenger Disaster: Organizational
Structure and the Design of Reliable Systems.” American Political Science
Review 87: 421-435.

Horn, Murrary J. 1995. The Political Economy of Public Administration: Institutional
Choice in the Public Sector. New York: Cambridge University Press.

Howell, William G. 2003. Power without Persuasion: The Politics of Direct Presidential
Action. Princeton: Princeton University Press.

Huber, John D., and Charles R. Shipan. 2002. Deliberate Discretion? The Institutional
Foundations of Bureaucratic Autonomy. New York: Cambridge University Press.

Huntington, Samuel P. 1952. “The Marasmus of the ICC.” Yale Law Journal 61: 467-
509.

Jacobs, Gary C. 2001. The Politics of Congressional Elections. New York: Longrnan.

Jacoby, William G. 1988. “The Impact of Party Identification on Issue Attitudes.”
American Journal of Political Science 32: 643—661.

Kaufman, Herbert. 1956. “Emerging Conflicts in the Doctrines of Public
Administration.” American Political Science Review 50: 1057-1073.

242

 

Kaufman, Herbert. 1981. The Administrative Behavior of Federal Bureau Chiefs.
Washington, DC: The Brookings Institution.

Keefe, William, and Morris Ogul. 1989. The American Legislative Process: Congress
and the States. Englewood Cliffs: Prentice Hall.

Keiser, Lael R., and Joe Soss. 1998. “With Good Cause: Bureaucratic Discretion and the
Politics of Child Support Enforcement.” American Journal of Political Science
42: 1133-1156.

Kelman, Steven. 1980. “Occupational Safety and Health Administration.” In The Politics
of Regulation. ed. James Q. Wilson. New York: Basic Books.

Kelman, Steven. 1981. Regulating America, Regulating Sweden: A Comparative Study of
Occupational Safety and Health Policy. Cambridge: MIT Press.

Kemell, Samuel. 1997. Going Public: New Strategies of Presidential Leadership.
Washington DC: CQ Press.

King, Gary. 1998. Unifying Political Methodology: The Likelihood Theory of Statistical
Inference. Ann Arbor: University of Michigan Press.

Kingdon, John W. 1995. Agendas, Alternatives, and Public Choices. New York:
Longrnan.

Knott, Jack H., and Thomas H. Hammond. 2000. “Congressional Committees and Policy
Change: Explaining Legislative Outcomes in Banking, Trucking, Airline, and
Telecommunications Deregulation.” In Governance and Performance: New
Perspectives. eds. Carolyn J. Heinrich and Laurence E. Lynn, Jr. Washington
DC: Georgetown University Press.

Koenker, Roger, and Gibert Bassett, Jr. 1978. “Regression Quantiles.” Econometrica 46:
33-50.

Koenker, Roger, and Kevin F. Hallock. 2001. “Quantile Regression.” Journal of
Economic Perspectives 15: 143-156.

Krause, George A. 1999. A Two- Way Street: The Institutional Dynamics of the Modern
Administrative State. Pittsburgh: University of Pittsburgh Press.

Krehbiel, Keith. 1991. Information and Legislative Organization. Ann Arbor: University
of Michigan Press.

Krehbiel, Keith. 1998. Pivotal Politics: A Theory of U.S. Lawmaking. Chicago:
University of Chicago Press.

Landau, Martin. 1969. “Redundancy, Rationality, and the Problem of Duplication and
Overlap.” Public Administration Review 29: 346-358.

243

Lewis, David E. 2003. Presidents and the Politics of Agency Design: Political Insulation
in the United States Government Bureaucracy, 1946—1 99 7. Stanford: Stanford
University Press.

Lindblom, Charles E. 1959. “The Science of ‘Muddling Through.”’ Public
Administration Review 19: 79-88.

Lipsky, Michael. 1980. Street-Level Bureaucracy: Dilemmas of the Individual in Public
Service. Beverly Hills: Sage Publications.

Long, Norton E. 1949. “Power and Administration” Public Administration Review 2:
25 7-264.

Maass, Arthur. 1983. Congress and the Common Good. New York: Basic Books.

Maltzman, Forrest. 1997. Competing Principals: Committees, Parties, and the
Organization of Congress. Ann Arbor: University of Michigan Press.

