University of Southern California Dissertations and Theses

Policy termination: a conceptual framework and application to the local public hospital context

(USC Thesis Other)

Policy termination: a conceptual framework and application to the local public hospital context

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

POLICY TERMINATION: A CONCEPTUAL FRAMEWORK AND
APPLICATION TO THE LOCAL PUBLIC HOSPITAL CONTEXT

by

Ke Ye

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PUBLIC ADMINISTRATION)

May 2007

Copyright 2007 Ke Ye
ii
Acknowledgements
This dissertation could not have been completed without the tremendous support and
encouragement of many people.

I am deeply indebted to my dissertation chair, Dr. Graddy Elizabeth, for her
continuous guidance and strong support throughout the development of this project
and my academic career at USC. I am also grateful to the other members of my
dissertation committee -- Dr. Glenn Melnick and Dr. Jeffrey Sellers, for their
invaluable comments on earlier drafts.

I would like to warmly thank many good friends and colleagues who gave me
immeasurable encouragement and help during the dissertation writing process: Chao
Guo, Chingheng Pan, Qing Liu, Hsin-ling Hsieh, Echu Liu, Erlyana Erlyana, Yuhua
Bao and Bin Chen for generously giving their time to provide me with insightful
feedback; Arleen Freeman for her careful work in polishing my dissertation; Dr.
Joanna Yu, Dr. Joyce Mann, Dr. Melissa Lopez, Ann Abrahamyan and other IPPAM
faculty and staff, who have given me many years of collegial friendship and support,
both academically and spiritually; and to the many IPPAM students who I’ve had the
pleasure to teach and work with; they never ceased to inspire me with their questions
and ideas.

I also want to thank SPPD for providing me with generous financial support, without
which I could not have been able to finish my academic work at USC. In the past
iii
years, many SPPD professors and staff members have given me much help, and I
would like to thank them all.

Finally, I owe a special debt to my parents, Qiuqiong Ye and Weier Zheng, and other
family members for their kind understanding and wholehearted support of my Ph.D.
education in the US, and to whom this dissertation is dedicated.
iv
Table of Contents

Acknowledgments

ii
List of Tables

vii
List of Figures

viii
Abstract

ix
Chapter One: Introduction

1
1.1. Government Reform Efforts

2
1.2. Policy Termination: A wrongly Under-attended Topic

6
1.3. Organization of the Dissertation

8
1.4. Significance of the Study

9
Chapter Two: Theoretical Development

12
2.1. Policy Termination Literature

12
2.2. Model Development

18
Triggering Factors

19
Fiscal Problems

20
Perceived Policy Failures

21
Ideological Change

22
Decision-Making Context Factors

22
Government Structure

25
Policy Characteristics

25
Interest Groups

27
Community Characteristics

29
Summary

32
Chapter Three: Application to the Termination of Public
Hospitals

34
3.1. Termination of Public Hospitals

34
3.2. The Role of American Public Hospitals

38
3.3. The California Public Hospital System

40
3.4. Termination Hypotheses

43
Triggering Factors

44
Fiscal Problems

44
1) Federal Fiscal Conditions

45
v
2) State Fiscal Conditions

47
3) Local Fiscal Conditions

48
Perceived Failures

49
Ideological Change

49
Decision-Making Context Factors

50
Government Structure

50
Policy Characteristics

51
Interest Groups

53
Community Characteristics

54
Chapter Four: Empirical Testing with a Binary Procedure

58
4.1. Data, Variables and Measures

58
Data Source

59
Dependent Variable

60
Independent Variables

60
Triggering Events

61
Decision-Making Context Factors

63
4.2 Analysis Method

68
4.3 Results and Findings

70
Descriptive Results

70
Multivariate Analysis and Discussion

73
Summary
81
Chapter Five: Distinguishing the Termination Form --
Hospital Conversion and Closure

83
5.1. Hospital Conversion versus Closure

83
5.2. Hypotheses

85
5.3. Empirical Testing

91
Methodology

91
Results and Findings

92
Conversion versus Open Decision

95
Closure versus Open Decision

96
Conversion versus Closure Decision

100
Chapter Six: Conclusion

104
6.1. Summary

104
6.2. Contribution

106
6.3. Implications

109
6.4. Limitations

112
vi
Bibliography

117
Appendix

126

vii
List of Tables
Table 1: Variable Definitions and Sources

67
Table 2: Descriptive Statistics by Dependent Variable Value

72
Table 3: GEE Analysis of the Decision to Terminate Public Hospitals

75
Table 4: Multinomial Logit with GEE Approach on the Decision to
Terminate Public Hospitals (Total Profit Margin)

93
Table 5: Multinomial Logit with GEE Approach on the Decision to
Terminate Public Hospitals (Operating Profit Margin)

93

viii
List of Figures
Figure 1: A Decision-Making Model for Policy Termination

33
Figure 2: The Termination Trend of Pubic Hospitals (comparing US and
CA)

36
Figure 3: A Decision-Making Model for Termination for Public
Hospitals

57
Figure 4: Trends in the Number of California Public Hospitals, 1981-
1995

71
Figure 5: Factors Distinguishing Hospital Closure and Conversion

90

ix
Abstract

The improvement of government performance requires that government carry out
periodical reviews of its policies and terminate those that are inefficient and
ineffective. The existing literature nevertheless reveals that the termination process
has been severely understudied. The paucity of knowledge has left many important
questions unanswered, thus limiting our capability to evaluate and guide government
reforms. The consequence is most acutely perceived at a time when governments are
struggling to do more with less and more terminations are expected to take place.

The main purpose of this study is therefore to develop an analytical framework of the
policy termination process by integrating theories from the policy termination,
decision-making and organization theory literature. With an emphasis on local
government, this study develops a two-stage model of the termination process: the
triggering factors, which include fiscal problems, perceived policy failures and
ideological change, and the local decision-making context, which encompasses
government structure, policy characteristics, the influence of interest groups and
community characteristics. The model is then empirically tested using binary and
multinomial procedures with data on California public hospitals over the period from
1981 to 1995. The results of the binary procedure show that state and local fiscal
conditions, the local private service market, the organization of interest groups, the
size of the beneficiary group, the local commitment to public service delivery, and
preference homogeneity are important determinants of policy termination. The
x
multinomial procedure reaches similar conclusions, but further reveals that different
termination forms are associated with different sets of factors. More specifically, the
conversion decision is affected by state and local fiscal conditions, the organization of
anti-termination interest groups, the private service market and local commitment to
the target services. The closure decision, however, is affected by local fiscal
conditions and the target program’s performance. Furthermore, the multinomial
results suggest that some factors distinguish the choice of conversion versus closure
form. Better operating efficiency, a less developed private hospital service market and
a weaker community commitment favor a conversion rather than a closure decision.
These findings highlight the importance of the termination form when examining
government termination decisions.

1
Chapter One: Introduction

Governments in recent years have been trying hard to improve performance, which
requires periodical reviews of government policies and termination of those that are
inefficient and ineffective. Yet compared to other policy processes such as agenda
setting, policy formation, policy implementation and policy evaluation, the process of
policy termination may be the least understood. The limited amount of research and
inconsistent findings available at the current stage pose great difficulty for us to
realistically evaluate government reform efforts and shed little light on guiding how
the government should carry out termination and harvest the expected benefits of a
more efficient, effective and responsive government. More importantly, integrative
frameworks for policy termination are almost completely lacking, further adding to the
difficulty in advancing our understanding of the termination process.

The purpose of this study is therefore to develop an analytical framework for the
termination process by integrating theories from different disciplines and testing the
explanatory power of the model. More specifically, a two-stage model of termination
determinants is proposed to include triggering events and the decision-making context.
An empirical test of the model is then conducted with California public hospitals data
over the 1981-1995 period.

In this chapter, the first section presents a brief introduction to the trend of governance
reform and the challenges that governments face. The second part evaluates the
2
current situation of termination research and identifies related research questions.
Finally, the organization of the dissertation and the significance of the study are
discussed.

1.1. Governance Reform Efforts:

In recent years, increasing attention has been focused on reforming the existing
governmental policymaking structure (Kettl, 2002; Stoker, 1998). Critics argued that,
compared to the private sector, government has distinctive characteristics that prevent
it from functioning efficiently and responsively (Weimer and Vining, 1991; Wolf,
1979). For example, government agencies operate in environments that lack
competition, have conflicting objectives, or an inflexible civil service system, just to
name a few. Some scholars even contend that the US government is designed to be
inefficient (Wilson, 1989; Moe&Caldwell, 1994). Much of the existing empirical
research supports that government production costs more than that in private and
nonprofit sectors (Savas, 1987; Ferris, 1986; Brooks, 2004). Therefore, it is argued
that in order to improve government performance, governments should conduct
governance reform and shift the focal point of service delivery to the private and
nonprofit sectors.

The new governance mode challenges the concepts of traditional public administration
in fundamental ways. Traditionally, government is both the rule maker and the service
provider. The new governance mode, however, seeks to fundamentally redefine the
3
role of government and its relationship with society. It argues that government should
do more steering rather than rowing (Osborne&Gaebler, 1992); government’s primary
role is not public service delivery, but rather coordination of the actors in different
sectors to jointly provide services in a more efficient and responsive manner.
Therefore, government is proposed to reduce its share of service provision, transfer
service delivery to the private sector, and engage in active collaboration with private
and the nonprofit organizations (Gaudin, 1998). Correspondingly, the existing
governance structure should be replaced with alternative ones that involve more of the
private sector. In particular, government is urged to commit to the practice of
privatizing, contracting out and establishing partnerships with both the for-profit and
nonprofit sectors (Krahmann, 2003).

The ideas of governance reform have been well embraced by both the Republican and
Democratic parties in the United States. Vice President Gore has sought to evaluate
agency performance on a nationwide scale and downsize the government accordingly
(Gore, 1993), while Congressional republicans passed the “Contract With America”
Plan, which proposed to eliminate a large number of selected programs. In practice,
governments have entered the action networks with private/nonprofit sector partners
more frequently; partnering or contracting out with private/nonprofit organizations as
an alternative form to carry out social services has become a common practice (Ferris
and Graddy, 1991).

4
The determination to carry out governance reform and take advantage of the more
efficient production mode in the private sectors is further influenced by two problems
that governments face. First, governments at different levels are plagued by financial
difficulties. In the past two decades the federal government has consistently had a
budget deficit (OMB, 2004). At the state level, 46 out of 50 states in the United States
had budget deficits in 2002 (CLA, 2002). A more dramatic case took place in
California in October 2003 when voters recalled Governor Gray Davis, mainly due to
a fiscal crisis of a more than 10 billion budget deficit. A similar situation is happening
at the local level. For example, Los Angeles County was projected to have a $700
million budget deficit by the end of 2005 (Cousineau, et al., 2003).

Second, while the public demands more services on the one hand, they are unwilling
to pay for them on the other. For example, a recent poll revealed that a majority of
respondents wanted to balance the federal budget and wanted tax cuts (Beneditto,
1995). Such a typical public mentality poses a great challenge and has important
implications for government officials. On the one hand, they have to appeal to the
voters by satisfying their new service demands. On the other hand, they may be in
political trouble if they want to levy more money from the taxpayers to pay for them.

Faced with such difficulties, government is forced to do more with less. What this
implies is that government has to sustain effective programs, eliminate ineffective
ones, and transfer the services it cannot provide efficiently to the private sector.
Indeed, the governance and public management reforms in the past decades carried the
5
expectation that governments regularly evaluate their programs and terminate those
that are ineffective or inefficient.

Though the expectations on governance reform are keen, the commitment is strong,
and more terminations are expected to occur, the results are less clear. There are
arguments that government is entangled in politics and the program cuts do not work
like those in the private sector. For example, most program cuts are inevitably faced
with intensive resistance from the beneficiaries (Bardach, 1976; Behn, 1978).
Furthermore, a termination decision has to overcome the often formidable political
and institutional constraints within the government system (deLeon, 1978). Actually,
one of the biggest warnings on the high profile “reinventing government” movement
is “don’t forget politics” (Kettl, 1997).

According to a study conducted by Kaufman (1976), out of the 421 agencies existing
between 1923 and 1973, only 27, or 15%, had been terminated. The mortality rate is
less than half that of private organizations calculated in the same manner. What the
result suggests is that public agencies tend to live long after they no longer serve their
original function. This finding, if true, can have important implications for
governance reform. To a large degree, how successfully government can reap the
benefits of reform depends heavily on how it goes through the termination process. If
the termination is so difficult to carry out in the public sector, then many outdated
agencies and programs may not be able to be eliminated in a timely manner, and the
benefits of governance reform may have been overestimated. Is the termination of
6
government agencies really so difficult? If so, what are the factors that make it
difficult? Why do some organizations survive the termination efforts while others do
not? What are the determinants of the termination process? Do efficiency and
effectiveness criteria play a role in the termination process? With many program cuts
planned and possibly implemented in the near future, it is of critical importance that
we understand the termination process well.

1.2. Policy Termination: A wrongly Under-attended Topic

Unfortunately, despite the fact that policy termination, a generic term for the
termination of government policies, functions, programs and organizations, is a vital
component of the policy process, it has been long under-studied. Since 1976, there
have been only three symposiums on policy termination. The number of publications
on this topic is small and the writers amount to only a handful. As observed by
Daniels (2001), policy termination continues to be under-attended, a term first applied
to the study of this field three decades ago (Biller, 1976).

The under-attention is said to be the result of the following reasons (Daniels, 2001;
Bardach, 1976; deLeon, 1978): First, there are relatively few examples of termination
available for study, which makes the phenomenon seemingly unimportant and the
generalizations difficult. Second, researchers tend to focus on new and innovative
policies, rather than outdated, flawed, or ineffective polices. Third, there are
challenges in the operationalization of the concept of termination, such as whether to
7
categorize policy events as termination or partial termination or just adjustment.
Fourth, policy termination is often the political expediency that attracts more attention
during political campaigns and economic downturn but gets ignored when the
situation has improved. Fifth, the lack of communication between different disciplines
inhibits the proliferation of policy termination research.

The lack of scholarly attention on policy termination has severely limited our
understanding of the termination process. The current literature has left many
important questions unanswered or answered with conflicting results. For example, is
policy termination a premeditated process with careful planning, or it is just a random
event? Are there any regular patterns of policy termination? If so, what are the
factors that lead to such patterns? Is there any way we can tell which factor is more
important than another under certain circumstances? What kind of termination
strategy should we employ under different conditions?

The lack of knowledge on termination is especially troubling at a time when
governments are concurrently facing increasing financial constraints and increasing
public service expectations, and more program cuts are expected to take place. For
instance, if government wants to cut a program, what kind of program will it cut? Is it
really the inefficient one that will be cut? Will the targeted program be cut smoothly?
When government wants to increase efficiency, will the inefficient programs be
terminated in a timely manner? For government decision-making in such a highly
politicalized arena, will the political considerations dominate the economic and
8
efficiency ones? The answers to these questions will fundamentally affect what
benefits government reform can bring about, how we evaluate the reform efforts and
the strategies we should adopt to improve government performance. For government
efforts to result in a better effect, it is a great necessity that we accumulate better
knowledge on this subject.

This dissertation seeks to advance our understanding of the termination decision by
developing a two-stage termination decision model based on termination literature,
decision-making theory and organization theory, and empirically test it with 1981-95
data on California public hospitals.

1.3. Organization of the Dissertation:

The study is organized into six chapters. The next chapter reviews the related
literature on termination phenomenon with a focus on the triggering factors and
contextual factors during the termination decision process. A two-stage theoretical
model is developed and corresponding propositions are generated.

The third chapter applies the theoretical model developed in chapter 2 to the public
hospital context, with an emphasis on the state of California. Testable hypotheses are
developed for the termination decision of local public hospitals.

9
The fourth chapter discusses the operationalization of the model specified in chapter 3.
The dependent variable, independent variables, and the data sources are identified and
the results of empirical testing with binary procedure presented.

Chapter 5 makes the distinction between two different termination forms, conversion
and closure, and explores the effects of important factors. The results of multinomial
logistical regression are reported.

Chapter 6 concludes the study with a discussion of the implication of the findings, the
limitations of the study and the suggestions for the direction of future research.

1.4. Significance of the Study:

This study attempts to make several important contributions to the existing literature.
First, it is one of the few attempts to develop a process model to explain the policy
termination decision. The model is unique in that it involves two stages, which is
expected to be closer to the reality of the complex nature of termination phenomena.
At the same time, the theoretical framework includes a more comprehensive set of
factors such as governmental structure and community environment that are often
absent in previous research, thus allowing the empirical exploration of a broader range
of influences on the determinants of policy termination decisions, and their relative
importance. The model also integrates the theories from different theoretical
perspectives, such as policy termination, decision-making, and organization and
10
interest group theory, thereby facilitating the communication between different
academic fields and contributing to the proliferation of policy termination research.

Second, given that most existing termination research focuses on the state or federal
level, this article makes a first attempt to develop a theoretical model under local
decision-making settings. Such knowledge of termination is particularly important
because more and more termination decisions are taking place at the local level. The
model and the empirical findings in this study help improve our understanding of local
government’s behavior and the impact of community factors on the termination
decisions.

Third, this study adds to the few quantitative studies that overcome the limitations of
typical case studies and makes the conclusions more generalizable to termination
phenomena in other settings. The multivariate analysis method employed in this study
enables us to better evaluate the relative importance of each factor by including them
in the same model. In addition, a GEE approach is adopted to control the repeated
measure bias that is often found in clustered data analysis. Compared to non GEE
methods, this approach produces more accurate estimates and generates more reliable
conclusions.

Fourth, this study is also the first one to make the distinction between different
termination forms in policy termination research. The analysis of California hospital
data shows that different forms of termination are associated with different sets of
11
determinants, and certain conditions favors one form of termination versus another.
For example, a conversion decision of public hospitals is affected by the state and
local fiscal conditions, the organization of anti-termination interest groups, private
service market and the local commitment to the target services, while a closure
decision is only affected by the local fiscal conditions and the target program’s
performance. Thus, this study not only helps decision makers to think about a change,
but also enables them to foresee the problems associated with specific types of
changes, thus enabling better preparation to overcome the challenges during the
termination process.

Finally, with the particular emphasis being on the termination of public hospitals, the
model developed in this study is also a first attempt to systematically examine hospital
behavior from a public sector perspective. The testing of ideological, governmental
and interest group factors that are often neglected in private decision models helps
advance our understanding of hospital conversion and closure in public settings. The
fact that factors often found important in private hospital conversion or closure were
not found significant in this study illustrates the fundamental difference between
public and private decision-making. Therefore, the conceptual framework developed
from a public perspective in this study contributes to the theoretical integration and
synthesis of health service research and serves as a bridge between the health field and
public administration.

12
Chapter Two: Theoretical Development

This chapter aims at developing a 2-stage decision-making model for the policy
termination process. The first section briefly reviews previous research on policy
termination. The second section develops a new model based on the literature of
policy termination, decision-making theory, and organization theory.

2.1. Policy Termination Literature

Policy termination has been regarded as an integral part of the policy process. In a
policy cycle, a policy process typically includes agenda setting, policy formation,
policy implementation, policy evaluation and finally, termination (May&Wildavsky,
1978). No matter how perfectly a policy was originally designed, it may lose its
validity after it accomplishes its goal, or may meet with other problems due to
environmental changes or implementation difficulties. In other words, policies can
become inefficient, out of date, or dysfunctional. Consequently, periodical review and
termination of government policies becomes an important step to safeguard the quality
of public policy and ensure government efficiency, effectiveness and responsiveness.
Although we tend to treat policy termination as the last step of a policy cycle, policy
termination can be both viewed as an end and a beginning – an end to the existing
policy, and a beginning for the alternative new policy and re-deployment of the
resources.

13
While there is agreement on the importance of the policy process, the term has been
defined in different ways. The major challenges to define the term come from two
sources: First, the termination can mean either a full stop of services as commonly
understood, or a partial termination, such as changes in policy emphasis, jurisdiction
or level of funding (Daniels, 2001), thus making identifying the case and timing of
termination challenging. Second, related to the inherently complex nature of
government decisions, termination can refer to different targets under different
occasions. While some scholars focus on specific policies, others also include the
termination of programs or organizations. For example, deLeon (1978) examined four
major types of termination decisions in government and argued that functions are
presumably the most enduring of public activities, followed by organizations, policies
and finally programs (Daniels, 1997). Following deLeon, the policy termination in
this study is broadly defined as the cessation of specific government functions,
programs, policies, or organizations, for the purpose of analysis convenience, as it is
usually possible to identify cessation. No distinction is made between different types
of termination, because in practice different kinds of terminations are often
intertwined with each other. Moreover, as argued by Bardach (1976), any efforts to
terminate any type of them would generate similar political contests, thus making it
unnecessary to differentiate between them in the context of policy termination.

Most of the existing literature has focused on identifying the patterns and factors of
policy termination. In his seminal work, Kaufman (1976) argued the deaths, or
terminations of government organizations are more likely attributed to chance, or bad
14
luck, than to any other factor, as the incremental government decision-making process
insulates government organizations from a natural life cycle. Although Kaufman
(1985) later speculated that the organization’s death results from the system’s failure
to evolve and adapt to the new environmental conditions, he insisted that the timing of
termination appeared to be random. In contrast to Kaufman’s view, some others
scholars assume that termination is a deliberated, highly political but still rational
policy process, and much effort has been invested in identifying the common patterns
and determinants. DeLeon (1978) developed the first general conceptual framework
for analyzing the factors that inhibit policy termination: 1) intellectual reluctance; 2)
institutional permanence; 3) dynamic conservatism; 4) anti-termination coalitions; 5)
legal obstacles; 6) high start up costs. He argued that due to these constraints,
termination is typically difficult to plan and carry out. The model was found useful in
empirical case studies conducted by other researchers, despite some modifications
(Frantz, 1992; Daniels, 1995).

While deLeon’s model (1978) is useful in explaining why policy termination is a rare
phenomenon, it fails to answer why terminations do take place on many occasions.
The efforts to explore the latter question suggest that factors contributing to the
termination decision fall into three main categories: financial imperatives,
governmental efficiencies and political ideology, with different perspectives
emphasizing the different relative importance of different factors (deLeon, 1983).

15
The financial model emphasizes the role of budget deficits and revenue shortfalls in
the policy termination decision. It argues that since public programs are supported
with a government budget, when there is a budget deficiency or revenue shortfall,
government may be forced to terminate certain programs. Findings from empirical
research have lent support to the role of budget deficits (Kirkpatrick, Lester &
Peterson, 1999). This suggests that cost savings play an important role in the
termination process. Financial models, however, offer at best a partial explanation, as
they do not explain which programs will be cut when financial problems occur.
Moreover, they cannot explain why some government programs survive even under
severe governmental financial insufficiencies while other terminations take place
when there are no financial difficulties. Finally, cutting in response to short-term
financial stress may be counterproductive as cutting is itself costly. Considerable
costs may be incurred to compensate the parties negatively affected by the termination
policy (Frantz, 1997).

The efficiency perspective, on the other hand, argues that the poor performance by
public programs is the main reason for termination. This approach insists that a public
program has the mission of providing services in an efficient and responsive way;
failing that, the program will lose legitimacy for existence and thus become the target
of program cutbacks. One case in point is that a Comprehensive Employment and
Training Act (CETA) program was terminated after it proved to be inefficient.
Another example is the decision made by Secretary of Defense Robert McNamara to
discontinue the Skybolt missile program because there were more efficient alternatives
for maintaining strategic deterrence (deLeon, 1983). Efficiency models, though
16
desirable, have difficulty explaining why public programs, whether efficient or not,
tend to exist for a long time (Kaufman, 1976), and many inefficient programs do
survive inefficiency complaints. Furthermore, there are considerable challenges
associated with estimating both costs and efficiencies, thus making applying efficiency
criterion difficult in many situations. More fundamentally, it seems unlikely that such
an argument would be persuasive unless it is put in the context of a more effective
way to achieve the desired goal. Therefore, this motivation might be more usefully
subsumed in a broader focus on ineffectiveness or failure. Public programs could be
judged ineffective either because they are inefficient (could be done more cheaply by
others), ineffective (there are better ways to achieve the same goal), or obsolete (the
goal is no longer worthwhile).

The political perspective, however, contends that policy termination is a political
process (Bardach, 1976). The fundamental reason for program termination, it argues,
lies not in the economics or efficiencies of the program, but in political beliefs
(deLeon, 1983; Harris, 1997). The examples cited by deLeon include President
Nixon’s assault on the Office of Economic Opportunity and President Reagan’s attack
on the California Rural Legal Assistance program. Several empirical studies support
the relative importance of political factors over economic ones in explaining
termination. Lewis (2002), for example, in his study of 426 federal agencies existing
between 1946 and 1997, found that political factors are a primary cause of termination.
Franz (1992), in a case study of federal policy with respect to leprosy, provides
evidence against cost savings and policy ineffectiveness as reasons for termination.
17
She concludes that the role of politics dominates, citing a 50-year effort to terminate
thwarted by a coalition of patients, staff, and the community that housed the
leprosarium. Political models highlight the importance of political dynamics in policy
termination in such a highly politicalized arena. However, they fail to explain why
some termination cases happened without obvious political ideology changes.
Furthermore, economy or efficiency assessment is usually carried out before
government makes a termination decision. If the decision is purely a political one,
such economic analyses would not have been needed.

It is unlikely that any of these forces are the sole source of policy termination
decisions. Some integration of financial, efficiency and political factors is likely at
work in almost all cases, and thus we need models that incorporate all three and
explore their relative importance in different contexts. At the same time, it is
important that we place factors inhibiting and facilitating the termination decision in
the same model so as to evaluate more accurately their roles and the interaction among
them. Kirkpatrick, Lester & Peterson (1999) take an important step in their
development of a descriptive model for terminating federal programs that includes
inherent program characteristics, the political environment, and specific barriers to
termination. However, much more work is needed to develop integrative frameworks
and test them systematically. As Kirkpatrick et al. (1999) observe, “Of all the phases
of the public policy process, policy termination remains as the phase with the greatest
potential for scholarly contribution because the theoretical framework for analysis is
still incomplete and/or not well developed.”
18

This study continues the effort to build more comprehensive models and conduct
model testing. Considering that most empirical work has taken the form of case
studies (Frantz, 1992; Daniels, 1995; Harris, 1997; Cameron, 1978; Kirkpatrick, et al.,
1999) and quantitative studies are rare (Daniels, 1997; Lewis, 2002), this study works
to test the model with quantitative data, in an attempt to make more valid
generalizations. Given that most of current researchers focus on federal policies,
programs, and organizations (e.g., Kaufman, 1976; Frantz, 1997, 2002; Kirkpatrick,
James & Peterson, 1999; Lewis, 2002), the model developed in this study maintains an
emphasis on local termination decisions that have not been widely studied.
Furthermore, local decision makers are closest to both the pro- and anti-termination
forces, and the consequences of ending policies or organizations will be especially
clear. Thus, this arena has the capacity to be particularly revealing about the
determinants of termination.

2.2 Model Development

The conceptualization of the termination process developed in this study differs from
the existing literature in an important way. Most of current research perceives that
policy termination is a direct result of the interaction between different factors, thus
implicitly assuming a one-stage model (deLeon, 1978; Kirkpatrick et al., 1999). This
study, however, proposes to treat the termination decision as a two-stage process.
Arguably, the status quo will prevail unless there is a triggering event that prompts
19
reconsideration of the policy. When such triggers achieve a threshold level, we enter a
decision-making context in which the decision is made. The perceived distinction
between the triggering factors and the decision context suggest that a two-stage model
is more appropriate to describe the termination process. The triggering events here
include fiscal problems, perceived policy failure, and changes in ideology. Decision-
making context factors include the characteristics of government structure, the
characteristics of the targeted policy, the influence of interest groups, and the
characteristics of the community served. The model is elaborated below and
illustrated in Figure 1. I begin with a discussion of the triggering factors, and then
turn to the decision-making context.

Triggering Factors

Termination decisions require the reconsideration of an existing policy. In order for
such a reconsideration to occur, something must place the issue before the decision
maker, i.e., put it on the policy agenda. As noted by Kingdon (1984), problem
recognition is critical to agenda setting. Problems create opportunities for a new
round of discussion and solicit necessary solutions. The likelihood of a proposal or a
subject rising to the policy agenda thus is markedly enhanced if it is connected to an
important problem. Both the termination and broader agenda-setting literature suggest
that the types of problems that are likely to trigger a reconsideration of a policy
include fiscal problems, organizational failure, and swings in political ideology. I
consider here the role of all three as triggering events for termination decisions.
20

Fiscal Problems

A program or organization’s survival depends on the resources of its environment.
When the capacity of the environment to support the organization declines, this
resource dependency will threaten its survival (Pfeffer and Salancik, 1978). The
greater the dependency level, the greater the risk. Public programs are almost totally
dependent on government funding, and thus are presumably vulnerable to reductions
in governmental support. Indeed, Levine (1978) views the decline and death of
government organizations as just a symptom of resource scarcity, which creates the
necessity for governments to terminate some programs, lower the activity level of
others, and confront tradeoffs between new demands and old programs. Budget
deficits and reduced revenue flows reduce the resources available for all public
spending. Severe reductions in available resources should promote the reconsideration
of public policies. As observed by deLeon (1983), financial or budgetary constraints
have been increasingly posed as stimulating program terminations or at least
retrenchments. Thus, it is expected that:
Proposition 1: The greater the decline in governmental resources, the more
likely is policy termination.

21
Perceived Policy Failures

Any program or organization has a mission to serve and a performance criterion to
meet. Failing that, it will face pressure for re-organization. Ample literature on
organizational change suggests that fundamental organization changes, such as
closure, are more likely to happen when faced with performance failures (Alexander,
1996; Greve, 1998). Performance problems are a common phenomenon in
organizations, but government programs are particularly vulnerable because the lack
of competition and market signals reduces the incentives as well as the information
needed for effectiveness (Downs and Larkey, 1986; Wolf, 1979). For public programs
or organizations, they are expected to serve local needs in an effective and responsive
way. It is from this mission that a program is seen by the public to possess or not
possess legitimacy for its existence (Kirkpatrick, et al., 1999). Failing to meet this
mission creates pressure for policy change, such as reform or termination. Such a
policy change is even more likely if there are effective alternative policy tools
available or there are alternative providers in the private sector (either nonprofit or for-
profit) that could replace public providers (Ferris and Graddy, 1991). Therefore,
Proposition 2: The poorer a public program’s performance is, the more likely
it is to be terminated.

22
Ideological change

Political ideology is the third factor to trigger termination. Ideology is said to provide
the intellectual and emotional pressure required to “convince” the system to change
(Cameron, 1978). Organizational inertia and entrenched interests ensure that there
will be substantial resistance to any policy change. Strong beliefs that require change
can provide the needed energy and legitimacy to overcome these forces. The most
relevant ideological change for the purposes of this study is the substantial change in
preferences about the role of government which first manifested in the early 1980s
with the elections of Ronald Reagan in the United States and Margaret Thatcher in
Great Britain, forming a phenomenon of “intellectual eagerness” for termination
(Frantz, 1992). Such changes in the preferences of citizens and politicians are likely
to produce efforts to reduce the size of government by terminating certain policies,
programs or agencies. It is observed that even in periods of relative financial plenty,
ideological precepts identify certain programs for attack (deLeon, 1978). Thus, it is
predicted that:
Proposition 3: An increase in the ideological preference for reduced
government increases the likelihood of policy termination.

Decision-Making Context Factors

Triggering events place public policies and their possible termination on the policy
agenda, but they do not mean termination will necessarily happen and, if termination
23
happens, they do not tell which program will be terminated. Therefore, here I consider
the second stage of the decision-making process, the decision context.

When triggering events force the reconsideration of a policy, decision makers have 3
broad options to consider – maintain the status quo, change the way the policy is
implemented, or terminate the policy. Maintaining the status quo is viewed as the
default policy option. The general forces of inertia will ensure this choice unless the
triggering events are strong and important to the decision maker. Even in this case,
the negative impacts of termination on the community and/or interest groups that are
valued by the decision maker favor the “do nothing” choice.

In some cases, changing policy implementation offers a viable response to triggering
events. Common alternatives include reducing the scope of the policy (e.g., by
cutting its budget), utilizing cheaper means of service delivery (e.g., contracting out),
reforming public provision to increase efficiency, or reducing the financial burden by
securing outside funding or raising taxes or user fees. The feasibility of these
alternatives is dependent on the policy characteristics and the specific triggering
events that define the problem to be solved.

Finally, the policy can be terminated. In some cases, this option will only be
considered after efforts to change the policy’s implementation have failed to achieve
the desired outcome.

24
The feasibility of policy termination is affected by the decision-making context within
which it takes places. There is a decision maker and a feasible choice set. Both are
influenced by the government structure within which the decision is made. By
structure, it means both the jurisdiction (e.g., state, county, city) and the governance
arrangement within the jurisdiction (e.g., city manager versus city council or mayoral
forms of city government). Since structural characteristics create incentives and
constraints for decision makers, we expect them to affect policy outcomes.

The decision context is broader than its institutional structure. It includes interests and
impacts that are unique to the policy and to the community served. How the decision
maker views the choice set alternatives and their impacts, as well as the weight given
to the perspectives of competing interests, depend on the characteristics of the
community served. In addition, the choice set itself is affected by the characteristics
of the policy.

To examine the role of the decision context in more detail, I first discuss the role of
government structure in a general sense, with a specific hypothesis developed in
chapter 3. Then I explore the expected effects of policy, interest group and
community characteristics on termination decision.

25
Government Structure

Government structure is an important part of the policy-making system that affects the
policy outcome (Dye, 1975; Moe&Cadwell, 1994). First, when making a decision,
decision-makers have to follow certain standard operating procedures and rules
(March and Olsen, 1976). This affects what decision can be made and how quickly it
can be made (Scott, 1995). Second, the designs of the government structure can
provide different managerial authority and political power for local officials to spend
the resources and deal with the issue of declining resources, including the termination
decision (Levine et al., 1981). Furthermore, government structure, as part of formal
and informal norms, shapes the value preferences of the actors within it, thus affecting
the weight they put on each alternative (Stone, 1982). Finally, structure also limits the
choice set, as available alternatives depend on the legal authority of the jurisdiction.
Such an institutional constraint is expected to play an important role in the termination
process. However, the specific role of structure depends on the policy arena, and I
develop it later for the analysis of local public hospitals.

Policy Characteristics

The feasibility of the termination option depends on the policy characteristics
operating within the jurisdiction. These include the characteristics of the program as it
is currently being implemented, the demand for the service, and the private sector
capacity for provision. These characteristics reveal the importance of the policy and
26
the consequences that are likely to be generated by termination, and thus are common
elements under scrutiny before any important policy changes are made on service
delivery. I consider here the impacts and interests associated with policy
characteristics, and how they are expected to affect the termination decision.

Impacts depend on the size of the program and the extent of need or demand for the
services provided. Large programs that currently service a significant proportion of
the population, as well as those for which there is substantial unmet demand are likely
to generate bigger repercussions and discourage officials from deciding on
termination. Thus,
Proposition 4: Large programs, and those in high demand are less likely to be
terminated.

The availability of alternative modes of service delivery may also impact the
termination decision. For example, if there is an existing private service provider,
decision makers may view private provision as an alternative to meeting service needs,
thus increasing the likelihood of termination. A case in point is that private ambulance
providers in the community decrease the rationale for public providers. Therefore:
Proposition 5: The development of private sector service increases the
likelihood of termination.

27
Interest Groups

The existence of interest groups is another important factor influencing the decision-
making process. In a democratic system, different groups actively participate in the
political system (McFarland, 1992). They are involved in electoral politics, lobbying
the legislature systems, the executive branches, and the courts (Wilson, 1990; Densau
and Munger, 1986). They draw attention to public problems, promote solutions,
introduce bills to legislators, or provide important information or expertise for
legislators (Grossman& Helpman, 2001). The existence of interest groups is a
constraint on governmental action in all political systems (Richardson, 1993).
Policies, in a sense, are not the product of a series of rational decision-making stages,
but the result of political competition, bargaining and compromise among different
actors or forces (Lindblom, 1959). As Dahl (1982: 52) observed, “Most of the actions
of government can be explained … as the result of struggles among groups of
individuals with differing interests and varying resources of influence.”

During the process of termination, active interest group activities have frequently
formed, which typically manifest as a termination or anti-termination coalition
(deLeon, 1978; Kirkpatrick, 1999). Two most important interest groups are the
beneficiaries and the public employees who are directly affected by the policy
termination decision. Terminating a policy has direct, personal and negative impacts
on those currently employed providing the service and thus they resist the termination
decision (Behn, 1978; Levin, 1978). Their opposition is likely to be most effective if
28
the group is large and well-organized (Olsen, 1965; Graddy, 1991). For example,
when public employees are unionized they may have substantial influence on
politicians. Therefore,
Proposition 6: The more powerful (in terms of size and organization) the
public employees, the less likely is policy termination.

The beneficiaries of public programs are also likely to resist termination. Their
effectiveness will depend on their capacity to organize and influence decision makers.
The vulnerability of a political group will limit its capability to resist budget decreases
and termination (Levin, 1978). A strong political opposition, however, can force a
public administrator to reverse a termination decision. For many public programs, the
beneficiaries are poor and unorganized, thus limiting their potential for resisting. In
some cases, nonprofit advocacy groups represent their interests. In others (e.g., trauma
services), beneficiaries are a much broader group and will have more political
influence. Thus, it is reasoned that:
Proposition 7: The more powerful (in terms of size and organization) the
beneficiary group, the less likely is policy termination.

In summary, the feasibility of the different alternatives in the choice set, as well as
how the decision maker evaluates the impacts and interests associated with those
alternatives will be affected by government structure, by the policy characteristics that
include the size of the program, the extent of the demand for the service, the
development of the private sector, and by the influence of interest groups of employee
29
and beneficiary groups. How the decision maker weighs the interests for and against
termination will also depend on the characteristics of the community served, which I
consider next.

Community Characteristics

Community characteristics affect the preferences of public officials and provide
incentives for certain kinds of behavior, thereby influencing policy outcomes (Stone,
1980; Lowery, 1987). This is consistent with broader decision-making literature that
suggests the environmental context is an important factor affecting policy outcomes
(Dye, 1967). A local government is especially affected by the community context
because, first, it is geographically small. Compared with state- and federal-level
government, local people more easily participate in the local policy-making process
directly (due to less time consumption and transportation cost and the accessibility of
local officials). Second, as most local policies are directly related to people’s daily
lives, people are more likely to express opinions in the local policy-making process.
Third, elected official’s behavior is more observable to voters in the local government
policy-making process. Therefore, we expect that local officials will pay much
attention to local community characteristics that determine the local preferences and
needs. Among them, four most important factors for the termination decision
considered here are the community’s commitment to the service arena, its political
philosophy, the extent of its homogeneity, and its wealth.

30
Communities differ in their commitment to different service arenas. Many residents
are attracted to communities because they value a particular service, e.g., education.
Presumably, programs and services in such favored policy arenas are less likely to be
terminated. Although commitment is difficult to measure, one indication may be the
proportion of a local budget allocated to the policy arena. For example, communities
that spend more than average on law enforcement may be signaling their commitment
to these services. Therefore, it is expected that:
Proposition 8: Functions or programs in policy arenas that represent larger
proportions of government budgets are less likely to be terminated.

A community’s preferences about the role of government are part of its broader
political philosophy. Communities that are more conservative will be skeptical of
government’s role in service delivery. This philosophy should encourage elected
officials to favor private sector provision of services, and to favor private-sector
interest groups over public sector unions. Termination is more likely to be an
acceptable option in such communities. Moreover, the preferences and
predispositions of public officials are likely to be a manifestation of underlying
community political characteristics (Stone, 1980). Indeed, interest in reducing the size
of government is likely to be a salient election issue, yielding like-minded public
decision makers. Therefore,
Proposition 9: Policy termination is more likely in politically conservative
communities.

31
Communities with homogeneous populations are more likely to have stable
preferences. These stable preferences can of course be conservative or liberal, but
homogeneous populations should have more agreement about what services they want
to provide. Once it is agreed that a service will be provided, it is less likely that it will
be terminated in an environment of stable preferences. However, researchers have
found local governments in homogeneous communities are more likely to contract
with the private sector (Ferris, 1986). This willingness to contract services might also
mean a willingness to privatize and rely completely on private service providers.
Therefore, only the importance of this characteristic, not its expected effect, is
considered:
Proposition10: Homogeneity of population in a community affects the
termination decision.