March, J ames G., and Herbert A. Simon. 1958. Organizations. New York: Free Press.

Mark S. Hurwitz, Roger J. Moiles, and David W. Rohde. 2001. “Distributive and
Partisan Issues in Agricultural Policy in the 104th House.” American Political
Science Review 95: 911-922.

Marvel, Mary K. 1982. “Implementation and Safety Regulation: Variations in Federal
and State Administration under the OSHA.” Administration and Society 14: 15-
33.

Mayhew, David. 1974. Congress: The Electoral Connection. New Haven: Yale
University Press.

Mayhew, David. 1991. Divided We Govern: Party Control, Lawmaking, and
Investigations, 1946-1990. New Haven: Yale University Press.

McCaffrey, David P. 1982. OSHA and the Politics of Health Regulation. New York:
Plenum Press.

McCubbins, Mathew D. 1985. “The Legislative Design of Regulatory Structure.”
American Journal of Political Science 29: 721-748.

McCubbins, Mathew D., and Thomas Schwartz. 1984. “Congressional Oversight
Overlooked: Police Patrols vs. Fire Alarms.” American Journal of Political
Science 28: 165-179.

McCubbins, Mathew D., Roger G. N011, and Barry R. Weingast. 1987. “Administrative
Procedures as Instruments of Political Control.” Journal of Law, Economics, and
Organization 3: 243-277.

244

McCubbins, Mathew D., Roger G. Noll, and Barry R. Weingast. 1989. “Structure and
Process, Politics and Policy: Administrative Arrangements and the Political
Control of Agencies.” Virginia Law Review 75: 431-482.

McCubbins, Matthew D., and Thomas Schwartz. 1984. “Congressional Oversight
Overlooked: Police Patrols Versus Fire Alarms.” American Journal of Political
Science 28: 165-179.

Meier, Kenneth J. 1993. Politics and the Bureaucracy: Policymaking in the Fourth
Branch of Government. Belmont: Wadsworth.

Mendeloff, John. 1979. Regulating Safety: An Economic and Political Analysis of
Occupational Safety and Health Policy. Cambridge: MIT Press.

Miller, Gary J. 1992. Managerial Dilemmas: The Political Economy of Hierarchy. New
York: Cambridge University Press.

Miller, Gary J ., and Terry M. Moe. 1983. “Bureaucrats, Legislators, and the Size of
Government.” American Political Science Review 77: 297-322.

Miller, Gary J., and Thomas H. Hammond. 1990. “Committees and the Core of the
Constitution.” Public Choice 66: 201-227.

Miller, Gary J., Thomas H. Hammond, and Charles Kile. 1996. “Bicarneralism and the
Core: An Experimental Test.” Legislative Studies Quarterly 21: 83-103.

Miqué, Jean-Luc, and Gerard Bélanger. 1974. “Toward a General Theory of Managerial
Discretion.” Public Choice 17: 27-47.

Moe, Terry M. 1982. “Regulatory Performance and Presidential Administration.”
American Journal of Political Science 26: 197-224.

Moe, Terry M. 1984. “The New Economics of Organization.” American Journal of
Political Science 28: 739-777.

Moe, Terry M. 1985. “Control and Feedback in Economic Regulation: The Case of the
NLRB.” American Political Science Review 79: 1094-1116.

Moe, Terry M. 1987. “An Assessment of the Positive Theory of Congressional
Dominance.” Legislative Studies Quarterly 12: 475-520.

Moe, Terry M. 1990. “Political Institutions: The Neglected Side of the Story.” Journal of
Law, Economics, and Organization 6: 213-253.

Moe, Terry M., and William G. Howell. 1999. “The Presidential Power of Unilateral
Action.” Journal of Law, Economics, and Organization 15: 132-179.

245

 

 

Needham, Douglas. 1983. The Economics and Politics of Regulation: A Behavioral
Approach. Boston: Little, Brown.

Nice, David, and Patricia Fredericksen. 1995. The Politics of Intergovernmental
Relations. Chicago: Nelson-Hall.

Niskanen, William A. 1971. Bureaucracy and Representative Government. Chicago:
Aldine-Atherton.

Niskanen, William A. 1975. “Bureaucrats and Politicians.” Journal of Law and
Economics 18: 617-644.