Finally, wealthy communities have more choices as to the services they provide and
how to fund them. This allows their decision makers flexibility that poorer
communities don’t possess (MacManus&Pammer, 1990). They can choose to offer
services longer in the face of reduced revenues from other governments, or can rely on
user fees or even increased taxes to fund desired services. Thus wealthy communities
should have less need to terminate some policies.
Proposition 11: Wealthier communities are less likely to terminate policies.

We thus expect four community characteristics to influence the termination decision.
Communities will be less likely to decide on termination in a policy arena to which
32
they are committed. Decision makers in politically conservative communities should
be more likely to terminate policies, and those in wealthy communities should be less
likely to do so. We also expect the homogeneity of a community to affect its
propensity to terminate policies. These characteristics represent the final component
of the model.

Summary

To summarize, I have developed a two-stage model of the policy termination decision.
Triggering events force the reconsideration of a policy and frame a problem for which
policy termination is one alternative. The decision to terminate or not is affected by a
decision-making context that includes government structure, the characteristics of the
policy, of the interest groups and of the community. The model is summarized in
Figure 1.

Decision
Policy Characteristics
• Program characteristics
• Service demand
• Private capacity
Community
Characteristics
• Service
commitment
• Political
philosophy
• Homogeneity
• Wealth
Interest groups
• Public employees
• Beneficiary groups
Impact & Interests
Decision
Maker
Choice Set
• Status Quo
• Change
implementation
• Termination
Government Structure

33

Figure 1: A Decision-Making Model for Policy Termination
Triggering
Factors
• Fiscal problems
• Perceived policy
failures
• Ideological change

34
Chapter Three: Application to the Termination of Public Hospitals

The public hospital system in California as well as the nation as a whole has
undergone considerable change over the past few decades. The large reduction in the
number of public hospitals has provided us an opportunity to empirically explore the
policy termination decision by local governments. This chapter applies the
termination decision model developed in the previous chapter to the public hospital
context, with a particular focus on California. I focus on the State of California
because the California Office of Statewide Health Planning and Development
(OSHPD) maintains a high-quality dataset for public hospitals within the state, which
greatly facilitates the analysis. Furthermore, the trend of termination of public
hospitals in California has been very similar to that of the nation, which enables
meaningful comparisons and enhances the generalizability of the results. In this
chapter, I first briefly discuss the trend of and research on public hospital termination.
Then I provide the background information of the US and California public hospital
system. Finally, I apply the termination model to the public hospital context and
develop specific hypotheses for the public hospital termination decision.

3.1. Termination of Public Hospitals

In recent decades, the number of public hospitals in the United States has suffered a
large decline. Between 1985 and 1995, the number of public hospitals in the United
States declined by nearly 14%. During this period, 293 public hospitals converted to

35

private ownership or management, and 165 closed; an additional 20 formerly public
hospitals closed after converting to non-public status (Legnini, 1999).

Consistent with this national trend (Andrulis and Duchon, 2005), ownership in the
California hospital industry has also undergone considerable changes. In 1964, 49 of
the state’s 58 counties operated 65 public hospitals. By 1981, the number had shrunk
to 29 counties operating 37 hospitals (Brown, 1981). Besides the county hospitals,
another important component of the California public hospital system – district
hospitals also experienced a similar declining trend. In 1972, there were 73 district
hospitals in California comprising about 11% of the state’s hospitals. In 2000, there
were 50 district hospitals, comprising about 9% of the state’s hospitals (Eldenburg and
Krishnan 2003). In a study conducted by Ferris and Graddy (1999), the relative share
of the California hospital industry held by public hospitals was found to have declined
from 20% in 1981 to 17.5% in 1990. The number of public hospitals declined 18%
from a combination of facility closures and transfers in ownership to the private
sector. In another study, between 1986 and 1996, thirteen California hospitals were
found to have switched from government to nonprofit ownership and one hospital was
sold by a government entity to a for-profit organization (Spetz et al., 1999).

The following two graphs (Figure 2a and 2b) illustrate the trend of termination of US
and California public hospitals in a comparable period. Between 1982 and 1995
1
, the

1
I wasn’t able to obtain the number for US public hospitals in 1981, or the number for CA public
hospitals in 1980. Thus the earliest year of comparison starts from 1982.

number of US public hospitals decreased from 1783 to 1334, which constituted a 21%
reduction (Chakravarty et al, 2005). During the same period, the reduction in the
number of California public hospitals was 22%, a number remarkably close to that of
the nation. From the following two charts, we can also see that the rates of
termination for both US and California public hospitals are pretty similar in all years
between, though the US curve seems a little smoother. Thus, the analysis of
California public hospital termination decisions is likely to produce very generalizable
results.

Figure 2: The Termination Trend of Pubic Hospitals (comparing US and CA):
2a:
US public hospital
0
200
400
600
800
1000
1200
1400
1600
1800
2000
1980 1982 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995
Year
N um be r of hos pita ls

36

Figure 2, continued
2b:
CA public hospital
0
20
40
60
80
100
120
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995
Year
N um b er of h ospitals

Although the termination of public hospitals is well observed and the eroding role of
the public hospital has important implications for public health, there is little research
in this area. As Needleman et al. (1999) observes: “Since 1980 about one-quarter of
U.S. public hospitals have closed or converted to for-profit or nonprofit status. This
has occurred with virtually no study of the reasons for or impact of closure or
conversion.” Most of the existing research has focused on private hospitals (Young &
Desai, 1999; Desai et al., 1998; Tami, 1999); on the few occasions when public
hospital behavior is studied, it is often analyzed along with private hospitals within a
theoretical framework of private hospitals (Needleman, 1999; Thorpe et al., 2000;
Sloan et al., 2003). Such an approach is problematic for public hospitals, given the
important difference between the private and public sectors (McKay&Coventry, 1993;
Alexander et al., 1996). For example, public hospitals depend heavily on government
37

38
money and are influenced by government financial policy; they are affected less by the
pressure of competition and more by institutional constraints and political factors, just
to name a few. Therefore, to better understand public hospital behavior, it is important
for us to examine public hospital phenomenon within a public context and employ
new conceptual frameworks.

What are the determinants of the termination decision of local public hospitals? Is it
due to local government’s financial problems, performance failures, a shift in
ideology, or other contextual factors? Which factor is most important? What insight
can termination theory shed on this phenomenon in the local health context? The
following section is dedicated to answering these important questions by applying the
theoretical model developed in chapter 2. Before formally exploring these questions,
it is necessary to first examine the background information about the US and
California public hospital system, which specify the institutional and political
constraints on hospital termination.

3.2. The Role of American Public Hospitals

American public hospitals include federal, state, and local public hospitals. Federal
hospitals typically have been designed for special beneficiaries such as veterans,
American Indians, merchant seamen and military personnel. State hospitals typically
provide long-term psychiatric and chronic care; in the past they have especially served
patients with tuberculosis. There are also state university or teaching hospitals that

39
provide short-term general acute care (Kovner, 1990). The largest proportion are city
or county hospitals. There are also district hospitals located in hospital/health care
districts created by statute and granted legal rights to assess property taxes to support
hospital operations (Little Hoover Commission, 2000).

Public hospitals have been playing an important role in protecting public health. They
provide a full range of services, the provision of emergency and specialized services,
and various other objectives (Shanks, 1988). With their large volumes of outpatient
and emergency visits, public hospitals are major resources for primary and other
ambulatory care in the community. For example, in 1990, public hospitals in the
nation’s largest central cities averaged 210,000 outpatient visits and 74,000 emergency
room visits, which was three times more than their private hospital counterparts
(Andrulis et al., 1995).

The primary mission of the public hospital, however, is to provide health care services
to all persons in the community regardless of their financial status, health status, race,
culture and social status (Andrulis et al., 1996). This especially helps those most
vulnerable populations, such as the disadvantaged, under-insured and uninsured
(NAPC, 1994). For a long time, the public hospital has been the last resort for the
uninsured and underinsured (Brown, 1981). For states with a large under-insured and
uninsured population, such as California, the role of public hospitals is especially
important.

40
A second critical mission of public hospitals is to provide costly, specialized care such
as emergency services, neonatal intensive care, bum care, and trauma care to the larger
community. During 1990, public general hospitals represented approximately 8% of
all institutions in the 100 largest cities, but provided 21% of the neonatal and pediatric
intensive care days and 38% of the burn care days (Andrulis, 1995).

Finally, many public hospitals also conduct medical education and research, which
benefit society at large (Andrulis et al., 1996).

3.3. The California Public Hospital System

In California, the public hospital system consists of state, county, city, city/county and
district hospitals. Though just 6 percent of all hospitals statewide, public hospitals in
California provide more than half the hospital care to the state’s 6.5 million uninsured,
deliver 11 million outpatient visits per year, operate more than 60 percent of the top-
level trauma and burn centers, train half the doctors in the state and provide more than
60 percent of the state’s psychiatric emergency care (CAPH, 2005). Except for a few
state, city and city/county hospitals, most of California hospitals are county and
district hospitals. For the purpose of model discussion, this study focuses only on
county and district hospitals.

County hospitals have been mandated as the healthcare providers of last resort since
1935 (codified under Section 17000 of California’s Welfare and Institutions Code),

41
with the obligation to “relieve and support all incompetent, poor, indigent persons, and
those incapacitated by age, disease, or accident”. California law (under Sections
1444-1445 of the Health and Safety Code) further authorizes the board of supervisors
of each county to establish and maintain a county hospital, prescribe rules for its
government and management, appoint staff, and perform other functions (Roemer and
Shonick, 1980). The state, however, offers little in the way of guidelines, leaving
counties with considerable latitude as to how services are organized and provided,
including the target population and the scope of services (Cousineau, et al., 2003).

County hospitals were initially a popular means to fulfill these obligations. Prior to
the passage of Medicare and Medicaid in 1965, county hospitals were the principal
source of healthcare for the poor in California (Shonick and Roemer, 1983). Counties,
however, began to close public hospitals in the early 1970s raising public concerns.
The 1974 Beilenson Act was designed to slow these closures by requiring local
governments to demonstrate that any proposed changes would not have a detrimental
impact on indigent healthcare, or requiring alternative service provision arrangements
if they did (Sections 1442 and 1442.5 of California’s Health and Safety Code).
Moreover, this Act required that the quality of care provided to indigents be the same
as that available to non-indigents from private providers in the county (Shonick and
Roemer, 1983).

Although the primary purpose of county hospitals is the provision of indigent
healthcare, they also often provide specialized services to the broader population. For

42
example, some provide emergency and trauma care, and/or burn units. Medical
training and research are also often provided by county hospitals.

Public hospital services are also provided by healthcare districts. These special
purpose governmental entities are created by statute and allow under-served
geographically-defined communities to offer healthcare services. California has 74
healthcare special districts, formed mostly in the 1940s and 1950s to build and operate
hospitals and deliver health care in rural areas, where the population density is low and
it is difficult for the county hospitals to reach those patients (Eldenburg and Krishnan,
2003). Districts are governed by boards appointed by local officials or publicly
elected, and board meetings are open to the public. District hospitals are not expected
to provide free services to patients who qualify for state and county support
(Eldenburg and Krishnan, 2003).

Though district hospitals in California tend to be found in rural areas and are often
smaller than other hospitals, they also play an important role in providing health
services to people within or outside the district. For example, several district hospitals
(e.g., the Mountain View Hospital in El Camino, Tri-city Hospital in San Diego, etc.)
are among the 15 largest hospitals in California in terms of bed size, some rank high in
specific services (e.g., Sequoia District hospital in Red Wood City), and some are
famous for their high quality (e.g., Plumas District Hospital).

43
California county and district hospitals are jointly financed by federal, state, and local
governments. Federal funding is primarily Medicaid (called Medi-Cal in California)
and Medicare based funding, and Medicaid Disproportionate Share Funding (to
provide additional assistance to hospitals serving a disproportionate number of low
income patients). State funding includes the state portion of Medi-Cal reimbursement,
and revenues from a variety of other sources, including the sales tax, vehicle licensing
fees, and the tobacco tax.

Local governments provide direct support through county revenues. In 1997,
California counties spent an average of 11% of their budgets on health services
(County and City Data Book). Special districts can also directly levy property taxes to
support their operations. However, the ability of all California local governments to
raise revenues was severely curtailed in 1978 by Proposition 13, which limited
property taxes to no more than 1% of assessed valuation and limited annual increases
in assessed valuation to no more than 2%.

Given this background on the institutional setting of local hospital services in
California, now I turn to the development of hypotheses.

3.4. Termination Hypotheses

As discussed in the previous chapter, the termination decision is affected by triggering
events (i.e., fiscal problems, perceived policy failure, and changes in ideology), and

44
decision-making context factors (i.e., the government structure, the available choice
set, the characteristics of the targeted policy, the influence of interest groups, and the
characteristics of the community served). The same sets of factors are expected to
affect the process of terminating local public hospitals. Next I analyze the effect of
each of these factors, starting with the triggering factors, followed by the decision-
making factors.

Triggering Factors

Fiscal Problems

Similar to other public programs, public hospitals rely heavily on government money.
According to a recent study, Medicare, Medicaid, and state or local governments pay
almost 80 percent of the services provided by the National Association of Public
Hospitals and Health Systems (NAPH) members (Singer et al., 2002). Such a heavy
dependency on government funding severely jeopardizes hospitals’ survival when
government faces financial difficulties. This is especially true when the program cuts
are big and the political voice of public hospital beneficiaries is weak. Indeed,
government financial problem is a most-often cited reason for governments to
terminate public hospitals. In a survey of more than one hundred mayors, one of the
primary factors attributed to the community hospital’s closure is insufficient
governmental reimbursement (Hart et al., 1991). Since California county and district
hospitals are jointly financed by federal, state, and local governments, fiscal problems

45
at any of these levels of government could trigger a termination. I examine the effect
of each level respectively in more detail.

1) Federal Fiscal Conditions

The federal government affects local decisions through a variety of means. It can
influence local decisions directly by providing money to local governments or
indirectly by providing money to state governments that will later allocate all or part
of the funding to local governments. The most common funding transfers include
revenue sharing, categorical grants, and bloc grants. It has been found that the federal
aid to state and local governments has increased as a percentage of state and local
revenues and thus expenditure (Rebovich, 1985). With the increase in dollars
amounts, the effect of the federal government on lower level governments also
increases.

Health care is one of the most important categories in which the federal government
exerts influence. The federal government funds Medicare that provides health
insurance to most persons over 65 years old and to certain disabled persons. It also
contributes approximately 50% of Medicaid (Medi-Cal in California), the primary
source of health care coverage for low-income individuals who lack medical insurance
but are the major patient group of public hospitals. Furthermore, it provides matching
funds to the Disproportionate Share Hospital (DSH) program that is aimed at making
up the short fall for hospitals when care is provided to a patient who has little or no

46
funds to cover the cost of care or who is a Medi-Cal beneficiary, which is the common
case for the public hospitals. In extreme cases, the federal government can also
provide financial assistance in the form of a Waiver to help local government
restructure the health service delivery system and avoid termination.

All these suggest that federal funding has an important influence on local
government’s health delivery decisions. When the federal government’s financial
situation is good, it is more likely for it to support those state and local governments
that are having financial difficulties and help rescue failing public hospitals. But when
its financial situation is deteriorating, it is likely that it will cutback the funding, thus
weakening state and local government’s capability to support public hospitals and
increasing the probability of hospital termination. For example, cutbacks in federal
Medicaid payments to states were found to significantly worsen the budgetary
problems states were already experiencing with their Medicaid programs (Altman et
al., 1989). Similar complaints were also found among hospital administrators that
public hospitals have been threatened disproportionately by federal (and state)
government decisions, because Medicaid is no longer affordable at current funding
levels (Siegel, 1996). Therefore, I expect that:
H1a: The greater the decline in federal financial resources, the more likely the
local public hospital is to be terminated.

47
2) State Fiscal Conditions

The state government has increasingly been found to play an important role in local
spending, either as a funding partner or by specifying the manner in which the money
is used (Florestano, 1985; Yinger, 1990); and health care spending is especially so. In
California, the state government supports local health services in different ways. First,
it pays the difference in the Medicaid program after the federal money. Second, the
State General Fund provides money to qualifying hospitals through the Acute Hospital
Outpatient Disproportionate Share (SB 1179) program, with a cap of $10 million. In
addition, the state government also provides funds through the Construction and
Renovation Reimbursement Program (1732). This program reimburses qualifying
hospitals for a portion of their debt service on revenue bonds that were issued to fund
hospital construction/renovations and Graduate Medical Education (GME) (SB 391),
which provides a portion of teaching costs associated with operating hospitals that
serve eligible Medicaid recipients through the Medi-Cal Selective Provide Contracting
Program. Furthermore, the state government also allocates money from sales tax,
vehicle license fees and tobacco taxes to support local health programs since revenues
from traditional health care sources are not sufficient to fund their operation.

Therefore, when the state government’s financial condition deteriorates, it is likely to
cut down its spending on health care, thereby forcing local governments to terminate
certain hospital services. Thus it can be expected that:

48
H1b: The greater the decline in state financial capability, the more likely the local
public hospital is to be terminated.

3) Local Fiscal Conditions

Besides federal and state government money, local governments are another important
funding source for local hospital’s operation. County government can help local
hospitals by contributing money to DSH programs, and Emergency Services and the
Supplemental Payment Fund (1255) so as to obtain federal matching funds, or directly
subsidize local hospitals through the local budget. Each year the county government
needs to set aside a certain budget to subsidize local health services, of which hospital
services are the most important item. In 1997, California counties on average spent
11% on health services. Special districts can also directly levy property taxes to
support their operations.

The ability of all California local governments to raise revenues was severely curtailed
in 1978 by Proposition 13, which capped the rate of property taxes. As the finances of
local governments become constrained while health service costs rise, some counties
and districts have continued service provision by adopting alternative delivery
methods, while others have been pressured for termination (Cousineau, et al., 2003).
Therefore,
H1c: The greater the decline in the local government’s financial situation, the
more likely the local public hospital is to be terminated.

49
Perceived Failures

Government programs are often the target of re-organization or termination when
functioning inefficiently. For hospitals, poor hospital management is an often-cited
reason for conversion or closure (Hart, 1991). Indeed, performance failure gives the
best legitimacy or justification for government’s decision to terminate such an
essential service as that provided by a public hospital. Moreover, the ample literature
on hospital conversion and closure suggests that termination events are usually
preceded by hospital financial difficulties; hospitals closed or converted are more
likely to have experienced a financial loss prior to the event than the average hospital
(Sloan et al., 2003; Tami, 1999; Wilson, 2000; Alexander, 1996). In California, a
recent report found that the hospital’s financial losses were the single most common
reason cited for the closure of 23 hospitals between 1995 and 2000 (Kagan et al.,
2001). Thus, it is predicted that:
H2: The more poorly a public hospital is performing, the more likely it is to be
terminated.

Ideological change

Ideology defines the extent of government’s obligation in providing public services.
A change of citizens and politicians’ preferences as to the government’s role in
relation to society is likely to trigger a change in the provision of services. The
ideology of federal and state governments in providing health services will affects

50
both their willingness to provide assistance to public hospitals and the management
style of hospital services (Andrulis, 1997). For example, a more conservative
ideology is likely to reduce funding to local public governments and support a
preference for hospitals to be run by the private sectors. Therefore, I expect:
H3: An increase in the ideological preference for reduced government
increases the likelihood of hospital termination.

Decision-Making Context Factors

After a reconsideration of public hospital services is brought up to the agenda by the
triggering factors, the decision maker is faced with the choices of maintaining the
status quo, changing the way the policy is implemented, or terminating the hospital.
What option to choose is influenced by a set of contextual factors, which include
government structure, policy characteristics, the influence of interest groups, and
community characteristics. The expected effects are detailed below.

Government Structure

Most public hospitals in California are operated by government entities of county and
special districts, which differ greatly in their purpose, power and function. Counties
are geographical and political subdivisions of the state and thus serve as important
administrative units for state, and often federal, laws, programs and services. In other
words, counties are bestowed with a broad range of traditional functions, such as to

51
build roads, maintain jails, care for the poor, and to keep records on property and
statistics on people. Correspondingly, a large proportion of county revenues also
come from state and federal sources. Special districts, on the other hand, are units of
local government established by the residents of an area to provide certain services not
provided by the county or city. In contrast to the broad constitutional and legal
authority under which counties and cities operate, the authority of special districts is
restricted to specifically enumerated powers and purposes. Special districts are
limited in their activity, their ability to raise revenue, and their power to regulate
planning and land use.