Occupational Safety and Health Act of 1970, Public Law No. 91-596.

Office of Technology Assessment (OTS). 1985. Preventing Illness and Injury in the
Workplace. Washington, DC: U.S. Government Printing Office.

Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of
Groups. Cambridge: Harvard University Press.

Olson, Mary. 1999. “Agency Rulemaking, Political Influences, Regulation and Industry
Compliance.” Journal of Law, Economics and Organization 15:573-601.

OSHA Field Inspection Reference Manual, CPL 2.103.

Padgett, John. 1981. “Hierarchy and Ecological Control in Federal Budgetary Decision
Making.” American Journal of Sociology 87: 75-129.

Paolino, Philip. 2001. “Maximum Likelihood Estimation of Models with Beta
Distributed Dependent Variables.” Political Analysis 9: 325-346.

Peltzman, Sam. 1976. “Toward a More General Theory of Regulation.” Journal of Law
and Economics 19: 21 1-240.

Peterson, Paul E. 1995. The Price of Federalism. Washington, DC: The Brookings
Institution.

Poole, Keith T. 1998. “Recovering a Basic Space from a Set of Issue Scales.” American
Journal of Political Science 42: 964-993.

Poole, Keith T., and Howard Rosenthal. 1997. Congress: A Political-Economic History
of Roll Call Voting. New York: Oxford University Press.

Redford, Emmette S. 1969. Democracy in the Administrative State. New York: Oxford
University Press.

Rohde, David W. 1991. Parties and Leaders in the Postreform House. Chicago:
University of Chicago Press.

246

 

Rothenberg, Lawrence S. 1994. Regulation, Organizations, and Politics: Motor Freight
Policy at the Interstate Commerce Commission. Ann Arbor: University of
Michigan Press.

Rourke, Francis E. 1984. Bureaucracy, Politics, and Public Policy. Boston: Little,
Brown.

Rourke, Francis E. 1986. Bureaucratic Power in National Policy Making. Boston: Little,
Brown.

Sabatier, Paul A., and Hank C. Jenkins-Smith. 1993. Policy Change and Learning: An
Advocacy Coalition Approach. Boulder: Westview Press.

Sabatier, Paul A., John Loomis, and Catherine McCarthy. 1995. “Hierarchical Controls,
Professional Norms, Local Constituencies, and Budget Maximization: An

Analysis of U.S. Forest Service Planning Decisions.” American Journal of
Political Science 39: 204-242.

Scholz, John T., and F eng Heng Wei. 1986. “Regulatory Enforcement in a Federalist
System.” American Political Science Review 80: 1249-1270.

Scholz, John T., and Wayne B. Gray. 1997. “Can Government Facilitate Cooperation?
An Informational Model of OSHA Enforcement.” American Journal of Political
Science 41: 693-717.

Scholz, John T., Jim Twombly, and Barbara Headrick. 1991. “Street-Level Political
Controls Over Federal Bureaucracy.” American Political Science Review 85:
829-850.

Selznick, Philip. 1949. TVA and the Grass Roots. Berkeley: University of California
Press.

Shepsle, Kenneth A., and Barry R. Weingast. 1987. “The Institutional Foundations of
Committee Power.” American Political Science Review 81: 85-104.

Shipan, Charles R. 2004. “Regulatory Regimes, Agency Actions, and the Conditional
Nature of Congressional Influence.” American Political Science Review 98: 467-
480.

Simon, Herbert A. 1946. “The Proverbs of Administration.” Public Administration
Review 6: 53-67.

Simon, Herbert A. 1947. Administrative Behavior. New York: Macmillan.

Spitzer, Robert J. 1993. President and Congress: Executive Hegemony at the Crossroads
of American Government. New York: McGraw-Hill, Inc.

247

 

 

Stigler, George J. 1971. “The Theory of Economic Regulation.” Bell Journal of
Economics and Management Science 2: 3-21.

Stimson, James A. 1991. Public Opinion in America: Moods, Cycles, and Swings.
Boulder: Westview Press.

Thompson, Frank J., and Michael J. Scicchitano. 1985. “State hnplementation Effort and
Federal Regulatory Policy: The Case of Occupational Safety and Health.”
Journal of Politics 85: 686-703.