Given such different institutional constraints, county and district governments are
likely to respond differently to triggering events. Presumably, as a single function
form of government, special districts are less likely to terminate public hospitals, as
doing so would reduce the justification for the existence of the district itself. County
governments, however, face no such threat and thus can be expected to more readily
consider termination. Therefore, it is expected that:
H4: The county government is more likely than special district government to
terminate public hospitals.

Policy Characteristics

The feasibility of hospital termination depends on policy characteristics of health
services. Among them, the size of the hospital, the demand for health services and the

52
capability of the private sector to provide the services are the three most important
elements local government needs to consider. A large hospital plays a bigger role in
accomplishing local government’s obligation in health provision by providing a larger
number of services and serving a larger community population. Terminating big
hospitals invites serious concern about and criticism for the government’s failure in
health provision. In addition, such termination often triggers substantial negative
impacts on the local community, which incurs considerable compensation costs to
relevant parties (Frantz, 1997). Therefore, I reason that:
H5: A larger public hospital is less likely to be terminated than smaller ones.

Similarly, a highly demanded service signals community needs local government
attempts to satisfy, thereby having a better chance to secure the needed amount of
resources. Thus, it is expected that public hospitals with higher service demands are
in a better position to obtain financial support and avoid termination. Therefore, I
predict that:
H6: A public hospital with a higher service demand is less likely to be
terminated.

The likelihood of hospital termination is also affected by the development of the local
private health service market. As a substitute for governmental hospital service,
private hospitals reduce the negative impact of public hospital termination by
providing an alternative to local service seekers, thus encouraging local officials to
choose the termination option. Thus, I expect:

53
H7: Public hospitals located in areas with more developed private health
services are more likely to be terminated.

Interest Groups

While interest groups are a common phenomenon in American politics, their role is
particularly active in the health field. As a dispensable component of health care, the
hospital termination process is heavily influenced by interest groups. Among the
different groups, two groups, public hospital employees and beneficiaries, are
especially important in the termination decision. Fearing the loss of their current jobs
and benefits, they vehemently resist hospital termination. Combined with support
from other groups, such an anti-termination coalition, if strong enough, can effectively
block the possibility of termination. For example, in New York, the sustained,
militant actions by such a coalition saved at least one of four municipal hospitals
slated for closure by a cost-cutting mayor (Behn, 1978). In San Mateo County of
California, the board of supervisors leaning toward closing Chope Community
Hospital reversed its plans after receiving strong pressure from anti-termination
groups (Benedict, 1977). To a large degree, the capability of these groups to influence
the decision depends on how effectively they can mobilize and organize themselves.
Thus it is expected that:
H8: The more powerful (in terms of size and organization) the public
employees and beneficiary groups, the less likely is hospital termination.

54
Community Characteristics

As discussed in chapter 2, community characteristics affect the termination decision
by influencing the decision makers as to how they weigh the interests for and against
termination. The important factors include the community’s commitment to health
service, its political philosophy on governmental service in general, the extent of its
homogeneity, and its wealth.

Local communities differ in their commitment to the provision of health services, as
indicated by their spending on these items. For example, in 1997, Inyo County in
California allocated 37.4% of its budget for health services, while the proportion in
Yuba County from the same state was only 1.1%. Such a commitment directly affects
the capability of public hospitals to provide health care access to the poor (Ginsburg,
1996). Understandably, a less committed community is more likely to adopt the
termination option. A recent study on public hospital conversion found that
conversions are typically a response to the unwillingness of local communities to
provide continued tax support to hospitals (Needleman et. al., 1997). Therefore, it can
be reason that:
H9: Hospitals located in communities with larger proportions of government
budgets are less likely to be terminated.

Hospital termination also depends on local communities’ attitude towards
government’s role in service provision as a whole. Communities with a more liberal

55
political philosophy are more likely to encourage government officials to maintain
health services, even under stressful financial pressure, while those with a
conservative preference are more likely to prefer provision by the private sector and
vote for termination. As observed by Andrulis (1997), local support of the public
sector’s mission will have perhaps the greatest influence on the public sector’s
survival. Naturally, such an effect is expected to extend to the fate of public hospitals.
Therefore, I expect:
H10: Termination of public hospital is more likely to take place in politically
conservative communities.

As argued in the previous chapter, the homogeneity of the local population affects the
termination decision, but the direction is more difficult to predict. Communities with
homogeneous populations are more likely to have stable preferences, which reduce the
termination probability. On the other hand, under certain circumstances, homogeneity
can also work to facility program termination. For example, the opposition to hospital
termination is likely to be smaller in a homogenous community when there is a shift of
preference favoring smaller and more efficient government. Therefore, I include the
hypothesis:
H11: The homogeneity of population in a community affects the decision to
terminate a public hospital.

Finally, consistent with the arguments in chapter 2, wealthy communities have more
flexibility in choices when facing triggering problems for hospital re-organization.

56
Such choices include options of user fees or tax increases, which are effective
alternatives to rescue hospitals having financial difficulties. Thus, the availability of
such solutions can greatly reduce the need for hospital termination, especially when
the problem is mainly a monetary one. Therefore, I expect:
H12: Wealthier communities are less likely to terminate a public hospital.

In summary, this chapter applies the two-stage model of the policy termination
decision to the public hospital context. A reconsideration of hospital service delivery
is triggered by problems of federal, state and local government fiscal deterioration,
hospital performance and ideological change. The decision to terminate or not is
affected by the context that includes the government structure within which the
decision is made, and the termination impact and interests that are connected to the
characteristics of the targeted hospital, the influence of interest groups and the
characteristics of the local community. Twelve corresponding hypotheses have been
developed, and the model is summarized in Figure 3.

57

Figure 3: A Decision-Making Model for Termination for Public Hospitals
Triggering
Factors
• Fiscal problems
• Perceived policy
failures
• Ideological change

Decision
Maker
Policy Characteristics
• Hospital characteristics
• Health Service demand
• Capacity of private
hospitals
Community
Characteristics
• Service
commitment
• Political
philosophy
• Homogeneity
• Wealth

Decision
Choice Set
• Status Quo
• Change
implementation
• Termination
Government Structure
Impact & Interests
Interest groups
• Public employees
• Beneficiary groups

58
Chapter Four: Empirical Testing with a Binary Procedure

This chapter discusses the operationalization of the model specified in chapter three
and presents the results of empirical testing with binary procedure. First I will define
the data source and measures of the dependent variable and independent variables.
Then I will elaborate the methodology of model testing. Finally the empirical result
will be presented.

4.1. Data, Variables and Measures

As discussed earlier, a local termination decision of public hospitals is affected by the
triggering factors of fiscal problems, perceived policy failures and ideological
changes, and the contextual factors of government structure, policy characteristics, the
influence of interest groups, and community characteristics. The model is specified as
follows:

Local termination decision = f (fiscal problems; perceived policy failures; ideology;
government structure; policy characteristics; interest group characteristics; local
community characteristics)

59
Data Source

The model of policy termination is applied to decisions by California local
governments to provide public hospital services over the period from 1981 to 1995.
The information of public hospitals is obtained from the California Office of
Statewide Health Planning and Development (OSHPD). OSHPD maintains a dataset
for all acute care hospitals licensed by the State of California, which includes literally
thousands of variables describing the finances, service offerings, ownership status,
expenses, and employment of each of the some 500 hospitals in the state. To increase
the accuracy of the data of hospital ownership change, I compared OSHPD data with
the American Hospital Association (AHA) annual data. The AHA publishes
ownership information of American hospitals each year (AHA Guide). When a public
hospital was indicated as closed or converted in the OSHPD data, I checked for
consistent information in AHA. Only hospitals that were indicated as closed or
converted in both sources are included as terminations in the analysis. Data for other
variables are collected by searching documents or datasets compiled by governments
and researchers, as specified in Table 1. The study is limited to 1981-1995 for the
reason of data availability. I was not able to obtain clean data before 1981 or after
1995. However, the number of termination cases during this period is large enough to
allow meaningful statistical analyses.

60
Dependent Variable

The dependent variable is a local government’s decision to terminate its provision of a
local public hospital in a given year. Termination occurs when the local government
stops operating the hospital. This occurs either with the closure of the hospital or the
transfer in ownership of the facility to a private (nonprofit or for-profit) entity.

The dependent variable, denoted Termination, is a binary variable, assuming a value
of 1 when a public hospital was terminated in a given year and 0 otherwise. Once a
public hospital was terminated it was no longer followed in this study, therefore I do
not include subsequent ownership changes that might have occurred over the 15-year
period.

Independent Variables

The independent variables in the model include triggering events, government
structure, policy characteristics, interest groups, and community characteristics. The
following develops the measures to be used for model estimation.

61
Triggering Events

I have hypothesized that 3 types of events are likely to trigger reconsideration of a
policy – fiscal problems, perceived policy failures, and change in ideology. The
effects of these events are captured with 7 variables.

Fiscal Problems. Since California public hospitals receive resources from federal,
state and local governments, I include 3 measures for the change in the financial health
of funding governments. Fiscal stress at the federal level is measured by the annual
percentage change in national GDP (denoted FedFinChg). The expected sign is
negative. When GDP is growing, the federal government has more resources with
which to support all of its programs; therefore it is less likely to cut funding to local
governments, thus reducing the chances of hospital termination. Changes in the state
government’s fiscal condition are measured by the annual percentage change in
California’s total revenue (denoted StateFinChg). The expected sign is negative.
Increasing state revenues make it less likely that funding for local government
programs will be reduced. Finally, changes in the fiscal conditions of the local
governments are measured by the annual percentage change in the local government’s
total revenue (denoted LocalFinChg). Again, the expected sign is negative. A local
government with growing revenue is less likely to reconsider its support of public
hospital services.

62
Perceived Policy Failures. The financial performance of the public hospital will be
used to provide an indicator of policy failure. I consider two measures of performance,
the hospital’s total profit margin (denoted TotalMargin) and its operating margin
(denoted OperMargin). Both ratios are commonly used to measure hospital
profitability (Bazzoli and Andes, 1995; Eldenburg and Krishnan, 2003). Total profit
margin is the difference between total revenue and total expenses as a proportion of
total revenue. Operating margin is the difference between operating revenue and
operating expenses as a proportion of operating revenue. The difference between the
two is that the former includes non-operating income such as government subsidies,
private donations, and commercial (non patient-care) activities, and their associated
expenses. Since public hospitals receive government subsidies, total profit margin is
the appropriate measure of their financial performance. Operating margin, however,
may be a better indicator of impending financial crisis (Harrison and Montalvo, 2001),
especially if governmental subsidies are inconsistent from year to year. Therefore, I
consider both measures separately. The expected signs are negative. Increasing
margins indicate better performance, and thus reduce the likelihood of a
reconsideration of the policy.

Ideological Change. I focus on two indicators of changes in ideology that might
trigger reconsideration of the role of government in general and less support for local
public hospitals in particular. The ideological preferences of the President are likely
to influence the behavior of the federal administration toward local governments and
be reflective of prevailing public preferences. The measure (denoted PresIdeology) is

63
a weighted index of presidential liberalism in economic and social policy ranked by
political scientists listed in the American Political Science Association Directory of
Membership (Segal, et al., 2000). Such an index is found to be more accurate than the
common practice of using president’s party affiliation as the proxy and increase
variety in the variable. The expected sign is negative. Higher values mean a more
liberal ideology, which is expected to be associated with fewer efforts to terminate
public programs. State government ideology is also likely to affect support for local
public hospital services. The measure (denoted StateIdeology) is the government
ideology score created by Berry, et al. (1998). This score was estimated based on the
ideological orientation of the state governor and the major party delegations in each
house of the state legislature. The empirical tests show that the measures have a high
level of reliability and usually outperform other indicators in multivariate models
where ideology is deemed an important causal variable. The expected sign is
negative. Higher scores mean a more liberal ideology in state government, and thus
less likelihood that policy terminations will be considered.

Decision-Making Context Factors

Government Structure. There are two types of local governments that provide public
hospitals with different decision makers and institutional constraints: special districts
and counties. Special hospital or healthcare districts are single function forms of
government and therefore expected to be less likely than county governments to
terminate a public hospital. I control for this difference in government structure with a

64
dummy variable (denoted District), which assumes a value of 1 for special districts
and 0 for counties. The expected sign is negative.

Policy Characteristics. The model identifies the size of the public hospital, the
demand for the hospital service, and the capacity of private hospitals to provide the
service as the policy characteristics most likely to impact the termination decision. To
explore these influences, I include 3 measures of the policy to provide public hospital
services. Program size is measured by the number of beds in the public hospital
(denoted Beds). The expected relationship is negative. Larger hospitals are less likely
to be terminated. The demand for public hospital services is measured by the public
hospital’s occupancy rate (denoted OccRate). The expected effect of this variable is
negative. Higher occupancy rates indicate higher demand for public hospital services,
and thus reduce the likelihood of termination. Finally, a large local private sector
capacity allows decision makers to consider private provision as an alternative to
public hospitals. The size of the private hospital sector is measured by the number of
private hospital beds in the county or district per 1000 population (denoted
PrivBedRate). The expected effect is positive.

Interest Groups. The role of employee and beneficiary interest groups in the hospital
termination decision is captured with 3 variables. For program employees, I consider
an indicator of size and one of organization. Size is measured by the number of
physicians employed by the public hospital (denoted Physicians). Employee
organization is measured by the proportion of public employees in California who

65
belong to labor unions (denoted Union). The expected effect of both variables is
negative. Larger employee groups have more influence. More organized members
imply lower organizational costs and thus more effective political mobilization to
oppose a termination decision. The primary beneficiary group for public hospitals is
the poor, so I measure the influence of this group by the percentage of the county’s
population with incomes below the federal poverty level (denoted Poverty). The
expected effect is negative. A larger beneficiary group is more likely to have political
influence that will reduce the likelihood of termination.

Community Characteristics. The model predicts that a community’s commitment to
the hospital and health services, its political philosophy, the homogeneity of its
population, and its wealth will affect the hospital termination decision. Their effects
are captured with 4 variables. The community’s commitment to public health is
measured by the proportion of the local government’s budget devoted to public health
and hospital services (denoted HealthBudget). The expected sign is negative. The
higher the budget percentage allocated for health, the less likely the local government
is to terminate public hospital.

Politically conservative communities are expected to be more positively predisposed
to terminate public hospitals. Political philosophy (denoted PoliticalPhil) is measured
by the percentage of voters within the county who registered as Republican during the
presidential election. Presumably, Republicans are more conservative than Democrats

66
(Erikson et al., 1989). The expected sign is positive, meaning voters that are more
conservative are more likely to support termination decisions.

Community homogeneity is relatively more difficult to measure. A common practice
in policy research is to measure it in racial terms as the proportion of nonwhite in each
county. Considering the ethnicity diversity in California, the proportion of nonwhite
may not accurately reflect the population homogeneity in California. Therefore, I
have constructed a Herfindahl-Hirschman Index (HHI):
n
HHI = ∑ s
i
2
i=1

Where s denotes the share of each of the 5 categories of ethnicity in the county –
Caucasian, Hispanic, Asian/Pacific Islander, African American and American Indian.
The population homogeneity increases when the HHI becomes larger.

Finally, community wealth is measured by per capita income (denoted PCIncome).
The expected sign is negative. Wealthier communities will have more options for
dealing with financial difficulties, thus reducing the likelihood of termination of public
hospitals.

To summarize, the model to be estimated with the expected signs of the coefficients
noted in parentheses is:

67
Termination= f [FedFinChg(-), StateFinChg(-), LocalFinChg(-), TotalMargin(-) or
OperMargin(-), PresIdeology(-), StateIdeology(-), District(-), Beds(-), OccRate(-),
PrivBedRate(+), Physicians(-), Union(-), Poverty(-), HealthBudget(-),
PoliticalPhil(+), Ethnicity, PCIncome(-)]

The variable definitions and data sources are summarized as below (Table 1).

Table 1: Variable Definitions and Sources
Variable Definition Source
Termination Dummy variable, which assumes a value of 1
if a termination occurs, 0 otherwise
OSHPD
Triggering Events
FedFinChg Annual percentage change in U.S. GDP OMB
StateFinChg Annual percentage change in total revenue of
California state government
OSC
LocalFinChg Annual percentage change in the local government’s
total revenue
OSC
TotalMargin Public hospital’s total profit margin, net income
divided by total revenue
OSHPD
OperMargin Public hospital’s operating profit margin, net income
from operations divided by total operating revenue
OSHPD
PresIdeology Presidential ideological ranking of liberality in economic
and social policy
STH
StateIdeology State government’s ideological ranking of liberality BRFH
Government Structure
District Dummy variable of local government form,
which assumes a value of 1 for special districts, 0 for
counties
OSHPD
Policy Characteristics
Beds Number of hospital beds in the public hospital OSHPD
OccRate Hospital occupancy rate; the average number of
inpatients, excluding newborns, divided by the average
number of hospital beds
OSHPD
PrivBedRate Private hospital beds per 1000 population in the county OSHPD

68
Table 1, continued

Interest Groups
Physicians Number of physicians in the public hospital OSHPD
Union Proportion of California’s public sector workers
belonging to labor unions
HM
Poverty Percentage of county’s population with income below the
federal poverty level
ARF
Community Characteristics
HealthBudget Percentage of local budget allocated for hospital and
health services
CCDB
PoliticalPhil Percentage of county voters who registered as
Republicans during a presidential election year
SOV
Ethnicity HHI index of the 5 categories of ethnicity in each county DOF
PCIncome Average per capita income in the county OSC

Notes:
OSHPD: State of California, Office of Statewide Health Planning and Development
OMB: U.S. Office of Management and Budget
OSC: State of California, Office of State Controller
STH: J.A. Segal, R.J. Timpone and R.M. Howard
BRFH: Berry, W. D., Ringquist, E. J., Fording, R. C., and Hanson, R. L.
ARF: Bureau of Health Professions Area Resource File
HM: Barry Hirsch and David Macpherson
CCDB: County and City Data Book
SOV: State of California, Secretary of State, Statement of Votes in California
DOF: State of California, Department of Finance

4.2. Analysis Method

The dataset contains observations on multiple individual units clustered in a higher
level, forming a multilevel model. More precisely, there can be multiple hospitals
operating in the same county, which means that the county characteristics variables are
measured repeatedly. It is likely that hospitals within the same county will be more
homogenous than hospitals in different counties, because they are exposed to a
common set of environmental factors. Therefore, it is reasonable to expect some
common unobserved variation among hospitals operating in the same county. If this

69
within-cluster correlation is not taken into account, the results could be biased (Guo
and Zhao, 2000).

Generalized Estimating Equations (GEE), a quasi-likelihood approach (Zeger and
Liang, 1986), has several features that make it an attractive choice for clustering
analysis. First, it estimates consistent standard errors in the presence of within-cluster
correlation, thus addressing the multilevel structure problem in the model. Second, the
method requires very few distributional assumptions (Zeger and Liang, 1986), thus
giving us confidence in the robustness of the results. Finally, it can be applied to
continuous, binary, categorical, and time-to-event dependent variables, and is
relatively simple to apply in practice through readily available software (Norton, et al.,
1996). GEE requires that the number of clusters be “relatively large,” with a rule of
thumb of 30 (Norton et al., 1996). The cluster size (county) in this study is 45, which
satisfies such a requirement. Although GEE methodology is not familiar to many
social scientists, it has been widely used in biological and epidemiological research in
the past decades (Liang and Zeger, 1986).

Given the binary structure of the dependent variable and the focus on the effects of
covariates, the most appropriate alternative methodology is random effect logistic
regression. This methodology is similar to GEE in that both allow time-dependent
covariates (Zeger and Liang, 1986), and neither address unobservable heterogeneity
that is correlated with other regressors in the model. Random-effect models tend to be

70
used more frequently when there are few clusters (Norton, et al., 1996), which makes
it inappropriate for our case of large cluster size.

Therefore, I use the GEE methodology to estimate the model. Considering the
absence of a logical ordering for observations within a cluster, I use an exchangeable
correlation structure (Horton and Lipsitz, 1999). Since the dataset is a pooled cross-
sectional time series and the dependent variable is dichotomous, I use STATA’s xtgee
procedure and specify a logit function. Finally, to capture the time lag between the
independent variables and the resulting decision to terminate, I use a 1-year lag, i.e.,
the independent variables in year t-1 are used to explain the dependent variable in year
t.

The unit of analysis is the individual public hospital located in a specific local
jurisdiction. Each observation is measured annually over the period from 1981 to
1995, yielding a pooled dataset with a binary dependent variable. An observation is
truncated after the first termination. The final dataset contains 1165 observations, of
which 32 are terminations.

4.3. Results and Findings

Descriptive Results
As showed in Figure 4, a trend of termination can be clearly identified. In 1981, there
were 105 local public hospitals in California -- 37 county hospitals and 68 district

hospitals. In 1995, there were 27 county hospitals and 54 district hospitals, a 23%
reduction in the total number of public hospitals -- 27% for county hospitals and 21%
district hospitals. These numbers are, however, net changes, as they include public
hospital entry (either from new hospitals or more commonly the conversion of private
hospitals to public ownership). A careful examination of the data reveals that between
1981 and 1995, 26 public hospitals converted to private sector ownership and 11 were
simply closed, bringing the total number of terminations to 37. Furthermore, figure 4
also suggests that public hospital terminations took place faster in the 1981-1985 and
1991-1995 periods, while they remained relatively stable between 1985 and 1991.

Figure 4: Trends in the Number of California Public Hospitals, 1981-1995
0
20
40
60
80
100
120
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
Number of Hospitals
County
District
Public total

The descriptive statistics for the dataset is summarized in table 2, disaggregated by the
dependent variable values.
71

72

Table 2: Descriptive Statistics by Dependent Variable Value

Termination = 0
1133 Observations
Termination = 1
32 Observations
T-test
Variable Mean Std Dev Min Max Mean Std Dev Min MaxDifference
(1-0)
T P
FedFinChg 6.97 2.36 3.412.06 6.92 2.58 3.4 12.06 -0.05 0.120.905
StateFinChg 6.93 4.92 -1.1114.28 5.96 5.08 -1.11 13.73 -0.97 1.10.272
LocalFinChg 7.97 9 -29.19 71.89 4.72 10.54-29.19 20.23 -3.25 2.010.045
TotalMargin 0.016 0.11 -1.42 0.41 -0.044 0.15 -0.57 0.14 -0.06 2.240.032
OperMargin -0.11 0.21 -1.48 0.39 -0.18 0.28 -1.35 0.11 -0.07 1.480.149
PresIdeology 29.95 18.04 17.8 67.55 28.28 16.51 17.8 67.55 -1.67 0.520.604
StateIdeology 52.93 14.09 31.42 84.88 52.61 11.94 43.75 84.88 -0.32 0.120.901
District 0.659 0.474 0 1 0.625 0.492 0 1 -0.034 0.40.687
Beds 162.55 205.15 12 1628107.08 118.82 5 428 -55.47 2.540.016
OccRate .56 .20 .05 1.0 .49 .21 .14 .81 -.07 1.90.058
Poverty 12.47 6.22 0.88 31 11.26 5.13 0.88 21.87 -1.21 1.090.277
Physicians 111.02 130.91 0 613.5 88.83 146.81 2 573.5 -22.19 0.940.346
Union 46.88 2.46 43.4 51.4 45.74 2.36 43.4 51.4 -1.14 1.220.224
PrivBedRate 1.58 1.03 0 4.1 1.53 1.1 0 3.58 -0.05 0.260.793
HealthBudget 10.4 4.65 2.77 31.25 8.34 2.68 4.18 14.94 -2.06 4.19 0
PoliticalPhil 0.377 0.052 0.226 0.495 0.38 0.053 0.25 0.474 0.003 -0.230.817
Ethindex 0.55 0.16 0.31 0.91 0.6 0.16 0.35 0.86 0.05 1.67 0.09
PCIncome 17100 4729 852138342 16217 4973 930431050 -883 1.040.299

A simple means difference analysis reveals 6 statistically significant differences
(denoted in red) between the observations of the 2 values of termination.
Terminations occur where local governments experience lower increases in revenues
(4.72% annual growth rate compared to 7.97%), spend less of their budgets on health
(8.34% compared to 10.4%), and in more homogenous communities (0.6 compared to
0.55). Terminated hospitals have substantially lower total profit margins than public
hospitals that remain open (-4.4% compared to 1.6%), are significantly smaller (119
beds compared to 163) and have a lower occupancy rate (49% compared to 56%).

73

These univariate results thus provide some preliminary support for financial and
performance triggers, and for program and community characteristics as influences on
the termination decisions. Now I will turn to the multivariate analysis to identify the
independent effect of each determinant.

Multivariate Analysis and Discussion

The multivariate results are produced by a GEE procedure. The results are presented
in Table 3. I consider the two indicators of hospital performance -- total profit margin
and operating margins – separately. I report coefficients with the probability levels
associated with the Z test for statistical significance noted in parentheses after the
standard errors. Statistically significant relationships are shaded. Since the
coefficients of a logit regression are more difficult to interpret, I also present estimates
of the partial effect of each independent variable on the probability of termination,
calculated at the mean values of each independent variable
2
.

I checked for significant multicollinearity (see Appendix). The only possible concern
is the relationship between the variables of union and presidential ideology, where the
correlation coefficient is .84. However, a VIF test shows that the VIF score for the
union variable is 6.7, and the score for the presidential ideology variable is 4.4. Both

2
In a logit model, the dependent variable is transformed into the log of the odds that a particular choice
will be made: log[P/(1-P)]= α+ β
i
X
i
, where P = Prob (Y=1) = 1/[1+e
-( α+ βiXi
)]. Thus the coefficient β
cannot be explained as that in ordinary linear regression that measures the change in the dependent
variable due to a unit change in a given independent variable. We thus use parameter estimates to
calculate the partial effect: dP/dX
i
= β
i
P(1-P), evaluated at the mean values of the independent variables.

74

are far less than 10, a rule of thumb to conclude whether a multicollinearity problem
exists. In fact, the STATA software can automatically detect a multicollinearity
problem by stopping the program from running or by releasing a warning message.
Without any problem in the STATA procedure, the possibility of multicollinarity can
be excluded
3
.

Considering that the termination rate may vary from year to year, this study attempted
to include year dummy variables so as to isolate the time effects. However, several
factors have prohibited their inclusion: First, the study doesn’t have a large enough
sample size to allow for a test of the 14 year dummy variables, in addition to all
existing independent variables. Second, an appropriate division criterion is difficult to
identify or, if identified, doesn’t work. Since it is impossible to include all year
dummies, the only alternative left is to divide the whole time period into several
intervals. This requires careful selection of appropriate and meaningful cutting criteria;
otherwise, the effect of year dummies is not of much use, because different methods of
division will produce different results. The most reasonable dividing criterion is the
change in the national administration every 4 or 8 years, which is part of the interest of
this study. Unfortunately, the year dummies created with this method have
multicollinearity problems with the presidential ideology and were dropped by
STATA automatically. Thus, year dummies are not included in the end. Nevertheless,

3
Another possible concern is that the correlation might conceal the effect of the union variable. The
statistical results, however, show that the union variable is significant. Therefore, the correlation should
not be a problem.

75
the muticollinearity problem indicates that the year effect should have been partially,
if not fully, captured by the presidential ideology variable in the model.

Table 3: GEE Analysis of the Decision to Terminate Public Hospitals
Total Profit Margin Operating Profit Margin
Coefficient Partials Semi-
Robust
Std Error
(p)
Coefficient Partials Semi-Robust
Std Error
(p)
Constant 15.36 8.68 (.077) 15.70 8.52 (.065)
Triggering Events
FedFinChg .039 0.00055 .094 (.674) .036 0.0004 .09 (.70)
StateFinChg -.0997 -0.0013 .054 (.067) -.098 -0.001 .05 (.057)
LocalFinChg -.05 -0.001 .021 (.017) -0.05 -0.001 .021 (.019)
TotalMargin -1.88 -0.026 1.18 (.11)
OperMargin -.49 -0.009 .84 (.56)
PresIdeology .023 0.00032 .019 (.23) .021 0.0003 .019 (.28)
StateIdeology -.021 -0.0003 .019 (.26) -.021 -0.0003 .018 (.26)
Government Structure
District -.23 -0.003 .52 (.66) -.16 -0.004 .61 (.79)
Policy Characteristics
Beds -.0018 -2.5E-05 .002 (.37) -.002 -3E-05 .0019 (.34)
OccRate -.0072 -0.0001 .010 (.48) -.007 -0.0001 .01 (.50)
PrivBedRate -.49 -0.007 .24 (.044) -.53 -0.008 .24 (.024)
Interest Groups
Physicians .0011 1.6E-05 .0029 (.70) .0009 2E-05 .0028 (.76)
Union -.33 -0.0045 .19 (.095) -.33 -0.005 .19 (.079)
Poverty -.088 -0.0012 .05 (.08) -.082 -0.001 .048 (.09)
Community Characteristics
HealthBudget -.27 -0.0038 .07 (.000) -.28 -0.004 .069 (.000)
PoliticalPhil 3.09 0.043 4.14 (.46) 2.87 0.029 4.10 (.48)
Ethnicity 1.73 0.024 .999 (.08) 1.70 0.02 .988 (.086)
PCIncome -6.7E-07
0.000
6E-05
(.99)
4E-06 0.000 6E-05
(.95)

Number of obs 1163 1163
Chi Sq 70.42 (.0000) 72.48 (.0000)

The results in table 3 suggest that overall, the model provides a good fit for the data, as
the chi-square value is significant at the .0001 level in both estimations. Although the
operating-profit-margin model provides a slightly better fit, there are no important

76
differences between the results using the two different indicators of public hospital
performance. Therefore, I will not differentiate between these two measures in the
discussion.

The results provide support for both stages of the model, as triggering events and
characteristics unique to the decision-making context both affect the decision to
terminate local public hospitals.

The results provide strong support for the role of fiscal stress in triggering policy
terminations. Decreases in the growth rates of both state and local revenues are
significantly associated with the termination of local public hospitals. The results
reveal that a 1% decrease in the growth rate of state revenues increases the probability
of hospital termination by 0.13%, holding other factors in the model constant at their
mean. Similarly, a 1% decrease in the growth rate of local government revenues
increases the chances of termination by 0.1%.

These findings of the importance of fiscal stress in triggering policy termination are
consistent with assertions by local decision makers. For example, Los Angeles
County decision makers cited insufficient government reimbursement as the reason for
more recent reductions in public health services in that community (Cousineau, et al.,
2003).

77
Federal fiscal changes, however, were not found to be significant triggers. This is
interesting as federal funding is often as important as state funding for local hospital
revenues. There is, however, less variation in changes in federal revenue compared to
state, and presumably this translates into less variability in federal support to local
governments.

Neither hospital performance nor ideological triggering events were significant
influences. The insignificance of hospital profitability comes as a surprise, as most
health research on hospital conversion or closure in general suggested it is an
important determinant (Sloan, 2001; Tami, 1999; Wilson, 2000; Alexander, 1996).
Such a result is likely to suggest a reduced role of hospital performance in the
termination decision under a public context. As noted earlier, however, the role of
decreases in the total profit margin as a trigger for termination is at least suggested by
the descriptive results. A larger sample of terminations or a more in-depth analysis
may be necessary to adequately explore its role. In any case, for this particular policy
it is found that fiscal stress dominates as the triggering event most likely to lead to
policy termination.

The results also provide support for the importance of the decision-making context.
Five variables, including two community variables, two interest group variables, and a
policy variable, have statistically significant impacts.

78
Recall that the proportion of the local budget spent on health activities is used as an
indication of the community’s commitment to providing health services. As expected,
communities that devote a higher proportion of their local budget to healthcare
expenditures are less likely to terminate public hospitals. The probability of
terminating a public hospital increases by 0.38% for every 1% decrease in a local
government’s budget for health and hospital services. The importance of community
commitment is consistent with research findings on the viability of safety net
providers. For example, in Los Angeles and Boston, local community together with
local officials’ effort successfully rescued public hospitals from being closed or
privatized (Norton and Lipson, 1998).

The variable of ethnicity index measures the homogeneity of local populations. As
mentioned before, a more homogeneous population composition is likely to have more
stable preferences for public services and would thus be less likely to terminate them
once they are provided. This speculation is not supported by the results. Table 3
reveals that a higher level of population homogeneity is actually associated with a
higher probability of termination. Holding other factors constant, a 1% increase in the
homogeneity of the local population increases the chances of termination by 2.4%.
This finding is consistent with the contracting out literature, where researchers found
that local governments in homogeneous communities are more likely to take
advantage of the private production mode (Ferris, 1986). Since all communities in this
study are dominated by Caucasians, the effect of the ethnicity index indicates that a
higher percentage of Caucasian population increases the termination probability. In

79
other words, a highly percentage of minority populations decreases the chances of
termination. This is probably explained by the fact that the minorities typically have a
higher demand for public health services.

Two variables for interest group influence are found to be significant. The proportion
of unionized public sector workers in the state is negatively associated with
termination. This variable captures changes in the extent of organization, and
presumably the power, of public employees over time. As expected, increases in this
proportion are associated with fewer terminations. A one percent increase in the
number of unionized workers in the public sector reduces the probability of
termination by 0.44%. This suggests that with increasing organization, public
employees are able to exert more influence on local decision makers and resist efforts
to reduce the number of public jobs.

The proportion of the poor, which captures the effect of the beneficiary group, is also
found to be significant. As expected, increases in the proportion of the poor reduce
the likelihood of termination. A one percent increase in the proportion of the poor
people in local communities reduces the probability of termination by 0.12%. This
effect is consistent with the policy termination literature that the beneficiary group is
an important force in the anti-termination coalition that can effectively block the
termination decision. While generally speaking the poor only have a weak influence
on political decisions, the huge stake in public hospital services can mobilize them to

80
fight fiercely when faced with the termination possibility (Behn, 1978), and the larger
the size of the beneficiary group, the more powerful the influence they have.

The final significant impact on termination was the number of private hospital beds.
The effect was not as predicted. The presence of a large private hospital market is
expected to allow local decision makers to more readily consider the termination of
public hospitals. Thus a positive relationship was expected. Instead, a negative
association is found; larger private hospital markets are associated with fewer
terminations. It may be that private providers are not interested in taking on indigent
care in the current low-reimbursement environment. They may thus lobby against the
closure of public hospitals, fearing the impact of more uncompensated care on their
financial performance.

Finally, government structure does not have a statistically significant impact. This
suggests there are no significant differences in the behavior of districts and counties in
these termination decisions. In this policy context at least, the particular institutional
structure within which the decision-making took place was not constraining.

To summarize, the results provide support for fiscal stress as the major triggering
event for termination decisions. Lower growth rates in state and local revenues are
associated with the termination of local public hospitals. In addition, local
governments operating in communities with a strong demonstrated commitment to
providing health services, with less homogeneous populations, with a higher

81
proportion of unionized public employees, a higher proportion of the poor, and those
with large private hospital markets are less likely to stop providing public hospital
services.

Summary

This chapter has empirically tested the 2-stage decision-making model of policy
termination developed in previous chapters, using the data of local public hospitals in
California. The results provide support for the importance of both stages of the model
– as triggering events and characteristics of the decision-making context both affected
the decision to terminate.

The specific factors found to affect the termination of local hospitals were state and
local fiscal conditions, the size of the local health budget, the homogeneity of local
populations, the proportion of unionized public employees, the proportion of the poor,
and the size of the private hospital sector. The results suggest that – at least in the case
of public hospitals – financial and political explanations dominate efficiency and
ideological explanations of policy termination.

The importance of fiscal stress in triggering policy termination is consistent with
general expectations of system inertia. A shock to the system is necessary to prompt
the reconsideration of a policy. The insignificance of the other two types of
hypothesized triggers – policy performance and ideology – is however interesting.

82
The apparent lack of an association between policy performance and termination
raises important governance issues. If policy termination is primarily triggered by
financial pressure, then the resulting terminations may not improve public sector
efficiency. Termination itself is costly and if efficiency is not considered in
determining which policies are terminated, it is unlikely that alternative solutions to
the underlying problems will be cost-effective ones, and both efficient and inefficient
programs may be cut. However, this result should be treated as preliminary because
the two different types of termination are combined together in the analysis:
conversion and closure. A deeper exploration is needed to uncover if any important
effect has been concealed.

The insignificance of ideological forces is surprising given the widely held belief that
changes in ideological preferences about the size of government has driven many of
the worldwide changes in governance. It may be that the indicators do not fully
capture the effect of ideological influences, or that ideology is less important in this
service area. Alternatively, the role of ideology may have been overemphasized in the
literature. More empirical work is needed to better assess its role.

83
Chapter Five: Distinguishing the Termination Form -- Hospital
Conversion and Closure

To this point, I have explored the effect of model determinants on hospital
termination, without considering the form of the change. Based on the model in
chapter 4, this chapter further examines how those factors are connected to different
termination forms by making the distinction between public hospital conversion and
closure. The first section briefly discusses the differences between hospital
conversion and closure. The second part examines the factors that influence local
government to take the conversion form rather than closure. Finally a multinomial
procedure is performed with the California public hospital data to test the relationship
between termination forms and their associated determinants, and results and findings
are discussed.

5.1. Hospital Conversion versus Closure

When a local government is faced with a termination decision, it has the choice of
taking the form of a conversion or a closure. A conversion refers to the action in
which a local government stops delivering hospital services by selling the hospital to a
private buyer, who is expected to resume the responsibility of service provision. A
closure, however, doesn’t involve such a market transaction and ownership transfer,
but simply shuts down all facilities. Although conversion and closure of public

84
hospitals are both deemed as policy termination, they are likely to generate different
impacts on the local community. For example, a closure is deemed as a more radical
action than a conversion, because the former results in a complete stop of health
services, while the latter at least maintains access to some services, despite the change
in ownership.

Given the differences between these two termination forms, they are likely to be
connected with different sets of factors. For example, a closure is likely to be
associated with certain conditions that overcome local officials’ hesitation for such a
radical policy change. Among them, a possible one is that the financial performance
of a public hospital is so low that the burden on the government budget is unbearably
heavy and finding private buyers is difficult, thus making a closure decision more
justified. On the other hand, a conversion is more likely to be associated with an
environment with good economic resources that attracts more potential buyers.
Therefore, to better understand local government’s termination behavior, it is
important to make distinction between these two termination forms. Indeed, some
scholars have criticized that simple binary choice models of combining conversion and
closure could oversimplify the decision options (Alexander et al, 1996). They argued
that once the decision to change is made, there is still an issue of what type of change
should be undertaken, and particular conditions distinguish the types of change a
hospital will adopt. In other words, different types of changes are connected to
different factors; to lump them together is likely to mask the effect of particular
factors.

85

Following the same line, this chapter continues to explore the effects of termination
determinants on different termination forms, with a focus on identifying the conditions
favoring a conversion rather than a closure. The next section develops hypotheses
related to such distinguishing conditions based on the contracting out theory and
health research literature.

5.2. Hypotheses

Although both conversion and closure are commonly practiced in public hospital
termination, little is known about why a local government chooses one form rather
than the other. Studies on hospital conversion and closure decision have appeared in
health research (McKay&Coventry, 1993; Alexander et al., 1996), but their
contribution to policy termination literature is constrained by a couple of factors.
First, the approach they adopt is mainly an organizational one. Little effort has been
put into identifying factors, especially ideological and political ones that are
distinctive in government decisions-making. Second, the empirical testing mixes
public hospitals with private ones, which may underestimate the effect of some factors
while overestimating others in an otherwise purely public hospital setting. To
overcome these problems, this study considers the conceptual framework developed
by the contracting out approach.

86
Contracting out theory has been widely applied to explain government’s decisions of
transferring service delivery to the private and nonprofit sectors in the past few
decades (Ferris, 1986; Ferris and Graddy, 1991). Despite that contracting out differs
from termination by retaining the financing responsibility and exerting substantial
influence on many issues after the decision, several features make the contracting out
models particularly useful in analyzing local government’s termination decision of
conversion versus closure. First, the setting to which the contracting out theory
applies -- government’s decisions on service delivery-- is exactly the same as this
study. Second, factors typically found important in the government contracting out
decision such as political, economy, service characteristics and community
characteristics are also key determinants in my model. Furthermore, the model has
most frequently been used in local decision making, which is the focal point of this
study. Most importantly, much similarity exists between a contracting out decision
and a conversion one. Both cases involve a transfer of service delivery from
government to another party, typically a private provider, thus making it a two-sided
decision. Such a two-sided characteristic is the fundamental difference between a
conversion and a closure, because the latter is just a one-sided decision, involving
government only. Therefore, by relating to contracting out theory and examining the
demand side of the decision, we may be able to identify important factors that favor a
conversion rather than closure when local government are faced with a termination
decision.