Tsebelis, George. 1995. “Decision Making in Political Systems: Veto Players in
Presidentialism, Parliamentarism, Multicameralism and Multipartysm.” British
Journal of Political Science 25: 289-325.

Tsebelis, George. 1999. “Veto Players and Law Production in Parliamentary
Democracies: An Empirical Analysis.” American Political Science Review 93:
591-608.

Tsebelis, George. 2002. Veto Players: How Political Institutions Work. Princeton:
Princeton University Press.

U.S. House, Committee on Small Business. 1997. “OSHA’s Contemplated Safety and
Health Program Standard.” Hearing. The 105th Congress, 1St Session.
Washington, DC: U.S. Government Printing Office.

U.S. House, Subcommittee on Workforce Protections of the Committee on Education
and the Workforce. 2002. “The OSHA Rulemaking Process.” Hearing. The 107th
Congress, 1St Session. Washington, DC: U.S. Government Printing Office.

U.S. House, Subcommittee on Workforce Protections of the Committee on Education
and the Workforce. 2002. “Reaching a Consensus to Update OSHA’s Permissible
Exposure Levels.” Hearing. The 107th Congress, 2"d Session. Washington, DC:
U.S. Government Printing Office.

U.S. Senate, Subcommittee on Employment, Safety, and Training of the Committee on
Health, Education, Labor, and Pensions. 2002. “Workplace Safety and Health:
Oversight of MSHA and OSHA Regulation and Enforcement.” Hearing. 107th
Congress, 2nd Session. Washington, DC: U.S. Government Printing Office.

Volden, Craig. 2002a. “Delegating Power to Bureaucracies: Evidence from the States.”
Journal of Law, Economics, and Organization 18: 187-220.

Volden, Craig. 2002b. “A Formal Model of the Politics of Delegation in a Separation of
Powers System.” American Journal of Political Science 46: 111-133.

Waldo, Dwight. 1984. The Administrative State: A Study of the Political Theory of
American Public Administration. New York: Ronald Press.

248

_ F 213' I?" Elf-lie, '—

 

 

 

Weingast, Barry R. 1984. “The Congressional Bureaucratic System: A Principal-Agent
Prespective (with Applications to the SEC).” Public Choice 44: 147-191.

Weingast, Barry R., and Mark J. Moran. 1983. “Bureaucratic Discretion or
Congressional Control: Regulatory Policymaking by the Federal Trade
Commission.” Journal of Political Economy 91: 756-780.

Weingast, Barry, and William Marshall. 1988. “The Industrial Organization of
Congress.” Journal of Political Economy 96: 132-163.

West, William F. 1995. Controlling the Bureaucracy: Institutional Constraints in Theory
and Practice. Annonk: ME. Sharpe.

West, William F., and Joseph Cooper. 1989. “Legislative Influence Versus Presidential
Dominance: Competing Models of Political Control.” Political Science
Quarterly 104: 581-606.

Whitford, Andrew B. 2005. “The Pursuit of Political Control by Multiple Principals.”
Journal of Politics 67: 29-49.

Wilson, James Q. 1980. The Politics of Regulation. New York: Basic Books.

Wilson, James Q. 1989. Bureaucracy: What Government Agencies Do and Why They Do
It. New York: Basic Books.

Wilson, Woodrow. 1887. “The Study of Administration.” Political Science Quarterly 2:
197-222.

Wood, B. Dan, and Richard W. Waterman. 1994. Bureaucratic Dynamics: The Role of
Bureaucracy in a Democracy. Boulder: Westview Press.

Wood, B. Dan. 1988. “Principals, Bureaucrats, and Responsiveness in Clean Air
Enforcement.” American Political Science Review 82: 213-234.

Wood, B. Dan. 1991. “Federalism “and Policy Responsiveness: The Clean Air Case.”
Journal of Politics 53: 851-859.

Wood, B. Dan. 1992. “Modeling Federal Implementation as a System: The Clean Air
Case.” American Journal of Political Science 36: 40-67.

Wooldridge, Jeffrey M. 2000. Introductory Econometrics: A Modern Approach.
Cincinnati: South-Westem.

Wooldridge, Jeffi'ey M. 2002. Econometric Analysis of Cross Section and Panel Data.
Cambridge: MIT Press.

249