87
The contracting out theory places an important emphasis on the demand side in the
government’s contracting out process. Government supplies public services to be
transferred to other sectors in the face of a re-organization. For the contracting
process to proceed smoothly, however, there must be a demand side, i.e., one or more
private providers to take the bid. Therefore, the availability of potential private
providers and their willingness to join the process is important in determining the
contracting out decision. Evidence from empirical studies supports that the more
providers there are, the more likely government is to make the contracting out decision
(Ferris and Graddy, 1986).

Following the same argument, the availability and interest of potential hospital buyers
is expected to affect local government’s choice of the termination form. For a
conversion decision of a public hospital to take place, there must be potential private
purchasers available. The lack of availability of private purchasers constitutes an
important constraint on the conversion decision. In this study, the availability of
potential private providers is indicated by the development of a private service market.
A more developed private market contains more service providers potentially
available to take the bid, and thus is likely to favor a decision of conversion in relation
to closure. On the other hand, the health service market in California is extremely
competitive. A developed private market means more intense competition, and thus
may deter potential buyers from taking the bid and reduce the likelihood of
conversion. Therefore, this study only considers the important effect but not the
expected direction of the private health services market:

88
H1: The development of the private health service market affects a public
hospital’s likelihood of being converted rather than closed.

The likelihood of conversion is also affected by the financial performance of the target
public hospital, which indicates the profit prospect of the transaction. Profit is one of
the most critical motivations to explain the behavior of the private sector. When a
private hospital takes over a failing public hospital, it expects to make profit through
reform efforts made after the transaction. It doesn’t make much sense for a private
buyer to purchase a hospital when there is no hope or it is too difficult to make it
profitable even after such restructuring efforts. Thus, the financial performance of the
target hospital affects potential buyers’ interest in taking the bit and the feasibility of a
conversion decision. Empirical studies showed that hospitals in the most difficult
financial condition appear to have been unable to find acquirers (Sloan, 2003). Given
the already low profit margin in the California hospital market and the even worse
financial situation of public hospitals, it is expected that the potential buyers will be
more interested in taking over those relatively efficient public hospitals rather than
those performing most poorly. Therefore, I expect that when faced with a termination
consideration, an efficient public hospital is more likely to connect with a conversion
choice rather than closure. Thus, it is predicted:
H2: Public hospitals with better financial performance are more likely to be
converted than closed.

89
Finally, the choice of termination form also depends on the wealth of the local
community. The local community is the survival niche for a private organization. In
the public hospital conversion scenario, the local community defines the post-
conversion market for a potential buyer, because after a public hospital is converted,
its development opportunity relies more on the resource pool in the local community.
Therefore, before taking over a public hospital, a private buyer is expected to examine
the wealth of the community. Communities with higher per capita incomes contain
more potential payment sources for alternative services after the conversion. Thus, all
things being equal, a targeted public hospital located in a wealthier community signals
a better post-conversion survival prospect for potential buyers and favors a conversion
decision. Empirical research on general hospital conversion demonstrates that
conversion is more likely to take place in wealthier communities, where resources are
more abundant (Alexander et al, 1996). Therefore, I expect that:
H3: Public hospitals are more likely to be converted in wealthier communities.

The effects of each of the above three factors are summarized in the following chart.
As illustrated in Figure 5, government is the supply side of a termination decision.
The chances it considers a termination decision is affected by both the triggering
factors and decision-making factors, as elaborated in previous chapters. The specific
form a termination will take, either closure or conversion, however, is affected by the
demand side, i.e., the availability and interest of potential buyers. The presence of
potential buyers favors a conversion decision, while the lack of this condition will lead
to closure. The factors influencing the demand side include the private service market,

hospital performance and community resources. The signs inside the parentheses
indicate the expected directions on the likelihood of conversion.

Supply Side
(government)

Termination
possibility
--Triggering factors
--Decision-making
contextual factors

Yes No
--Private service market (?)
--Hospital performance (+)
--community resources (+)
Conversion Closure
Demand Side
(availability and interest
of potential buyers)

Figure 5: Factors Distinguishing Hospital Closure and Conversion
90

91
5.3. Empirical Testing

Methodology

The same dataset on California public hospitals is used to test how different factors
affect local government choice of the termination form. The same set of independent
variables is included in the model.

The dependent variable in this section is the form of termination local government
chooses to stop its operation of a local public hospital in a given year. The specific
form includes conversion and closure. In addition to the comparison group, no
change, there are three possible types of events in the dependent variable. Thus, the
new dependent variable, denoted TerminationMul, is a multiple categorical variable.
It takes a value of 0 when a public hospital remains open, 1 when it is converted and 2
when it is closed. Since the dependent variable is a three-level one, a multinomial
logit procedure is considered.

As discussed in the previous chapter, the dataset used in this study contains repeated
measures, as hospitals are clustered within counties. Such repeated measures, if not
controlled, could introduce bias to the results. Therefore, I use a multinomial logit
with a generalized estimating equations (GEE) approach. I estimate model parameters
with the mutlilog function in SUDAAN, which is currently the only available software

92

that combines the features of multinomial and GEE
4
. To ensure the difference in
results is not due to the software difference (STATA in chapter 4 and SUDAAN here),
I compare the results in SUDAAN and those produced by the mlogit command in
STATA, which doesn’t control the repeated measure problem. The coefficients and
significant variables in both programs are the same, showing that the estimation is
robust. The results also reveal that the standard errors in SUDAAN are usually larger
than those in STATA, suggesting that SUDAAN program does control the possible
inter-cluster correlation (repeated measures) bias and makes the estimates more
accurate.

Results and Findings

Since the two indicators of hospital performance, total profit margin and operating
margins, are considered separately, the results of the multinomial procedure are
presented in table 4 and table 5 respectively. I report the coefficient and the standard
error of each independent variable. The p-value for statistical significance of each
coefficient is included in parentheses. Statistically significant relationships at a 0.1
level are shaded.

4
The xtgee command in STATA is only able to perform a binary but not a multinomial logit procedure.

93
Table4: Multinomial Logit with GEE Approach on the Decision to Terminate Public
Hospitals (Total Profit Margin):

Conversion Vs Open Closure Vs Open Conversion Vs Closure
Coefficient Std Error (p) Coefficien Std Error (p) Coefficien Std Error (p)
Constant
27.28 9.81 (0.008) 2.13 12.67 (0.87) 25.09 16.33 (0.13)
Triggering Events

FedFinChg
0.08 0.14 (0.567) -0.02 0.23 (0.919) 0.1 0.30 (0.756)
StateFinChg
-0.16 0.08 (0.037) -0.01 0.11 (0.964) -0.15 0.14 (0.299)
LocalFinChg
-0.06 0.03 (0.066) -0.04 0.02 (0.052) -0.02 0.04 (0.594)
TotalMargin
0.88 2.38 (0.712) -3.16 1.32 (0.021) 5.7 3.70 (0.130)
PresIdeology
-0.01 0.03 (0.839) 0.04 0.04 (0.235) -0.05 0.05 (0.299)
StateIdeology
-0.02 0.02 (0.415) -0.01 0.03 (0.821) -0.01 0.05 (0.829)
Government Structure

District
-0.43 0.82 (0.605) 1.43 0.95 (0.138) -1.71 1.28 (0.187)
Policy Characteristics

Beds
0 0.00 (0.751) -0.01 0.01 (0.554) 0.01 0.01 (0.660)
OccRate
-0.01 0.02 (0.745) 0 0.02 (0.922) 0 0.03 (0.900)
PrivBedRate
-1.18 0.37 (0.003) 0.47 0.27 (0.093) -1.63 0.48 (0.001)
Interest Groups

Physicians
0 0.00 (0.449) 0 0.01 (0.778) 0.01 0.02 (0.671)
Union
-0.59 0.22 (0.010) -0.12 0.31 (0.697) -0.47 0.39 (0.235)
Poverty
-0.05 0.08 (0.521) -0.06 0.08 (0.481) 0 0.13 (0.991)
Community Characteristics

HealthBudget
-0.43 0.10 (0.0001) -0.07 0.07 (0.365) -0.36 0.12 (0.004)
PoliticalPhil
8.5 6.37 (0.189) -2.35 7.14 (0.743) 10.85 0.15 (0.291)
Ethnity
1.49 1.81 (0.417) -0.68 2.40 (0.777) 2.17 3.78 (0.569)
PCIncome
0 0.00 (0.876) 0 0.00 (0.800) 0 0.00 (0.980)

Number of obs 1163
Chi Sq 13189
P-value 0.0000

Table 5: Multinomial Logit with GEE Approach on the Decision to Terminate Public
Hospitals (Operating Profit Margin):

Conversion Vs Open Closure Vs Open Conversion Vs Closure
Coefficient Std Error (p) Coefficien Std Error (p) Coefficien Std Error (p)
Constant
28.07 9.79 (0.006) 1.55 11.74 (0.90) 26.3 14.50 (0.08)
Triggering Events

FedFinChg
0.08 0.14 (0.579) -0.01 0.21 (0.977) 0.12 0.28 (0.673)
StateFinChg
-0.16 0.07 (0.034) -0.01 0.10 (0.911) -0.15 0.13 (0.236)
LocalFinChg
-0.07 0.03 (0.064) -0.03 0.02 (0.066) -0.03 0.04 (0.397)
OperMargin
2.06 1.66 (0.220) -2.19 1.10 (0.052) 3.68 1.87 (0.055)
PresIdeology
-0.01 0.03 (0.702) 0.04 0.03 (0.222) -0.05 0.04 (0.264)

94
Table 5, continued

StateIdeology
-0.02 0.02 (0.421) -0.01 0.03 (0.643) -0.01 0.04 (0.893)
Government Structure

District
0.01 0.78 (0.987) 0.73 1.09 (0.508) -0.77 1.32 (0.561)
Policy Characteristics

Beds
0 0.00 (0.668) -0.01 0.01 (0.527) 0 0.01 (0.684)
OccRate
-0.01 0.02 (0.676) 0 0.02 (0.971) -0.01 0.03 (0.812)
PrivBedRate
-1.14 0.35 (0.003) 0.44 0.29 (0.142) -1.61 0.49 (0.002)
Interest Groups

Physicians
0 0.00 (0.494) 0 0.01 (0.730) 0.01 0.01 (0.614)
Union
-0.62 0.22 (0.009) -0.08 0.28 (0.784) -0.54 0.35 (0.126)
Poverty
-0.04 0.08 (0.589) -0.06 0.08 (0.464) 0.01 0.12 (0.964)
Community Characteristics

HealthBudget
-0.44 0.10 (0.0001) -0.05 0.07 (0.513) -0.39 0.12 (0.002)
PoliticalPhil
8.93 6.24 (0.159) -4.52 6.86 (0.514) 13.44 9.75 (0.175)
Ethnicity
1.52 1.75 (0.390) -0.42 2.49 (0.866) 1.95 3.56 (0.588)
PCIncome
0 0.00 (0.691) 0 0.00 (0.746) 0 0.00 (0.653)

Number of obs 1163
Chi Sq 19294
P-value .0000

The model overall provides a good fit for the data, as the chi-square value is
significant at the .0001 level in both estimations
5
. With respect to the effects of
independent variables, there is no big difference between the total-profit-margin and
operating-profit-margin models regarding the conversion versus open and closure
versus open decision. When coming to the conversion versus closure decision,
however, a significant effect is detected in the operating profit margin but not the total
profit margin. Therefore, I will not differentiate between these two measures of
hospital performance in the discussion except comparing the conversion versus
closure decision.

5
Due to the software difference, the chi-square values are not comparable to those in the binary model.

95
Conversion versus Open Decision

Regarding the factors affecting a conversion versus open decision, the multinomial
model reveals a similar pattern as that identified in the binary model in the previous
chapter. The results confirm the important role of both the triggering events and
decision-making context factors. The variables of state and local government fiscal
conditions are both statistically significant at a 0.1 level. The direction of the impact
is as predicted: Increases in the growth rates of state and local revenues reduce the
chances of hospital conversion. Such a finding is consistent with the existing
termination literature that a fiscal problem is an important trigger of policy
termination.

As expected, the proportion of unionized public sector workers in the state again is
found to be significant in the conversion decision. The negative sign shows that the
increase in the proportion of unionized public workers reduces the chances of
conversion. This suggests that with better organization public employees are able to
exert influence on the local termination decision, thus the interest group explanation is
supported in at least the conversion decision of local public hospitals.

Similarly, the increase in the proportion of the local budget spent on health activities
significantly reduces the likelihood of a public hospital’s conversion. This provides
evidence that the community’s commitment to providing health services exerts an
important impact on the conversion decision.

96

The variable of private hospital beds reveals a significant but negative effect. The
results show that more private hospital beds is associated with fewer conversions,
which suggests that the intensive competition in the California hospital service market
actually deters potential buyers from taking the bit and reduces the likelihood of
conversion.

Finally, similar to the effects in the binary model, neither the hospital’s financial status
nor ideological factors or government structure has a statistically significant impact on
the conversion decision.

Closure versus Open Decision

When it comes to the closure versus open decision, however, the results reveal a
different story. The negative effect of local financial condition is still statistically
significant at a 0.1 level, which is consistent with the argument that better financial
situation of local government decreases the need of termination. The significant effect
of state fiscal conditions in conversion vanishes in the closure decision. Such a
change may indicate that for closure, a more radical course, the decision is more of a
local choice, decided by the local government according to its own fiscal condition.
This is understandable because local government is the key decision maker of public
hospital services. Furthermore, hospitals to be closed are usually financially desperate

97
and the state government’s financial support for the local hospital is not sufficient for
a long-term solution.

On the other hand, the insignificant factor of public hospital financial performance in
the binary model is now found significant in the multinomial procedure, suggesting
that the important effect of performance failure as a triggering factor may have been
masked in the binary model. Both the total-profit-margin and operating-profit-margin
models in the multinomial procedure suggest that a public hospital’s financial
performance is negatively associated with its likelihood of being closed. Such a result
is consistent with general research on hospital closings that closed hospitals tend to be
less efficient in terms of operating margins, financial losses and costs (Cleverley,
1990). This indicates that a local government does evaluate a program’s efficiency
status before it resorts to the more radical decision of closure. Thus the performance
failure explanation of policy termination is supported at least in the public hospital
closure scenario. From another angle, the re-discovery of the significant effects of
performance indicators highlights the importance of distinguishing the termination
form in policy termination analysis.

Opposite to the effect of the performance measures, the significant effect of the
proportion of unionized public sector workers in the conversion scenario disappears in
the closure case. The insignificance of the union measure suggests that the effect of
interest group factors in local government’s termination decision is limited. Though
they can exert important influence in the conversion decision, their influence is greatly

98
reduced when faced with a more radical action such as closure. Considering the
significant effect of hospital performance in the closure case, the insignificant effects
of interest group factors suggest that economic consideration dominates interest group
consideration in the local hospital closure decision.

Similarly, the variable of the proportion of the local budget spent on health activities
also loses its significance in the closure case. Since this variable indicates the
community’s commitment to providing health services, the insignificance suggests a
limited role for local community’s commitment in preventing hospital closure.
Though it reduces the likelihood of local government’s moderate termination decision
such as conversion, it loses its effect in the more extreme decision such as closure. It
indicates that local officials with a strong commitment may try to fight against the
conversion decision, but under the desperate situation of local financial stress and
hospital financial loss, their commitment to local services does not enable them to
oppose the closure decision.

The effect of the private bed variable is also changed in the closure case. In the total-
profit-margin model, the relationship between the private bed variable and the
likelihood of closure is significant with a positive sign, which suggests that a more
developed private market increases the likelihood of public hospital closure. But the
effect is marginally significant, and such an effect disappears in the operating-profit-
margin model. Therefore, overall, the effect of the private service market seems to be
minimum in the public hospital closure scenario.

99

Finally, other insignificant factors in the binary model, such as the effect of federal
fiscal condition, ideology change, government structure and hospital characteristics,
are not found significant in the multinomial model, either. This may suggest that the
effects of these variables are not masked in the binary model, but indeed insignificant.

It should be noted that in both the conversion and closure scenarios, the significant
effects of two variables found in the binary model are not discovered in the
multinomial one: the proportion of people below the poverty level and the
homogeneity of the local community. However, we cannot easily conclude that these
factors are indeed unimportant, because the disappearance of their effects could be
caused by the small sample size. When we make the distinction between conversion
and closure, the case number for each comparison is reduced, thereby increasing the
standard error and jeopardizing the stability of the significance. Indeed, a careful
comparison between table 3 and table 4 and 5 reveals that the coefficients for these
two variables do not change much, while the standard errors in the multinomial
models are much larger than those in the binary model. Thus, the change of the effect
of the poor and of population homogeneity in the multinomial model is more likely to
be attributed to the small sample size of this study.

100
Conversion versus Closure Decision

A particular interest in this section is to examine the conditions that distinguish the
conversion and closure decision. Three factors stand out as significant in this regard:
operating profit margin, private bed number and the proportion of the local budget
spent on health activities.

It is hypothesized that more efficient hospitals is more likely to find potential buyers
and thus in favor of the conversion form. The direction of the significant effect of the
operation profit margin supports this hypothesis: better operating efficiency increases
the likelihood of hospital conversion rather than closure. Since operation profit
margin is a more accurate measure of the operation efficiency of the hospital (Harrison
and Montalvo, 2001), the significant effects of the operating profit margin but not the
total profit margin suggest that potential buyers are more concerned about how
efficiently the hospital is operating rather the overall financial status. A main reason is
that the operating efficiency defines the competency of the hospital to compete in the
private market after the conversion. This finding provides support to the argument
made by the contract theory: the conversion decision is not just a one-sided decision
that is made by the government; it is also decided by whether the private sector, a
buyer, wants to take the deal. Such a dual control illustrates the dilemma of public
hospital termination: termination usually is the result of poor performance, and
conversion carries the expectation of improving the efficiency of the failing hospital.
However, in order to be converted, it needs to meet a certain level of operation

101
efficiency. The implication of such an apparent paradox is that the conversion
decision should be planned before the hospital reaches a dead end; otherwise the
hospital will have no opportunity to be converted and have to be closed. To
government and the general public, this can be both bad news and good news. The
bad news is that efficient public hospitals also become the target of termination with
the form of conversion, which to some extent betrays the intention of using
termination as a tool for efficiency improvement. The good news is that despite its
new private ownership, more efficient hospitals are after all still functioning and
provide service access to the local community, thus the damaging effect of the
termination decision may be reduced, compared to a closure decision.

The effect of the private bed variable is interesting. Though the contracting out theory
predicts that a more developed private service market would favor conversion because
it would mean that more potential buyers are available, the negative sign of the private
bed variable suggests that a more developed private market actually decreases the
chances of conversion rather than closure. A possible reason is that a community with
more private beds means a more competitive environment. Given the relatively poor
performance of a public hospital and an extremely competitive hospital services
market in California, a more developed private market may actually deter potential
buyers from taking the bit. Such a finding may not necessarily contradict what the
contract theory predicts, but rather, it complements the contract theory by highlighting
the effect of the community environment on the hospital service market.

102
The negative effect of the proportion of the local budget spent on health activities
shows that local commitment reduces the chances of conversion compared to closure.
This at first glance may be a little counter-intuitive, because conversion after all is
better than a closure that provides no service at all. However, careful thought provides
a reasonable explanation: the strong commitment to public services provision is often
associated with a strong anti-privatization mentality. Thus a strong commitment can
work against conversion and end up with closure when a termination decision has to
be made.

While existing health research literature predicts that the wealth of the local
community helps attract potential buyers, the effect of the variable of income per
capita is not significant in this study. Such an insignificant effect is probably
connected to the distinctive characteristics of public hospitals. The potential buyers
may perceive that public hospitals tend to serve a large proportion of the poor, thus
making it difficult to take advantage of community resources after it is converted –
after all, lots of customers may still be the poor. However, it is also likely that income
per capital is not the best measure of the community resources. Thus more research
with better indicators is needed to assess the effect of community resources on the
likelihood of conversion.

To summarize, the multinomial results confirm the effect of both triggering factors
and decision-making context factors. The findings further reveal that different
termination forms are connected with different sets of factors. The conversion

103
decision is associated with a lower growth rate in state and local revenues, better
organization of the interest group of public employees, stronger local commitment for
health services, and fewer private beds. The closure decision, however, is affected
only by local fiscal conditions, the financial performance of the public hospital and
possibly the private service market. The state fiscal condition loses its impact on the
closure decision, suggesting a local control over the closure decision. The diminished
effect of the organization of the interest groups and local commitment suggested a
limited role of interest groups and community factors on the relatively more radical
action of closure. Such changes of the effects highlight the importance of the
termination form when examining government termination decisions. Similar to the
binary model, the ideological and government structure factors are not found to be
important in the conversion and closure decisions. Therefore, the overall pattern in the
multinomial model suggests that during local termination decisions, the fiscal and
efficiency considerations dominate the political and ideological ones.

An additional set of factors distinguishes the choice of conversion versus closure form.
Better operating efficiency favors a conversion decision, reminding us of the
important role of the willingness of potential buyers to engage in the termination
process. More private beds and a stronger local commitment to health services, on the
other hand, work against conversion and favor a closure decision. The effects of these
two variables highlight the impact of policy and community environments,
respectively. All together, these findings remind us of the importance of considering
termination forms when modeling the termination decision.

104
Chapter Six: Conclusion

6.1. Summary

This study has developed a 2-stage process model for the government termination
decision. Two empirical procedures have been carried out with California public
hospital data to examine the usefulness of the conceptual framework. The results
provide preliminary evidence that the conceptual framework is useful. The findings in
both the binary and multinomial procedures evidence the important role of triggering
factors including fiscal problems and perceived policy failures, and the decision-
making context factors including policy characteristics, interest groups and
community characteristics. More specifically, the binary procedure demonstrates that
the worsening state and local fiscal conditions increase the likelihood of termination,
while decision makers in communities with a stronger service commitment, a lower
level of population homogeneity, better organized public employees, a larger size of
the beneficiary group, and a more developed private service market decreases the
chances of a termination decision.

The multinomial procedure makes a distinction between the termination forms of
conversion and closure. The results reveal that different termination forms are
associated with different sets of factors. The conversion decision is affected by state
and local fiscal conditions, the organization of anti-termination interest groups, local
commitment for the target services and private service market. The closure decision,

105
however, is affected by local fiscal conditions, the target program’s performance, and
possibly the private service market that is significant in the total profit model but not
the operating profit one. Thus, the empirical test provides substantial support to the
important role of fiscal problem on the local government termination decision. It also
provides evidence that efficiency is an important criterion for at least radical
termination form such as closure.

The multinomial results also suggest that some factors distinguish the choice of
conversion versus closure form. Better operating efficiency, a less developed private
hospital service market and a weaker community commitment favor a conversion
rather than a closure decision. These findings highlight the importance of the
termination form when examining government termination decisions. It also contains
important implications for governments that are undertaking termination processes.
For example, when faced with a termination possibility, local offices may need to start
an early search of potential buyers before the performance of the target program gets
so poor that only the choice of closure is left. They may also need to consider
providing extra incentive for potential buyers when the target hospital performs
extremely poorly or is located in a competitive environment; otherwise it is unlikely
for private providers to take up the hospital services.

The finding that differs the most from the literature is the role of ideological and
political factors. In both multivariate procedures, the effects of ideological variables
have not been found significant. In addition, the role of interest groups is suggested to

106
be limited, affecting only the conversion decision but not the closure one. Thus,
unlike previous studies, the empirical results in this study suggest that fiscal and
efficiency considerations dominate the ideological and political ones, at least in the
public hospital scenario. Researchers for a long time have been concerned with the
dominance of ideological and political considerations, which makes the reform effort a
political game and reduces the efficiency gain. The results in this study suggest that
such a concern may have been overstated.

6.2. Contribution

This study makes several important contributions to the termination literature and
substantially advances our understanding of policy termination. First, it represents one
of the first attempts to develop a process model of policy termination. While previous
models (deLeon, 1978; Kirkpatrick et al., 1999) implicitly treated termination as a
one-stage phenomenon, this study uniquely argues for a two-stage model. The
distinction made between the triggering factors and decision-making contextual
factors helps illustrate better the complex nature of the termination process. In fact,
the multiple factors, actors and forces participating in the termination process and the
different times they get involved, may well suggest a two-stage or even three-stage
model. Furthermore, the theoretical framework developed here includes a more
comprehensive set of factors such as governmental structure and community
environment that are often absent in previous research. Thus, it allows the empirical
exploration of a broader range of influences on the determinants of policy termination

107
decisions, and their relative importance. The empirical test of California hospital data
reveals that such a conceptual framework is useful in analyzing determinants of
termination decisions. In addition, this study integrates theories from different
theoretical perspectives, such as policy termination, decision-making, and organization
and interest group theory. Thus, it contributes to the development of policy
termination research by helping overcome the problem of lack of communication
between different fields, a factor inhibiting the proliferation of policy termination
research (Daniels, 2001).

Second, the local perspective of this study sheds insight on local termination decisions
that haven’t been widely studied. As the devolution and decentralization trend
continues in the US, more and more government decisions are taking place at the local
level. The local level is a potentially fruitful area for additional work on termination
decisions. The model and the empirical findings in this study help improve our
understanding of local government’s behavior and the impact of community factors on
the termination decisions.

Third, the empirical tests performed in this study strengthen the model testing effort
that is desperately needed in policy termination studies (Daniels, 1997). The
quantitative analysis method used in this study helps overcome the shortcomings of
the case study method that has long been criticized for its lack of hypothesis testing
and generalization power in policy termination literature (Behn, 1978). At the same
time, the multivariate analysis method employed enables us to better evaluate the

108
relative importance of each factor by placing them in the same model. Thus, the
empirical result contributes to a beneficial dialogue between financial, efficiency and
ideological/political models and helps foster consensus. More importantly, compared
to the random and fixed effect models that are commonly used for longitudinal
studies, the GEE approach adopted in this study produces more accurate estimates and
generates more valid conclusions. The results in this study thus provide a good point
of comparison for future empirical work that considers different sets of services and
decision makers – necessary steps if we are to enhance our understanding of the
circumstances under which public decision makers prove willing to terminate public
policies and programs.

Fourth, this study is the first one to make the distinction between different termination
forms in policy termination research. The findings show that different forms of
termination are associated with different sets of determinants. Thus, not only the
determinant factors are important, the form of termination also matters. With this
knowledge in mind, decision makers may be able to predict and prepare for challenges
when making certain forms of termination decisions. For example, given the
significant effect of interest groups on the conversion decision, local decision makers
may need to work harder to get their understanding and support when planning a
conversion decision. Considering that lower performance and a competitive service
market inhibit conversion, local government should make a better effort to search for
potential buyers and consider providing incentives for them when the target program

109
performs extremely poorly or is located in a very competitive environment.
Otherwise, it is difficult for the private sector to participate in the process.

Fifth, the model developed in this study is also the first attempt to systematically
examine hospital behavior from a public sector perspective. The testing of factors
such as government factors and interest groups, which are often neglected in the
private decision model, advances our understanding of hospital conversion and closure
in public settings. The fact that some factors often found important in private hospital
conversion or closure (Needleman, 1997; Mark, 1999; Sloan et al, 2003), such as bed
size and environment resources, were not found significant in this study illustrates the
big difference between public decision making and private decision making.
Furthermore, government fiscal conditions, interest groups, and community
commitments usually not tested for private hospital conversion or closure have been
found significant in this study. Therefore, results in this work contribute to the
development of health service research and serve as a bridge between the health field
and public administration.

6.3. Implications

The findings in this study contain important implications for researchers and
government practitioners. As mentioned earlier, due to the fiscal constraints and
increasing government responsibility, there is a worldwide trend to transform the
governance mode in modern society. Both researchers and government officials

110
expect that governments can identify and terminate inefficient and ineffective
programs so as to harvest the benefit of a more efficient, effective, and responsive
government. During the termination process, a frequently raised concern is that
ideological and political considerations will dominate the efficiency one. The results
in this study provide evidence to support that program performance is an important
criterion for local government in making the termination decision, at least in the
choice of hospital closure. Compared to the insignificant effect of the ideological
factors and the marginal effect of the political factors, findings in this study suggest
that in the local termination process, the fiscal and efficiency considerations are likely
to dominate the ideological and political ones. This is consistent with findings in local
government restructuring research that local governments are more concerned with
practical issues such as service quality and efficiency and less with ideology and
politics (Warner and Hebdon, 2001). While this finding is encouraging, it is not so
clear if other levels of government and programs receive similar kinds of efficiency
evaluation during the termination process. Therefore, an important research task in
the future is to compare the termination process at different levels of government, and
government officials need to try hard to implement the efficiency evaluation at every
level of government and in different kinds of programs.

On the other hand, while we don’t want to see that political considerations dominate
the performance criterion, neither do we want the voices of weak groups to be ignored,
especially those whose benefits will be affected by a disproportionate infringement. In
the case of hospital termination, though the public employees have a say in the

111
conversion decision, the marginal effect of the beneficiary group in the binary model
and its insignificant effect in the multinomial model raises the concern that their
interests may not be well represented in the termination process. Therefore, for
services that significantly affect the politically weak groups, government officials
should give a bigger weight to the opinions of those groups, or other community
groups or advocates need to speak out for them so that their interests can be well
protected.

This study also reveals that fiscal pressure is an important factor that promotes the
termination decision. While this motivation itself is not problematic, governments
need to make sure that the program cuts really save government money based on a
cost-benefit analysis and the cuts reflect service demands in the community. The
insignificance of the service demand in the model indicates that government doesn’t
make a distinction between programs with different service demands during the
termination process. Therefore, government may need to consider giving more weight
to those services with higher social demands so as to meet public expectations better.

Finally, this research reveals that different forms of termination are associated with
different sets of determinants, and certain conditions favor one form of termination
versus another. In other words, different termination forms are associated with
different social impacts and face different barriers. Thus, it is important for decision
makers to choose a certain termination from versus the other by analyzing the
constraints placed on them. If possible, they should explore more termination options

112
and evaluate their impacts and difficulties so that the termination process can be
undertaken more smoothly.

Considering the trend of increasing fiscal constraint and social expectations, it is
expected that government will continue the effort of improving performance.
Identifying and analyzing the factors associated with the termination decision helps us
to better understand the decision for termination and provides a foundation for
examining the effectiveness of these decisions, and thus the reform efforts as a whole.
It is the author’s hope that this study will help advance our understanding of the
termination process and help government officials to make more informed public
policy in relation to future termination decisions.

6.4. Limitations

Despite the contributions of this study, some limitations should be noted. First, one
must be cautious about the generalizability of the results in this study as it focuses on a
single function – hospital services. Public hospitals may not be broadly representative
of public programs. Since most of their services primarily benefit the poor, the
political dynamics during the public hospital termination process may differ
substantially from services that support broader and more powerful sectors of the
community. In addition, the performance of public hospitals is relatively easy to
measure in terms of dollar amounts (e.g., total profit margin and operating profit
margin), which makes it easier to apply the efficiency criterion and base the

113
termination decision on performance evaluation. It is doubtful, however, whether
governments can apply this kind of evaluation to other programs in which output and
performance are more difficult to determine. Thus, the significant effect of
performance failure in the termination decision may be unique to public hospitals,
compared to other public programs.

Second, the generalizability of the conclusions is also limited by the focus on a single
state. California differs from other states in many political, economic, and
institutional aspects, which may cause the behavior of California local governments to
be different. For example, the state’s local governments have been severely
constrained by property tax limitations. Unable to increase revenue with new taxation,
local government in California has to rely more on state transfers. Therefore, they
may be more responsive than their counterparts in other states to fiscal stress at higher
levels of government. At the same time, California has a more competitive healthcare
market, which may impact governments and private health service providers’ behavior
during the termination process. Under such an environment, for example, a more
developed private service market could actually reduce the likelihood of conversion.

Third, due to data availability, the number of termination decisions is limited in this
study. The small number of observations poses as another constraint on the validity of
the study, in particular the multinomial analysis, where the dependent variable is
separated into three levels. For example, while significant in the binary model, both
variables of the beneficiary group and community population homogeneity lose their

114
significance in the multinomial model, perhaps due to the small sample size.
Therefore, more quantitative research with larger sample sizes is needed in order to
draw more reliable conclusions. Given that the number of terminations occurring in
one place or service area tends to be small, future researchers should consider
compiling a more comprehensive dataset that combines termination cases from
different places and service areas.

Fourth, related to the issue of the small sample size, the time effect of the termination
decision is not well examined. It is possible that the termination rate varies from time
to time. An inclusion of year dummy variables can help separate the time effect and
provide more accurate estimations. Therefore, if conditions allow, future research
using longitudinal datasets should try to explore the effect of year variables; or, if the
exploration of all year dummies not possible, develop explicit criteria to divide the
time period into different comparable sessions.

Fifth, as mentioned earlier, the measures of several factors can be further improved.
For example, the ideological factor in this study is measured by the ideological
preferences of the President and the state government. Thus, they may not capture the
ideological influence of the congressional and professional preferences that are also
important in the termination process. Similarly, the measures for the influence of
interest groups and community resources are also worth further refinement. For
example, a better measure of the interest group for public hospitals might be the
proportion of uninsured people rather than the poor; community resources for the post-

115
conversion private market are more than what income per capita can capture.
Therefore, how to compile a more accurate index for each factor becomes an
important task lying ahead for future researchers. Before we have confidence in our
measures of these factors, we remain cautious about the conclusions drawn in this
study. In particular, more empirical studies with better measures are needed so as to
better compare the relative importance of the ideological and political factors versus
fiscal and efficiency ones.

Finally, the interaction effects of some variables are not tested in the model. For
example, it is possible that the poor and public employees (e. g., physicians) may work
together to fight against any termination decision, thus forming an interaction effect.
Also, the influence of physicians is affected by their organizational capability (e.g.,
union membership). Therefore, it may be good to include the interaction terms
between the poor and physicians, and between physicians and union membership.
Nevertheless, this study doesn’t include these interaction terms, mainly for the
following considerations: First, connected to the issue of small sample size, the
number of independent variables in this study is relatively large in relation to the
number of the observations in the dataset. Adding more independent variables will
pose a greater threat to the stability and reliability of the empirical results. Second, the
data about the uninsured and the union membership of physicians in each target
hospital are not available. The current measures are just proxies, casting further
doubts on the usefulness of creating interaction terms between the poor and
physicians, or between physicians and the union variable. After balancing the

116
potential pros and cons, I decided not to include interaction terms in the model. Future
researchers, however, are encouraged to test the interaction effects between different
variables, if the quality of their data allows them to do so.

117
Bibliography
Alexander, J. A., D’Aunno, T. A., Succi, M. J. (1996). “Determinants of Profound
Organizational Change: Choice of Conversion or Closure among Rural Hospitals.”
Journal of Health and Social Behavior; Vol. 37, Iss. 3, p. 238 (14 pages).

Altman, S. H., Brecher, C., Henderson, M. G., Thorpe, K. E. (1989). Competition and
Compassion: Conflicting Roles for Public Hospitals. An Arbor: Health Administration
Press.

American Hospital Association (AHA) (1980-1995). AHA Guide to the Health Care
Field. Chicago.

Andrulis, D. P., Shaw-Taylor, Y., Ginsberg, C., Martin, V. (1995). Urban Social
Health: Chartbook Profiling the Nation’s One Hundred Largest Cities. Washington,
DC: National Public Health and Hospital institute.

Andrulis, D. P., Acuff, K. L, Weiss, K. B., Anderson, R. J. (1996). “Public Hospitals
and Health Care Reform: Choices and Challenges.” American Journal of public
health,Vol. 86, Iss. 2, p. 162 (4 pages).

Andrulis, D. P. (1997). “The Public Sector in Health Care: Evolution or Dissolution?”
Health Affairs, Vol. 16, Iss. 4; p. 131 (10 pages).

Andrulis, D. P. and Duchon, L. M. (2005). Hospital Care in the 100 Largest Cities
and Their Suburbs, 1996-2002: Implications for the Future of the Hospital Safety Net
in Metropolitan America. SUNY Downstate Medical Center and Robert Wood
Johnson Foundation.

Bardach, E. (1976). “Policy Termination as a Political Process.” Policy Sciences, 7,
123-131.

Bazzoli, G. and Andes, S. (1995). “Consequences of Hospital Financial Distress.”
Hospital & Health Services Administration, 40:4, 472-495.

Behn, R. D. (1976). “Closing the Massachusetts public Training Schools.” Policy
Sciences, 7, 151-171.

Benedict, B. (1977). “San Mateo to keep Chope Hospital,” San Francisco Chronicle
(March 2), p.14.

Beneditto, R. (1995) “Poll: American People Want Action.” USA Today, September 1,
A4.

118
Berry, W. D., Ringquist, E. J., Fording, R. C., Hanson, R. L. (1998). "Measuring
Citizen and Government Ideology in the American States, 1960-93." American
Journal of Political Science 42:1, 327-348. Data retrieved October 14, 2003, from
http://www.icpsr.umich.edu, #1208.

Biller, R. P. (1976). “On Tolerating Policy and Organizational Termination: Some
Design Considerations.” Policy Sciences 7, 133-149.

Brooks, R. C. (2004). “Privatization of Government Services: an Overview and
Review of the Literature.” Journal of Public Budgeting, Accounting & Financial
Management, Vol. 16, Iss. 4, pg. 467.

Brown, R. E. (1981). Public Medicine in Crisis: Public Hospitals in California.
Institute of Governmental Studies, University of California, Berkeley.

Bureau of Health Professions (1940-1995). Area Resource File. U.S. Dept. of Health
and Human Services, Public Health Service, Health Resources and Services
Administration, Bureau of Health Professions, Office of Data Analysis and
Management, retrieved December 3, 2003, from http://www.icpsr.umich.edu, #1208.

Cameron, J. M. (1978). “Ideology and Policy Termination: Restructuring California’s
Mental Health System.” Public Policy, 38:4, 533-570.

California Association of Public Hospitals and Health System (CAPH), 2005.
California’s Essential Public Hospitals.
http://www.caph.org/CAPHPublicHospitalFactsNov2005.pdf

California Library Association (CLA) (2002).
http://www.cla-net.org/legislation/021219_update.php

Cleverley, W. O. (1990). “After the Fall: Reasons behind 1989 Hospital Closings.”
Healthcare Financial Management, p. 22-25.

County and City Data Book, 1992-2002. Lanham, MD: Bernan Press.

Chakravarty, S., Gaynor, M., Klepper, S., Vogt, W. B. (2005). “Does the Profit
Motive Make Jack Nimble? Ownership Form and the Evolution of the U.S. Hospital
Industry.” National Bureau of Economic Research, Working Paper 11705.

Cousineau, M., Graddy, E., and Tranquada, R. (2003). An Analysis of Alternative
Governance for the Los Angeles County Department of Health Services. Report to the
John Randolph and Dora Haynes Foundation.

Daniels, M. R., (1995). “Organization Terminal and Policy Continuation: Closing the
Oklahoma Public Training Schools.” Policy Sciences, 28, 301-316.

119

Daniels, M. R. (1997). Terminating Public Programs: An American Political
Paradox. M.E. Sharpe, New York.

Daniels, M. R. (2001). “Policy and Organizational Termination.” International
Journal of Public Administration 24:3, 249-262.

Dahl, R. A. (1982). Dilemmas of Pluralist Democracy: Autonomy vs. Control. New
Haven: Yale University Press.

deLeon, P. (1978). “A Theory of Policy Termination.” In J.V. May and A.B.
Wildavsky (eds.), The Policy Cycle. Beverly Hills: Sage, 279-300.

_____(1983). “Policy Evaluation and Program Termination.” Policy Studies Review
2:4, 631-647.

Densau, A. T., and Munger, M. C., (1986). “Legislators and Interest Groups: How
Unorganized Interests Get Represented.” American Political Science Review, 80 (1),
89-106.

Downs, G. W. and Larkey, P. D. (1986). “Business versus Government: an Invidious
Comparison?” in The Search for Government Efficiency. New York: Random House.

Dye, T. R., (1967). “Government Structure, Urban Environment, and Educational
Policy.” Midwest Journal of Political Science, 43: 213-239.

Dye, T. R., (1975). Understanding Public Policy, 2
nd
ed., N.J.: Prentice-Hall.

Eldenburg, L. and Krishnan, R. (2003). “Public versus private governance: A study of
incentives and operational performance.” Journal of Accounting & Economics 35:3,
377-404.

Erikson, R. S., Wright, G. C. Jr., McIver, J. P. (1989). “Political Parties, Public
Opinion, and State Policy in the United States.” The American Political Science
Review, Vol. 83, Iss. 3, p. 729 (22 pages)

Ferris, J. (1986). “The Decision to Contract out: An Empirical Analysis.” Urban
Affairs Quarterly, December, 289-311

Ferris, J. and Graddy, E. (1991). “Production Costs, Transaction Costs, and Local
Government Contractor Choice.” Economic Inquiry 29(July), 541-554.

_____(1999). “Structural Changes in the Hospital Industry, Charity Care, and the
Nonprofit Role in Health Care.” Nonprofit and Voluntary Sector Quarterly 28:1
(March), 18-31.

120

Florestano, P. S. (1985). “State-local Relationship, State Limitations on Local Fiscal
Authority.” in state and local government administration, edited by Jack Rabin and
Don Dodd, 1985. Marcel Dekker, Inc. New York and Basel.

Frantz, J. E. (1992) “Reviving and Revising a Termination Model.” Policy Sciences,
Vol. 25, Iss. 2; p. 175 (15 pages).

Frantz, J. E. (1997). “The high cost of policy termination.” International Journal of
public Administration, 20(12), 2097-2119.

_____(2002). “Political Resources for Policy Terminators.” Policy Studies Journal
30:1, 11-28.

Gaudin, Jean-Pierre (1998). “Modern Governance, Yesterday and Today: Some
Clarifications to be Gained from French Government Policies”. International Social
Science Journal (English edition), Vol. 50, Iss. 1; p. 47 (10 pages)

Ginsburg, P.B., 1996. “The RWJF Community Snapshots Study: Introduction and
Overview," Health Affairs: Vol. 15, Iss. 2; p. 7 (14 pages).

Gore, A. (1993). Creating a Government that Works better & Costs less: the Report of
the National Performance Review. Washington, D.C.: The Office of the Vice
President.

Greve, H. R. (1998). “Performance, Aspirations, and Risky Organizational Change.”
Administrative Science Quarterly, Vol.43, Iss. 1; pg. 58 (29 pages).

Grossman, G. M. and Helpman, E. (2001). Special interest politics. Massachusetts:
The MIT Press.

Guo, G. and Zhao, H. (2000). “Multilevel Modeling for Binary Data.” Annu. Rev.
Sociology 26, 441-462.

Harris, M. (1997). “Policy Termination, Uncovering the Ideological Dimension.”
International journal of public administration, 20(12), 2151-2175.

Harrison M. G. and Montalvo, C. C. (2002). “The financial health of California
hospitals: a looming crisis.” Health Affairs 21, 118-126.

Hart, L. G., Pirani, M. J., and Rosenblatt, R. A. (1991). “Causes and consequences of
rural small hospital closures from the perspectives of mayors.” Journal of Rural
Health. Summer, 7(3): 222-45

121
Hirsch, B. and MacPherson, D. Union Membership and Coverage Database (1983-
1995), retrieved December15, 2003, from http://www.unionstats.com/

Horton, N. J. and Lipsitz, S. R. (1999). “Review of Software to Fit Generalized
Estimating Equation Regression Models.” The American Statistician 53 (May), 160-
169.

Kagan, R., Maiuro, L. M., Schmittdiel, J., Yu, W. (2001). California’s Closed
Hospitals, 1995-2000. A Project of the Nicholas C. Petris Center on Health Care
Markets and Consumer Welfare, University of California, Berkeley, School of Public
Health. Available at: http://www.petris.org/Docs/ClosedHospitals.pdf

Kaufman, H. (1976). Are government organizations immortal? Brookings:
Washington, DC.

Kaufman, H. (1985). Time, Chance and Organizations: Natural Selection in a
Perilous Environment. N.J.: Chatham House

Kettl, D. F. (1997). “The Global Revolution in Public Management: Driving Themes
and Missing Links.” Journal of Policy Analysis and Management, Vol. 16., No. 3, pg.
446-462.

Kettl, D. F. (2002). The Transformation of governance: public administration for
twenty-first century America. Maryland: The Johns Hopkins University Press.

Kirkpatrick, S., Lester, J., and Peterson, M. (1999). “The Policy Termination Process:
A conceptual Framework and Application to Revenue Sharing.” Policy Studies Review
16:1, 209-236.

Kingdon, J. W. (1984). Agendas, Alternatives, and Public Policies. Boston: Little,
Brown.

Kovner, A., R. ed. (1990). Health Care Delivery in the United States. Springer
Publishing Company. New York.

Krahmann, E. (2003). “National, regional, and global governance: One phenomenon
or many?” Global Governance, Vol.9, Iss. 3, pg. 323.

Liang, Y. K. and Zeger, S. L. (1986). “Longitudinal Data Analysis Using Generalized
Linear Models.” Biometrika 73, 13-22.

Legnini, M., et al. (1999). The Privatization of Public Hospitals, Economic and Social
Research Institute, prepared for the Kaiser Family Foundation, January, 1999.

122
Levine, C. H. (1978). “Organizational decline and cutback management.” Public
Administration Review 38:4 (July/August), 316-325.

Levine, C. H., Rubin, I. S., Wolohojian, G. G. (1981). “Resource Scarcity and the
Reform Model: The Management of Retrenchment in Cincinnati and Oakland.” Public
Administration Review, Vol. 41, Iss. 6; p. 619 (11 pages)

Lewis, D. (2002). “The Politics of Agency Termination: Confronting the Myth of
Agency Immorality.” The Journal of Politics. 64:1, 89-107.

Lindblom, C. E. (1959). “The Science of ‘Muddling through.’” Public Administration
Review, 19, 79-88.

Little Hoover Commission (2000). Special Districts: Relics of the past or resources
for the future? State of California.

Lowery, D. (1987). “The Distribution of Tax Burdens in the American States: The
Determinants of Fiscal Incidence.” Western Political Quarterly, 40, 137-58.

MacManus, S. A. and Pammer, W. Jr. (1990). Cutbacks in the county: retrenchment in
rural villages, townships, and counties. Public Administration Quarterly, Vol. 14, Iss.
3; p. 302 (22 pages).

March, J. G. and Olsen, P. J. (1976). Ambiguity and Choice in Organizations. Bergen:
Universitetsforlaget.

May, J. V. and Wildavsky, A. B. ed. (1978). The Policy Cycle. Beverly Hills : Sage
Publications, c1978.

McFarland, A. S. (1992). “Interest Groups and the Policymaking Process: Sources of
Countervailing Power In America.” In The Politics of interests: Interest Groups
Transformed edited by Mark P. Petracca. Boulder: Westview Press.

McKay, N. L., Coventry, J. A. (1993). “Rural Hospital Closures: Determinants of
Conversion to an Alternative Health Care Facility.” Med Care, 31(2): 130-40.

Moe, T. and Caldwell, M. (1994) “The Institutional Foundations of Democratic
Government: A Comparison of Presidential and Parliamentary Systems.” Journal of
Institutional and Theoretical Economics, vol. 150, pp. 171-195.

National Association of Public Hospital (NAPC) (1994). Meeting the Needs of our
most Vulnerable Patients. Washington, DC.

Needleman, J., Chollet, D. J., Lamphere, J. A. (1997). “Hospital Conversion Trends.”
Health Affairs, Vol. 16, Iss. 2; p. 187 (9 pages)

123

Needleman, J., Lamphere, J. A., Chollet, D. J.(1999). “Uncompensated Care and
Hospital Conversions in Florida.” Health Affairs, Vol. 18, Iss. 4; p. 125 (9 pages)

Norton, E. C., Bieler, G. S., Ennett, S. T., Zarkin, G. A. (1996). “Analysis of
Prevention Program Effectiveness with Clustered Data Using Generalized Estimating
Equations.” Journal of Consulting and Clinical Psychology, 64:5, 919-926.

Norton A. S., Lipson, D. J. (1998). “Public Policy, Market Forces, and the Viability of
Safety Net Providers”. Urban Institute, URL:
http://www.urban.org/url.cfm?ID=308041

Office of Management and Budget (OMB). US GDP data, retrieved January 11, 2004,
from http://www.whitehouse.gov/omb/budget/fy2004/spreadsheets.html.

Olsen, M. (1965). The Logic of Collective Action. Cambridge, Mass.: Harvard
University Press.

Osborne, D. and Gaebler T. (1992). Reinventing Government: How the
Entrepreneurial Spirit is Transforming the Public Sector. Mass.: Addison-Wesley
Pub. Co.

Pfeffer, J, and Salancik G. R. (1978). The External Control of Organizations: A
Resource Dependence Perspective. New York: Harper and Row.

Rebovich, D. P. (1985). “Fiscal Stress in the American States,” in State and Local
Government Administration, edited by Jack Rabin and Don Dodd, 1985. Marcel
Dekker, Inc. New York and Basel.

Richardson, J. J. (1993). Pressure groups. Oxford University Pressure.

Roemer, R. and Shonick, W. (1980). Private Management of California County
Hospitals. Institute of Governmental Studies, University of California, Berkeley,
1980.

Savas, E. S. (1987). Privatization: the Key to better Government. N.J.: Chatham
House Publishers.

Segal, J. A., Timpone, R. J., and Howard, R. M. (2000). “Buyer Beware? Presidential
Influence through Supreme Court Appointments.” Political Research Quarterly.
53(3): 557-573. President ideology data retrieved April 3, 2004, from
http://psweb.sbs.ohio-state.edu/faculty/rtimpone/research/pressc_main.htm#pres

Scott R. W. (1995). Institutions and Organizations. Thousand Oaks: SAGE

124
Shanks, N. H. (1988). “The Changing roles of public hospital?” Dissertation,
university of Colorado at Denver Graduate School of Public Affaires. ADI, 51, no.
06B, (1988): 2805.

Singer, I., Kuzner, J., Fagnani, L. (2002). Hospitals and Health Systems. Results of the
2000 Annual NAPH Member Survey

Shonick, W. and Roemer, R. (1983). Public Hospitals under Private Management:
The California Experience. Institute of Governmental Studies, University of
California, Berkeley.

Siegel, B. (1996). “Public Hospitals--a Prescription for Survival,” Minority Health.

Sloan, F. A., Ostermann, J., Conover, C. J. (2003). “Antecedents of Hospital
Ownership Conversions, Mergers, and Closures.” Inquiry, Vol. 40. Iss. 1; pg 39.

Spetz, J., Seago, J. A., Mitchell, S. (1999). Changes in Hospital Ownership in
California. Public policy Institute of California.

State of California Office of State Controller. County Financial Data, Fiscal years
1984-1995. Retrieved November 17, 2003, from http://www.cicg.org/data/

_____ County Demographic and Financial Data, 1980-1995. Retrieved
November 17, 2003, from http://www.cicg.org/data/

State of California, Department of Finance, Race/Ethnic Population with Age and Sex
Detail, 1970–1989. Sacramento, CA, May 2004. Retrieved June 25, 2006, from
http://www.dof.ca.gov/HTML/DEMOGRAP/DRU_datafiles/Race/RaceData_70-
89.htm

State of California, Department of Finance, Race/Ethnic Population with Age and Sex
Detail, 1990–1999. Sacramento, CA, May 2004. Retrieved June 25, 2006, from
http://www.dof.ca.gov/HTML/DEMOGRAP/Data/RaceEthnic/Population-90-
99/RaceData_90-99.asp

State of California Office of State Controller. Financial Transactions Concerning
Counties of California: Annual Report (1980-1984). Sacramento.

State of California Office of State Controller. Financial Transactions concerning
Special Districts of California: Annual Report (1980-1995). Sacramento

State of California Office of Statewide Health Planning and Development. Hospital
data retrieved October 1, 2004, from http://www.oshpd.cahwnet.gov/

125
State of California Secretary of State. Statement of vote: General election / State of
California. (1980-1992). Sacramento: California State Printing Office

Stoker, G. (1998). “Governance as theory: Five propositions”. International Social
Science Journal. Mar 1998. Vol. 50, Iss. 1; p. 17(12 pages)

Stone (1982). “A revisionist View of Community Politics.” in Classics of Urban
Politics and Administration, edited by William J. Murin. Oak Park, Ill.: Moore Pub.
Co..

Tami, L. M. (1999). “Analysis of the Rationale for, and Consequences of, Nonprofit
and For-profit Ownership Conversions.” Health Services Research, Vol. 34, Iss. 1, p.
83 (19 pages).

Thorpe, K. E., Florence, C. S., Seiber E. E. (2000). “Hospital Conversions, Margins,
and the Provision of Uncompensated Care.” Health Affairs, Vol. 19, Iss. 6, p. 187.

Warner, M., and Hebdon, R. (2001). “Local government restructuring: privatization
and its alternatives.” Journal of policy analysis and management, vol. 20, no. 2, 315-
336

Weimer, D. L., Vining, A. R. (1991). Policy Analysis: Concepts and Practice. 2nd ed.
N.J.: Prentice Hall.

Wilson, G. K. (1990). Interest groups. Oxford: Basil Blackwell.

Wilson, J. Q. (1989). Bureaucracy: What Government Agencies Do and Why They Do
it? New York: Basic Books

Wilson, T. G. (2000). “Community Attitudes, Operational Indicators, and Hospital
Administration: Examining the Determinants Surrounding the Conversion of
Government-owned Rural Acute Care Facilities in Mississippi to Critical Access
Hospitals,” Mississippi State University; 0132, DAI, 61, no. 07A (2000): p. 2916

Wolf, C. Jr. (1979). “A Theory of Nonmarket Failure: Framework for Implementation
Analysis.” Journal of Law and Economics, P. 107-139.

Yinger, J. (1990) “States to the Rescue? Aid to Central Cities Under the New
Federalism.” Public Budgeting & Finance,Vol.10, Iss. 2, pg. 27.

Young, G. J, Desai, K. R. (1999). “Nonprofit Hospital Conversions and Community
Benefits: New Evidence from Three States.” Health Affairs, Vol. 18, Iss. 5; p. 146.

Zeger, S. L. and Liang, K. Y. (1986). “Longitudinal Data Analysis for Discrete and
Continuous Outcomes.” Biometrics 42, 121-130.

Appendix: Results of the multicollinarity test

Variable | VIF 1/VIF
-------------+----------------------
Union | 6.70 0.149239
PresIdeology | 4.47 0.223811
PCIncome | 3.21 0.311444
Poverty | 2.55 0.392474
Physicians | 2.53 0.395770
PrivBedRate | 2.23 0.449354
OperMargin | 2.21 0.453276
HealthBudgt | 2.19 0.456173
StateIdeology| 2.03 0.492657
Beds | 1.99 0.503106
TotalMargin | 1.92 0.521672
Ethnicity | 1.73 0.576424
District | 1.57 0.635297
OccRate | 1.53 0.653339
StateFinChg | 1.44 0.692152
PoliticalPhil| 1.19 0.843745
LocalFinChg | 1.08 0.928940
-------------+----------------------
Mean VIF | 2.39

| FedFinChg StateFinChg Local~g TotalMargin Oper~n PresIdeology State~y
-------------+---------------------------------------------------------------
FedFinChg | 1.0000
StateFinChg | 0.2996 1.0000
LocalFinChg | 0.0978 0.0605 1.0000
TotalMargin | -0.0433 -0.0854 0.0064 1.0000
OperMargin | -0.0344 -0.0746 0.0139 0.6207 1.0000
PresIdeology | -0.3906 -0.4819 -0.1670 0.1688 0.1367 1.0000
StateIdeology| 0.3571 -0.0132 -0.0644 -0.0606 -0.0478 -0.4236 1.0000
District | -0.0002 -0.0048 0.0096 -0.0293 0.3219 -0.0021 0.0057

126

Bed | -0.0237 -0.0049 0.0033 0.1836 -0.0025 0.0256 -0.0312
OccRate | -0.0221 -0.0339 -0.0024 0.2309 0.1030 0.0490 0.0050
Poverty | -0.3351 -0.2130 -0.1060 0.0217 0.1065 0.5096 -0.3166
PrivBedRate | 0.0716 0.0487 0.0351 0.0280 -0.1844 -0.1113 0.0841
Physicians | -0.0297 -0.0291 -0.0068 0.2310 0.0386 0.0558 -0.0247
Union | -0.3977 -0.3787 -0.0705 0.1224 0.0918 0.8446 -0.6214
HealthBudgt | 0.0930 -0.0738 -0.0772 0.0484 0.1859 0.0473 0.1509
PoliticalPhil| -0.1926 -0.0753 0.0124 0.0286 0.0832 0.1655 -0.1736
Ethnicity | 0.1287 0.0615 0.0118 -0.1315 0.0401 -0.1798 0.1599
PCIncome | -0.3656 -0.1542 -0.0498 0.1389 -0.0321 0.4980 -0.4642

| District Bed OccRate Poverty PrivBedRate Physicians Union
-------------+---------------------------------------------------------------
District | 1.0000
Bed | -0.3064 1.0000
OccRate | -0.3315 0.4308 1.0000
Poverty | 0.1026 -0.1420 -0.2223 1.0000
PrivBedRate | -0.2963 0.3343 0.2513 -0.2296 1.0000
Physicians | -0.3596 0.6696 0.4808 -0.2314 0.3261 1.0000
Union | -0.0079 0.0369 0.0318 0.5396 -0.1190 0.0518 1.0000
HealthBudgt | 0.1924 -0.2224 -0.0789 -0.0575 -0.6383 -0.2229 -0.0512
PoliticalPhil| 0.1721 -0.0990 -0.0159 0.0853 -0.1576 -0.1233 0.2244
Ethnicity | 0.1476 -0.4106 -0.2222 -0.2008 -0.3448 -0.3997 -0.2019
PCIncome | -0.1437 0.3355 0.2916 -0.0403 0.1273 0.4637 0.5963

| HealthBudgt Politi~l Ethnicity PCIncome
-------------+------------------------------------
HealthBudgt | 1.0000
PoliticalPhil| 0.1735 1.0000
Ethnicity | 0.4283 0.1898 1.0000
PCIncome | -0.1723 0.0379 -0.3177 1.0000

127

Abstract (if available)

Abstract The improvement of government performance requires that government carry out periodical reviews of its policies and terminate those that are inefficient and ineffective. The existing literature nevertheless reveals that the termination process has been severely understudied. The paucity of knowledge has left many important questions unanswered, thus limiting our capability to evaluate and guide government reforms. The consequence is most acutely perceived at a time when governments are struggling to do more with less and more terminations are expected to take place.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Civic associations, local governance and conflict prevention in Indonesia

PDF

Water-resources development in a decentralizing Indonesia: a challenge for policy design

PDF

Challenging migrant worker policies in Korea: settlement and local citizenship

PDF

The economic and political impacts of U.S. federal carbon emissions trading policy across households, sectors and states

PDF

Environmental policy reforms in a global and regional world: mimetic, diffusive, and coercive policy adoptions in France and Korea

PDF

Moving classical public radio into the new media future

PDF

A view from below: the development and role of organizational social capital in neighborhood regeneration in Los Angeles

PDF

The production of public spaces: design dialectics and pedagogy

PDF

Local ground-level ozone policies: Governance structure and institutions In achieving federal attainment

PDF

Using decision and risk analysis to assist in policy making about terrorism

PDF

Community foundations and new governance networks: three studies exploring the role of regionally-networked philanthropic organizations in local problem-solving

PDF

Cities on the environmental justice frontline: the intractability of hazards and the governability of amenities

PDF

Longitudinal effects of kindergarten policies and early childhood practices

PDF

How does collaborative governance work? The experience of collaborative community-building practices in Korea

PDF

Pathways of local economic development: tales of cities in the United States and South Korea

PDF

The institutional context of Korean philanthropy and the role of government and (quasi-) community foundations

PDF

A framework for good local governance: achieving prosperity in an increasingly complex environment

PDF

Satisfaction with local "public goods" and services: the effects of household income and privatization in southern California

PDF

The institutionalization of nonprofit management: emergence, development, and legitimization

PDF

The impact of public expenditures on health care on total health expenditures: an exploratory analysis of selected OECD countries

Asset Metadata

Creator Ye, Ke (author)

Core Title Policy termination: a conceptual framework and application to the local public hospital context

School School of Policy, Planning, and Development

Degree Doctor of Philosophy

Degree Program Planning / Public Administration

Defense Date 12/07/2006

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag closure,Conversion,OAI-PMH Harvest,policy termination,public hospital

Language English

Advisor Elizabeth, Graddy (committee chair), Melnick, Glenn (committee member), Sellers, Jefferey M. (committee member)

Creator Email kny@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m289

Unique identifier UC1221855

Identifier etd-Ye-20070222 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-159985 (legacy record id),usctheses-m289 (legacy record id)

Legacy Identifier etd-Ye-20070222.pdf

Dmrecord 159985

Document Type Dissertation

Rights Ye, Ke

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu