Close
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
The effect of rate regulation and price competition on labor efficiency in ancillary and nursing service areas of hospitals in California, 1983-1991
(USC Thesis Other)
The effect of rate regulation and price competition on labor efficiency in ancillary and nursing service areas of hospitals in California, 1983-1991
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter free, while others may be
from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely afreet reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zed) Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
THE EFFECT OF RATE REGULATION AND PRICE COMPETITION
ON LABOR EFFICIENCY IN ANCILLARY AND NURSING SERVICE AREAS
OF HOSPITALS IN CALIFORNIA
1983-1991
by
Suzanne Lucille Smith
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)
August 1996
Copyright 1996 Suzanne Lucille Smith
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 9705177
UMI Microform 9705177
Copyright 1996, by UMI Company. Ail rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES. CALIFORNIA 90007
This dissertation, written by
SUZANNE LUCILLE SMITH
under the direction of te. Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School in partial fulfillment of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
Dean of Graduate Studies
Date ...
4-COMMITTEE
j Ouirperson
........
%
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To My Fan Club:
Ma & Pa
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLED(34ENTS
The process of writing a dissertation requires
tremendous perseverance. I would like to thank the very
special people who had the perseverance to stick by me.
First, I would like to thank my Dissertation Committee : Dr.
Sampath Rajagopalan for providing me the feedback necessary
to keep the study on track; Dr. David Lopez-Lee for hashing
through the research methodology with me and for being
accessible for any methodological crisis; and especially my
Chairperson, Dr. Robert Myrtle, for patiently guiding me
through the maze of bureaucratic policies and degree
requirements during my entire doctoral experience.
I am also deeply grateful to my very special friend
Pascal Giraco. His wizardry with computers and his generous
nature were a constant comfort during my most trying
moments. He was indeed pivotal to the completion of this
degree.
Finally, I would like to thank the two people in my
life whose unconditional love and support have given me the
self-confidence to face life's challenges: my Ma and Pa. I
thank them both for keeping the financial engines burning
when the fuel was running low and for riding the emotional
roller coaster with me until we could all finally get off.
ill
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ABSTRACT
The purpose of this study was to examine the effects
of rate regulation and price competition on hospital labor
efficiency in 149 private, for-profit and non-profit
hospitals in California between 1983 and 1991. The study
hypothesized that hospitals with a high percent of Medicare
patients (rate regulation), Medi-Cal or managed care
patients (price competition) were more likely in areas of
high market competition, than in areas of low market
competition, to have had an increase in labor efficiency
overtime in their a) Ancillary Service Area and b) Nursing
Service Area.
The study used Data Envelopment Analysis (DEA) to
generate scores for labor efficiency for the Ancillary
Service Area and Nursing Service Area of each hospital.
Logistic regression tested the hypotheses, controlling for
ownership and membership in a multi-hospital system.
Interaction terms between market competition and payer
measured differences between areas of high and low market
competition.
The results of the study did not find that hospitals
in areas of high competition were more likely than
hospitals in areas of low competition to respond to rate
iv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
regulation and price competition by increasing labor
efficiency in their Ancillary or Nursing Service Areas
between 1983 and 1991. However, in both high and low areas
of market competition, hospitals that had an increase in
the percent of Medi-Cal or managed care patients (price
competition) were more likely, than hospitals that did not
have an increase in patients, to have had an increase in
labor efficiency in their Nursing Service Area between 1983
and 1991. In both areas of competition, hospitals that
joined a multi-hospital system were more likely, than
hospitals that did not change membership, to have had an
increase in efficiency in their Ancillary Service Area
between 1983 and 1991. Similarly, in both areas of
competition, for-profits were more likely than non-profits
to have had an increase in labor efficiency in their
Nursing Service Area between 1983 and 1987. The
implications for public policy are that hospitals do not
increase labor efficiency in response to rate regulation
but do respond to price competition regardless of market
competitiveness.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
ACKNOWLEDGMENTS .................................... iii
ABSTRACT ............................................iv
TABLES ............................................. ix
FIGURES ............................................ xi
I. PROBLEM STATEMENT ................................ 1
Reimbursement - .......................... 1
Hospital Inefficiency....................... 2
II. HISTORY TO THE PROBLEM............................8
The Era of the Great Depression.............. 8
The Era of Cost Containment .................15
The Era of Competition ..................... 23
Summary ................................... 28
III. THEORETICAL FRAMEWORK............................31
Models of Hospital Behavior .................31
Profit Maximization ....................32
Utility Maximization ...................33
Price Competition.......................... 36
Hospital Competition ....................... 37
Managed Care ...............................39
Rate Regulation............................ 40
Prospective Payment ........................ 41
Efficiency.................................44
Parametric Measures ....................45
Data Envelopment Analysis .............. 48
IV. LITERATURE REVIEW ............................... 52
Competition and Regulation ..................52
Hospital Labor............................. 61
Data Envelopment Analysis (DEA) ............. 64
Summary ................................... 73
V. HYPOTHESES ......................................78
VI. METHODOLOGY .....................................81
Study Sample ...............................81
Dependent Variable ......................... 82
Changes in DEA Labor Efficiency Scores ... 82
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table of Contents, Continued
DEA inputs and outputs ................. 84
Ancillary Service Area............ 84
Nursing Service Area.............. 85
Case-Mix Adjusted DEA Outputs ...........86
DEA Model Assumptions .......................88
Envelopment Surface ....................90
Input/Output Orientation ............... 91
Evaluation System......................92
Categorical Variables .................. 94
Discretionary/Non-Discretionary
Variables ............................. 96
Substitution Ratios ....................97
Model Degeneracy.......................98
Control Variables ......................... 100
Ownership ............................ 100
Multi-Hospital Membership ............. 101
Independent Variables ......................104
Market Competition ....................104
Rate Regulation .......................107
Price Competition .....................108
Interaction Terms ......................... 109
Logistic Regression Models ................. Ill
Model Variables and Coding............ 112
VII. RESULTS ....................................... 114
Regression Model Diagnostics ............... 114
DEA Scores ................................117
Summary .............................. 122
Regression Results ........................ 123
Goodness of Fit .......................124
Hypothesis I: ........................ 125
Hypothesis II: ........................129
Hypothesis III: .......................133
Control Variables .....................137
Summary .............................. 140
VIII. DISCUSSION....................................142
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table of Contents, Continued
IX. IMPLICATIONS FOR PUBLIC POLICY.................. 150
Rate Regulation ........................... 150
Price Competition......................... 151
Quality Care .............................. 152
X. STUDY LIMITATIONS ..............................155
Data Envelopment Analysis (DEA) ............ 155
Data ..................................... 155
Logistic Regression........................156
Case-Mix ..................................157
XI. FURTHER RESEARCH ...............................158
BIBLIOGRAPHY ....................................... 160
APPENDIX A: Market Competitors........................172
APPENDIX B: Logistic Regression Collinearity and
Correlation Diagnostics .................. 174
APPENDIX C: DEA Scores ..............................182
APPENDIX D: Percent Reduction of FTEs in Inefficient
Hospitals ...............................186
APPENDIX E: Logistic Regression Model Output .........189
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLES
Table 1:
Table 2:
Table 3:
Table 4:
Table 5:
Table 6:
Table 7 :
Table 8a:
Table 8b:
Table 9:
Table 10a:
Table 10b;
Table lia:
Table 11b:
Table 12a:
Table 12b:
Table 13a:
Percent Distribution of Study Hospitals .....81
Study Hospitals by Bed Size ................ 82
Logistic Regression Models ................ 112
Logistic Regression Model Variables
and Coding...............................113
Highly Correlated Terms of Regression
Models .................................. 115
DEA Efficiency Scores
Hospitals with a DEA Score of 1 ........... 117
Increase in DEA Efficiency Scores
Hospitals with an Increase in DEA Scores
During Time Periods ...................... 118
Ancillary Service Area
Average Percent Reduction of FTEs in
Inefficient Hospitals .....................121
Nursing Service Area
Average Percent Reduction of FTEs in
Inefficient Hospitals .....................121
Regression Models Goodness of Fit Scores .... 125
Ancillary-Regression Coefficients-Medicare ..128
Nursing-Regression Coefficients-Medicare ....128
Ancillary-Regression Coefficients-Medi-Cal ..132
Nursing-Regression Coefficients-Medi-Cal ....132
Ancillary-Regression Coefficients-HMOs .....136
Nursing-Regression Coefficients-HMOs .......136
Ancillary-Regression Coefficients
Control Variables ........................ 139
ix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Tables, Continued
Table 13b: Nursing-Regression Coefficients
Control Variables ....................... 139
Table 14: Total Average FTEs per Total Average
Discharge and Total Average Visit......... 143
Table A.l: Number of Competitors Within a 15 Mile
Radius of Hospital .......................173
Table B.l: Collinearity Tolerance Scores .............175
Table B.2a: Model A1 Correlation Coefficients ......... 176
Table B.2b: Model A2 Correlation Coefficients ......... 177
Table B.2c: Model A3 Correlation Coefficients ......... 178
Table B.3a: Model N1 Correlation Coefficients ......... 179
Table B.3b: Model N2 Correlation Coefficients ......... 180
Table B.3c: Model N3 Correlation Coefficients ......... 181
Table D.la: Ancillary 1983-87
Percent Reduction of FTEs in Inefficient
Hospitals by Percentile ..................187
Table D.lb: Ancillary 1987-91
Percent Reduction of FTEs in Inefficient
Hospitals by Percentile .................. 187
Table D.lc: Ancillary 1983-91
Percent Reduction of FTEs in Inefficient
Hospitals by Percentile .................. 187
Table D.2a: Nursing 1983-87
Percent Reduction of FTEs in Inefficient
Hospitals by Percentile .................. 188
Table D.2b: Nursing 1987-91
Percent Reduction of FTEs in Inefficient
Hospitals by Percentile .................. 188
Table D.2c: Nursing 1983-91
Percent Reduction of FTEs in Inefficient
Hospitals by Percentile .................. 188
X
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FIGURES
Figure 1: Average DEA Score-Ancillary.............. 183
Figure 2: Average DEA Score-Nursing................183
Figure 3: Average DEA Score-Ancillary-By Bed Size
Category................................184
Figure 4: Average DEA Score-Nursing-By Bed Size
Category................................184
Figure 5: Increase in DEA Score-Ancillary-Percent Out
of Total Hospitals With Increase in DEA
Score ...................................185
Figure 6: Increase in DEA Score-Nursing-Percent Out
of Total Hospitals With Increase in DEA
Score ...................................185
XX
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I. PROBLEM STATEMENT
The purpose of this study was to examine whether
hospitals in areas of high market competition were more
likely, than hospitals in areas of low market competition,
to respond to rate regulation and price competition by
increasing labor efficiency in their Ancillary Service Area
and Nursing Service Area between 1983 and 1991. The study
examined 149 non-teaching, private for-profit and non
profit hospitals with 95 to 500 beds in California.
Reimbursement
The escalation of health care expenditures overtime
has been associated with the lack of incentives on the part
of consumers to utilize services appropriately and on the
part of providers to produce services efficiently. Prior to
the 1980s, the lack of incentives was a function of cost-
based fee-for-service (FFS) reimbursement by third-party
payers. The presence of third-party payers in the health
care market has allowed consumers and providers to be less
sensitive to the rising prices of medical services.
Feldstein (1993) writes that "under a retrospective cost-
based reimbursement system any risks of higher operation
costs are shifted to the patients through their third-party
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
payers." Therefore, under cost-based FFS, the fee that
hospitals received for services would be based on costs or
charges which would allow the providers to pass on higher
costs due to inefficient operations. Since the
implementation of Medicare in 1965, total health care
expenditures have grown from 74.3 billion in 1970 to 884.2
billion in 1993. In addition, the percent of the GNP
devoted to health care has grown from 7.4 percent in 1970
to 13.9 percent in 1993. In 1993, acute care accounted for
61.3 percent of total expenditures (Levit, 1994) .
Hospital Inefficiency
The large amount of health care expenditures devoted
to hospital care may be due to inefficiencies in producing
hospital services (Rosko and Broyles, 1988). Efficiency
consists of 1) economic efficiency and 2) technical
efficiency. Economic efficiency refers to the least
expensive combination of inputs utilized in maximizing
output. Technical efficiency refers to the least number of
inputs utilized in maximizing output. Cost minimization
would be achieved when a firm is both economically and
technically efficient; or when the slope of the isoquant
line equals the slope of the isocost line (Dorfman, 1978).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Under economic theory, price sensitive firms within a
competitive environment would attempt to minimize the
quantity and cost of inputs to maximize profits given a
market driven price constraint. Therefore, in price
sensitive markets, price reflects efficiency. However,
given the presence of retrospective cost-based third-party
reimbursement, the health care market would lack a market
driven price constraint. Rate regulation and price
competition have been two approaches to impose price
constraints on the market.
The Medicare DRG based Prospective Payment System
(PPS), implemented in 1983, was one means to impose price
constraints on hospitals. Under PPS, hospitals would
receive a predetermined set fee for Medicare patients based
on the Diagnostic Related Group (DRG) in which the patient
was coded. However, Rosko and Broyles point out that:
Although an extensive body of empirical evidence
suggests that state mandated prospective payment
systems have constrained the growth of hospital
expenditures, there is less evidence documenting
how the savings were achieved. Thus, it is not
clear whether prospective payment has evoked
desirable behavioral responses, such as increased
efficiency, or less desirable reactions, such as
cost shifting or reducing the quality of care.
(Rosko and Broyles, 1988)
As Rosko and Broyles suggest, it is not clear if hospitals
have responded to prospective payment by reducing the cost
of inputs, the number of inputs or both.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Similarly, the effects of price competition on
hospital economic and technical efficiency has not been
well documented. Competition is a function of the number of
hospitals, physicians and payers within a defined market
area. Studies have found that the effect of competition on
hospital costs depends on whether hospitals are located in
a physician-dominated market or a payer-dominated market.
In a physician-dominated market, hospitals compete on a
service basis to attract physicians. Competing on a
service basis contributes to duplicative services within a
defined market and ultimately excessive health care
expenditures.
In a payer-dominated market, health plans control the
beneficiary's choice of hospitals and physicians
(Zwanziger, 1987; Porter, 1980). Managed care plans are
health insurance plans that restrict the beneficiary's
choice of hospitals and physicians. Managed care plans
guarantee care to an enrollee in exchange for a fixed,
monthly premium (capitation). Given the capitated premium,
managed care plans have the incentive to contract with
providers that can provide quality care at the least
expense. Hospitals are expected to compete for managed
care contracts by offering lower prices. The higher the
concentration of managed care plans within a market area.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the greater the incentive of the hospital to maximize both
economic and technical efficiency to minimize cost in order
to keep prices competitive (Enthoven, 1989) .
In 1982, California implemented legislation that would
allow both state and private payers to selectively contract
with hospitals and physicians. By 1986, 60 percent of
California beneficiaries were enrolled in managed care
plans and by 1990, 80 percent were enrolled (Zwanziger,
1994b). Therefore, since 1982, the California health care
market has become a payer-dominated market.
A number of studies of California hospitals have found
that hospitals have responded to Medicare PPS (rate
regulation) and public and private managed care contracting
(price competition) by reducing total hospital expenditures
and the rate of hospital expenditures (Phibbs and Robinson,
1989, 1993; Luft et al., 1986; Zwanziger et al., 1988,
1989, 1994a and 1994b) . For example, Zwanziger et al.
(1994a and 1994b) found that in California both PPS and
managed care were correlated with the overall decline in
total costs between 1983 and 1990. Moreover, hospitals in
areas of high competition had a greater decline in
expenditures than hospitals in areas of low competition.
The decline appeared to be a function of lower cost per
unit of service, on average. However, none of the studies
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
identified whether the decline in expenditures was due to
lower cost of inputs (economic efficiency) or the use of
fewer inputs (technical efficiency) or both (cost
minimization).
Under rate regulation and price competition, profits
would be more difficult to maintain given the presence of
price constraints. Consequently, hospitals in areas of high
competition would need to find ways to reduce the cost and
volume of inputs in order to offer competitive prices while
maintaining profit. Therefore, hospitals in areas of high
competition would be more likely, than hospitals in areas
of low competition, to increase economic and technical
efficiency.
Although one purpose of health care public policy is
to contain the growth of health care expenditures, it is
also the purpose of public policy to ensure that the public
does not suffer from lower quality of care as a result of
those policies. Therefore, from a public policy
perspective, it is important to know if lower rates of
health care expenditures under rate regulation and price
competition are a function of lower cost per input or use
of fewer inputs.
Given that hospital labor is a significant input in
the production of hospital ancillary and nursing services.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
hospitals may respond to rate regulation and price
competition by reducing their use of labor full-time-
equivalents (FTEs) . A reduction in labor FTEs may imply
greater technical efficiency in the use of labor inputs to
produce inpatient discharges and outpatient visits.
However, it may also suggest greater risk of lower quality
care. Therefore, this study will examine if hospitals in
California's payer-dominated markets have responded to
Medicare PPS (rate regulation) and to public and private
managed care contracting (price competition) by increasing
labor efficiency in their Ancillary Service Area and
Nursing Service Area.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
II. HISTORY TO THE PROBLEM
Health care legislation implemented by Congress
throughout history has assisted in solidifying a health
care system based on poor incentives for efficiency.
Beginning with the Great Depression, Congress encouraged
the utilization of medical services by passing legislation
that would encourage the spread of public and employer-
based cost-based fee-for-service (FFS) health insurance as
well as the excess supply of health care providers. By the
1970s, health care expenditures were rising at an alarming
rate. By the end of the 1970s, state regulatory agencies
had failed to contain rising expenditures since regulatory
agencies did not change the incentives inherent in cost-
based FFS reimbursement. By the 1980s, legislators turned
to rate regulation of Medicare reimbursement and price
competition of the non-Medicare population to induce
hospitals to improve economic and technical efficiency in
producing medical services.
The Era of the Great Depression
Access to Care
The availability of health insurance induces demand
for medical care. The availability of health insurance has
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
historically been encouraged by labor, insurers, the
medical profession and the federal government. Up until the
Depression in 1935, the role of the federal government in
the provision of health care was limited to assisting
states with programs including grants for venereal disease
control, quarantine services, child and maternal hygiene
(Starr, 1982). During the Depression, the role of the
federal government became more prominent in the area of
health care with the enactment of the Social Security Act
and the National Labor Relations Act of 1935. The Social
Security Act expanded assistance to state public health
programs and to the age, blind and destitute families
(Starr, 1982).
The growth of employer-based health insurance was
encouraged by organized labor and the federal government
under the National Labor Relations Act which stated that
wages and conditions of employment were subject to
collective bargaining. However, it was not until a Supreme
Court ruling in 1948, involving Inland Steel, that the law
was interpreted to include "benefits of health and welfare"
as a "condition of employment" (Wilkinson, 1989; Renn,
1988). In essence, the legislation enacted by the federal
government facilitated public access to medical care which
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
set into motion the spiraling inefficient utilization of
medical services by both consumers and providers.
Employer-Based Health Insurance
Employer-based health insurance was further encouraged
by the federal government with the favorable tax treatment
for employers and employees under the Internal Revenue Code
of 1954. Under this code, employer contributions to health
insurance premiums on behalf of employees was exempt from
income tax. Thus, the federal government encouraged the
expansion of employer-sponsored health insurance thereby
encouraging demand for medical services and subsequently
expenditures (Starr, 1982). By 1986, over 90 percent of
the population under 65 were covered under private health
insurance for inpatient and outpatient services (Wilkinson,
1989; Koch, 1988) .
Physician and Hospital Sponsored Health Insurance
Physicians and hospitals have also had a vested
interest in the availability of health insurance to
consumers. During the Great Depression, hospitals formed
Blue Cross to contract with individuals or employers to
provide services for a set monthly premium. Similarly,
physicians formed Blue Shield. In order for Commercial
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
plans to compete with the Blues for market share, they were
forced to relax cost containment efforts as well as expand
their benefit offerings. The physician sponsored plans
gained position in the market place through endorsement by
the American Medical Association (AMA) . Moreover, the non
profit status of both Blue Cross and Blue Shield provided
financial advantages which allowed them to compete with
for-profit plans. Moreover, the competitive advantage of
the Blues gave them stronger bargaining power with
employers and providers. In the years to follow, physicians
and hospitals continued to influence the payment structure
of health care by forming the American Medical Association
(AMA) and the American Hospital Association (AHA) which
provided extensive lobbying efforts regarding regulation of
provider reimbursement (Greenberg, 1991).
Thus, the competition between the Blues and Commercial
insurers encouraged expansion of cost-based fee-for-service
(FFS) reimbursed medical benefits. Consequently, FFS
reimbursement exacerbated provider insensitivity to rising
cost of medical services. Moreover, rising costs reflected
insensitivity to inefficient production of medical
services.
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
MEDICARE
The health insurance system that evolved out of the
Great Depression depended primarily on employer-based
coverage which excluded those who were not employed,
primarily the elderly. There was little incentive on the
part of insurers to offer insurance coverage to the elderly
since they had a higher demand for medical care and
consequently, could drive up the cost of insurance
premiums. On July 30, 1965, Congress amended the Social
Security Act (PL 89-97) to establish the federal insurance
program Medicare (Title XVIII) for those over 65 and the
federal state insurance program Medicaid (Title XIX) for
the poor. To gain acceptance by the medical community, the
legislation adopted retrospective cost-based fee-for-
service reimbursement for providers (Starr, 1982). Between
Medicare, Medicaid and employer-based insurance, the
majority of the population had access to medical services.
Thus, the U.S. Health Care System that evolved out of the
Great Depression was based on a reimbursement system that
induced demand and utilization among consumers and
providers without incentive to minimize cost or maximize
efficiency.
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Federally Funded Hospital Construction
The enactment of the National Labor Relations Act and
the Social Security Act of 1935, and subsequent Medicare
amendments in 1965, contributed to the demand for medical
services. Consequently, further government intervention
was increasingly necessary to create the supply of
providers and medical technology to meet the growing demand
for services. The scientific and technological advancements
of World War II were applied to the health care sector.
Government encouraged these advancements by establishing
the National Institutes of Health (Starr, 1982).
In addition. Congress enacted the Hospital Survey and
Construction Act, or Hill-Burton Act in 1946, which was a
federal aid program to states to subsidize the construction
of hospital beds. Over the next 15 years, hospital
construction expenditures increased at an annual rate of
12.5 percent between 1946 and 1960 (Zwanziger, 1987). More
hospitals would mean more patients needed to fill their
beds which in turn would require more labor and non-labor
inputs. Given that hospitals were reimbursed on a cost
basis for each service rendered, hospitals had little
incentive to restrict utilization of services.
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Federally Funded Medical Education
In addition to technological advances and hospital
construction, the supply of medical providers was also
encouraged through government programs. The Health
Amendments Act of 1956, subsidized training for public
health personnel and nursing education. Until the mid
1960s, the American Medical Association, through its
control over licensing, had controlled the rate of entry of
physicians into the medical school and provider market
place. The restrictions by the AMA over the supply of
physicians created a shortage which the federal government
attempted to counter by enacting the Health Professions
Educational Assistance Act in 1963. The Act authorized
matching grants for the construction of medical schools and
loans for medical education (Greenberg, 1991).
In 1966, the Allied Health Professions Personnel
Training Act was also implemented to increase the supply of
allied personnel. The final act to encourage supply of
health manpower was the Comprehensive Health Manpower
Training Act and the Nurse Training Act of 1971. However,
by the mid-1970s, the over supply of health manpower was
becoming evident, and in subsequent years Congress enacted
legislation to retract aid to medical schools and by 1983,
federal support for health training had been significantly
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
minimized (Wilkinson, 1989; Sorkin, 1986; Wilson and
Neiihauser, 1982) .
The ramifications of government intervention in the
supply of health manpower was that, by 1980, the health
care industry began to experience an excess supply of
providers (Wilkinson, 1989; Wilson and Neuhauser, 1982) . An
over supply of providers induced further demand for medical
services. Under cost-based fee-for-service reimbursement,
hospitals had limited incentive to develop efficient ways
of producing medical services since they could pass on
excess costs to third-party payers. In the decades to
follow, policy makers began to consider approaches for
containing the escalation of health care expenditures.
The Era of Cost Containment
Regulation
Between 1960 and 1980, total health care expenditures
rose from 5.3 percent to 9.5 percent of the GNP (Zwanziger,
1987) . By the mid-1970s, it was clear that the rate of
health care expenditures had to be brought under control.
The 1970s began an era of cost containment that raised
debate over the basic structure of the health care system.
Nonetheless, government created more regulatory policies to
contain rising health care expenditures. The expenditures
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
were the outcome of previous government policies which had
encouraged expansion of health insurance and the excess
supply of providers.
The rational for government regulation may be based on
two different behavioral models of regulatory agencies. The
traditional model is the Public Interest Model that assumes
regulatory agencies are "established to protect the
consumer from the abuses of big business and to provide
consumer protection in those cases where consumers are
unable to judge the quality of the product they are
purchasing" (Feldstein, 1993). Under this model, one of the
major objectives of regulation is to ensure that consumer
prices under monopolized industries are similar to prices
that would be paid in competitive industries.
In direct contrast to this model, is the Economic
Theory of Regulation that assumes "that regulation enables
what is or would be a competitive industry to act as though
it were in fact a monopoly" (Feldstein, 1993). Under this
model, it is assumed that industries, attempting to
maximize consumer prices, seek government regulation to
bare entry from other competitors. It is further
hypothesized that industries often succeed because they
provide political support to legislators who in turn enact
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the legislation that benefits the industry. In addition,
Feldstein writes that:
The bureaucrats responsible for the regulatory
agency may, in addition, have their own goals, such
as to increase the size and authority of their
agency, thereby justifying higher salaries. To
increase its budget, the agency must receive
support from the legislative subcommittee having
control over its jurisdiction. (Feldstein, 1993)
Thus, if the outcome of government regulation is barring
market entry from competitors, then the economic theory of
regulation may explain why regulatory approaches to health
care cost containment have not succeeded at inducing
hospitals to produce services with greater technical and
economic efficiency.
Wage Freezes
Freezing wages was another regulatory approach for
containing the rise in health care expenditures. The
Economic Stabilization Act of 1971 froze wages and prices
for the health care industry for three years, including the
rate of increase in physician fees. However, once the
controls were lifted, prices began to rapidly rise (Starr,
1982) . Consequently, this regulatory approach would not
have had lasting effects on inducing hospitals to produce
services more efficiently.
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Health Planning
Health planning was a significant regulatory attempt
at bringing the rate of health care expenditures under
control. Feldstein writes that:
The Health Planning Act of 1974 required states to
inclement Certificate of Need (CON) programs which
required hospitals, nursing homes, HMOs and free
standing surgery centers to obtain prior state
approval for capital expenditures and acquisition
of expensive medical equipment. However, a number
of studies showed that CON was not successful at
slowing the rate of health care expenditures.
(Feldstein, 1993)
In Feldstein's (1993) review of the literature, he found
that some studies indicated that regulatory agencies were
"influenced" by providers because they had to rely on
providers for data and information. In addition, studies
indicated that the "influence" of providers on agencies
reduced entry by more efficient competitors, including HMOs
and free-standing facilities. Moreover, the greater the
competition, the more likely states would require CON. In
1986, Congress refused funding for CON programs although
over half of the states continued to require it. Thus, the
studies would suggest that CON was not effective at
reducing the rate of health care expenditures because CON
actually functioned to bare low cost competitors from
entering the market. Thus, in the absence of more efficient
competitors, hospitals would have less incentive to find
more efficient ways of producing medical services.
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Professional Standards Review Organizations (PSROs)
Another regulatory approach to promote efficiency in
the health care industry was the establishment of PSROs.
Under the 1972 Social Security Amendments, PSROs were
established to oversee quality of care by curtailing
unnecessary utilization. Although quality was proposed as
an objective, the organizations focused mostly on the
appropriateness of admissions and length-of-stay. However,
a study by the Congressional Budget Office concluded that
the cost of the program was about equal to its savings.
Moreover, several studies showed that there was no real
overall effect on hospital utilization or admissions
(Feldstein, 1993).
The PSROs were composed of panels of local physicians.
The PSROs were expected to regulate the local physicians
who were also the same physicians represented on the boards
of the PSROs and who were to develop the standards that the
PSROs were to enforce. In 1984, PSROs were replaced with
another regulatory approach called the Professional Review
Organization (PROs) . PROs were responsible for reviewing
appropriateness of admissions. Although there is no clear
evidence to date, there is speculation that the work of the
PROs have been partly responsible for the decline in the
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
number of admissions since the implementation of Medicare
PPS (Feldstein, 1993).
Partly to blame for the failure of health planning
agencies to bring the rate of health care expenditures
under control was that under the legislation "no changes
were required in the method of hospital reimbursement and
provided no new incentives to change patient or physician
behavior" (Feldstein, 1993). Without incentives to change
behavior, hospitals would continue to operate less
efficiently then they might under different methods of
reimbursement.
State Rate Regulation
During the 1970s a number of states in the Northeast
attempted to contain the rate of health care expenditures
by setting hospital reimbursement rates. Some states set
rates based on charges and others by per diem. In some
states, such as New Jersey, participation in the state
rate-setting program was mandatory for all hospitals and
payers. In other states, participation by payers was
voluntary. The weight of the evidence indicated that
mandatory all-payer participation was more effective at
slowing the rate of increase in health expenditures,
compared to non-mandatory states, since hospitals were not
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
able to shift costs to other payers (Rosko and Broyles,
1988) . The evidence indicated that rate regulation
restrained the rates of increase in cost per admission,
cost per day and costs per capita. However, studies also
indicated an increase in the number of admissions.
Moreover, there was less evidence regarding whether the
cost of care was being shifted to non-hospital settings
(Rosko and Broyles, 1988).
Given that hospitals responded to state rate
regulation by raising prices to other payers, rate
regulation may not induce hospitals to operate more
efficiently unless participation in the rate setting
program is mandatory for all payers. However, if a large
percentage of the non-Medicare population is enrolled in
managed care plans, then the market may produce the same
results as an all-payer PPS system.
New Jersey DRG Prospective Payment
New Jersey was one of the states that chose to
implement an all-payer system of rate regulation. In 1975,
the state implemented the Standard Hospital Accounting and
Rate Evaluation (SHARE) program that regulated a cost-based
per diem reimbursement rate. Toward the end of the 1970s,
studies revealed that patient days were a poor measure of
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
output because variations in resource use were not captured
(Rosko and Broyles, 1988). To improve upon the all-payer
system, in 1980 New Jersey received support from the Health
Care Financing Administration to adopt DRGs as a method for
determining hospital reimbursement rates since DRGs were
expected to better reflect case-mix. Broyles (1990)
compared the effects of the SHARE reimbursement system with
the DRG system and concluded that "the all-payer DRG system
reduced average annual increases in the cost per admission,
the cost per day, length-of-stay but increased the number
of admissions per hospital." Despite the increase in the
number of admissions, the results of the New Jersey all
payer DRG based prospective payment system indicated that
the rate of health care spending per capita could be
brought under control. Consequently, DRG based prospective
payment was soon adopted by the Medicare program with
similar expectations.
Given that the rate of hospital expenditures declined
in spite of the rise in the number of admissions, would
suggest that hospitals responded to the all-payer DRG based
prospective payment system by increasing efficiency.
However, the studies did not indicate whether this
increased efficiency was a function of lower cost per
inputs or use of fewer inputs.
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The Era of Con^etition
Managed Care
Toward the end of the 1970s, policy makers began to
consider competitive strategies to contain the rising costs
of health care. More specifically, "advocates of
competition argued that regulatory interventions have not
constrained significant increases in health care costs and
that centralized regulations provide incentives for health
care providers to behave in socially undesirable ways"
(Rosko and Broyles, 1988). Socially undesirable ways
included barring market entry from more efficient
providers.
Some of the strongest arguments for injecting
competition into the health care industry has been offered
by Alain Enthoven, Professor of Economics at Standard
University. His theory of Managed Competition Is based on
economic principles that would design the incentive
structure of the health care industry to look more like the
incentive structure of a competitive market place. One of
the requirements of his theory would be the presence of
markets dominated by competing managed care plans. Managed
care plans would be reimbursed a fixed fee per month, per
member (capitation), which would create an incentive to
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
selectively contract with providers who could offer quality
care at the least expense. In effect, managed care plans
would compete on a price basis thereby injecting price
competition into the health care market place. This would
in turn create the incentives for hospitals to compete on
the basis of price rather than on the basis of inefficient
service offerings (Enthoven, 1988). In order to offer
competitive prices to managed care plans, hospitals would
have an incentive to find ways to produce services with
greater economic and technical efficiency.
In 1973, Congress passed the Health Maintenance
Organizations (HMO) Act to encourage the growth of managed
care plans through subsidies, loans and legal mandates. In
order to reduce barriers to HMO entry into the market
place, employers were required to offer employees an HMO
option if available in the area. However, development of
HMOs during the 1970s was slow due to provisions under the
Act which made it more difficult for HMOs to compete. The
provisions included mandatory annual open enrollment,
community versus experience rating, and a broad range of
benefit offerings (Feldstein, 1993). The HMO Act was
amended in 1976 and 1977 to relax the benefit requirements
thereby allowing HMOs to design more competitive benefit
packages. In 1979, Congress loosened the CON requirements
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
for HMOs to assist them in overcoming barriers to market
entry from existing providers (Greenberg, 1991). Studies
indicate that HMOs are successful at reducing expenditures
between 10 to 40 percent primarily through fewer hospital
admissions and shorter lengths-of-stay (LOS) (Manning,
1984; Luft, 1978) . With fewer admissions and shorter LOS,
hospitals would have incentive to reduce the cost and
number of labor and non-labor inputs thereby producing
services with greater economic and technical efficiency.
Cost Sharing
Cost sharing is another concept included in
Enthoven*s theory of Managed Competition. Cost sharing
creates the incentive among consumers to be cost conscious
in their selection of health plans. The practice of cost-
sharing has been adopted by both public and private payers.
The RAND Health Insurance Experiment examined the effects
of coinsurance on the utilization of medical services
(Manning et al., 1987). The study concluded that "as the
coinsurance rate rose, overall use and expenditures fell,
for adults and children combined" (Feldstein, 1993). The
degree of cost sharing that a plan requires of consumers is
one method that health plans use to compete for market
share. Thus, cost sharing has been a competitive means to
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
inject the industry with incentives to find ways to produce
services with greater economic and technical efficiency.
Competitive Bidding
In 1982, California enacted legislation that would
allow the Medicaid program and private insurers to
selectively contract with hospitals and physicians. In this
way, payers could encourage subscribers to enroll in
managed care plans that would agree to accept negotiated
fees. By 1986, more than 60 percent of California
beneficiaries were enrolled in either an HMO or PPG and by
1990, 80 percent of California beneficiaries were enrolled
(Zwanziger, 1994b). Zwanziger and Melnick (1988) examined
the effects of price competition in California between 1980
and 1985. They found that "the rate of increase in cost
per discharge of hospitals in highly competitive markets
was 3.5 percent less than the rate of increase of hospitals
in less competitive markets" (Zwanziger and Melnick, 1988).
In addition, Melnick et al. (1992) examined hospital prices
charged to California Blue Cross between 1980 and 1987 and
found that Blue Cross was able to negotiate lower prices
from hospitals located in more competitive markets.
Although the effects of price competition in
California have been insightful, the rest of the country
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
would require a greater degree of HMO market penetration to
produce similar results. In 1990, the HMO penetration among
Medicare and non-Medicare enrollees across the country was
32 percent in California and less than 10 percent in half
of all states (Feldstein, 1993). Feldstein (1993) writes
that "it is too early to judge how successful market
competition will be in reducing the rise in expenditures.
The degree of market competition varies greatly from region
to region." Therefore, if competition is to be effective
at containing the rate of health care expenditures
throughout the United States, then more regions will have
to consist of markets that are dominated by managed care
plans. Given capitated reimbursement, managed care plans
have the incentive to selectively contract with hospitals
who can produce quality care through greater economic and
technical efficiency.
Anti-Trust
Historically, medical and hospital societies have been
able to inhibit competition by encouraging providers to
boycott insurance plans that attempt to enforce strict cost
containment efforts or undercut reimbursement rates. In
1975, the Supreme Court ruled that the "learned
professions" were not exempt from anti-trust laws, which
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
would include the medical profession. This ruling limited
the ability of the medical associations to engage in
practices that would discourage competition through price
fixing, boycotts, mergers, and discouragement of
advertising. In the years following this Supreme Court
decision, actions brought by the Federal Trade commission
have undermined the "industry-supported" barriers to
competition (Greenberg, 1991). Thus, if competition is to
be an effective public policy approach to containing health
care expenditures by promoting efficiency among providers,
then barriers to competition must be minimized through
anti-trust legislation.
Summary
The rate of growth in health care expenditures since
the Great Depression has been the result of the dynamics
between the growing availability of public and private
health insurance and a retrospective cost-based method of
reimbursement. The result of these dynamics has been
inefficient utilization of medical services. The regulatory
attempts by the federal government to promote efficiency
and quality through Health Planning Agencies, CON and PSROs
were appropriate in theory but in reality functioned to
only further the status quo by protecting the position of
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
current incumbents. Moreover, without restructuring the
method of reimbursement to induce consumers and providers
to behave more cost consciously, the regulatory programs
were destined to be ineffective at slowing the rate of
health expenditures and encouraging efficient methods of
production.
The implementation of the Medicare DRG based
Prospective Payment System (PPS), in 1983, was an attempt
to correct the incentive flaws inherent in retrospective
cost-based reimbursement. In addition, enforcing anti-trust
laws to promote price competition among private capitated
health plans h^s been an attempt to induce providers to
curtail utilization of inappropriate care and adopt more
efficient ways of delivering services. Similarly,
increased cost sharing has been an attempt to induce
consumers to be more responsible in their utilization of
medical services.
Although fate regulation and price competition have
shown promise at slowing the rate of health expenditures to
varying degrees, it is important to understand how these
results are being accomplished and whether they are being
accomplished at the expense of quality. Given that hospital
labor is a significant input in the production of hospital
ancillary and nursing services, hospitals may respond to
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
rate regulation and price competition by increasing their
efficient use of labor FTEs. Technical efficiency is the
utilization of the least number of inputs for a given level
of output. Reducing the quantity of labor inputs to produce
a given level of inpatient discharges and outpatient visits
would suggest greater technical efficiency. Thus, this
study will examine the effects of rate regulation and price
competition on changes in the efficient use of labor FTEs
in the Ancillary Service Area and Nursing Service Area of
California hospitals between 1983 and 1991.
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
III. THEORETICAL FRAMEWORK
Models of Hospital Behavior
How a hospital responds to rate regulation or price
competition is a function of the hospital's objectives.
Under the theory of price competition, it is assumed that
the objective of the firm is to maximize profits. However,
given the non-profit status of hospitals, their objectives
may be other than profit maximization. Rosko and Broyles
(1988) write that "there is no consensus as to a theory of
behavior for health care organizations." Moreover, "each
model was developed to represent a particular phenomenon
that occurred in the health care field" (Rosko and Broyles,
1988). Economic models of hospital behavior consist of two
components: 1) an objective function which are the goals of
the hospital and 2) constraints that limit the courses of
action available to the institution (Rosko and Broyles,
1988) .
Berki (1972) makes the point that the multiplicity of
models is a function of the diversity of analytic foci.
Moreover, the appropriateness of the model "depends on the
questions to be asked and the framework within which they
are posed" (Berki, 1972). As Berki suggests, the usefulness
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of conclusions drawn from analysis depends on the
assumptions made regarding the objectives of the hospital.
Profit Maximization
This model of hospital behavior assumes that the goal
of the hospital is to maximize profit. Moreover, this
model assumes that one of the goals of the non-profit
hospital would be to return the "profits" to the community
in the form of subsidized care to the poor or to lower
overall hospital expenses. Given the objective of profit,
non-profit and for-profit hospitals would be expected to be
cost minimizers. Hospitals would invest only in those
facilities or services expected to make profits.
Facilities or services found to result in loses would be
closed or discontinued. Moreover, cross subsidization
would not be used to off-set losses. Hospitals would also
be expected to charge higher prices if demand became
insensitive to increases in prices (Feldstein, 1993).
Studies prior to the 1980s, indicate that actual
behavior of hospitals was not congruent with the profit-
maximization model. Feldstein writes that prior to the
1980s:
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hospitals invested in facilities and services that
were known in advance to result in substantial
losses; cross subsidies were provided to certain
facilities and services to offset their losses
rather than closing them. These money losing
facilities and services were not the sole source of
such services in the community but were instead
duplicative of others. (Feldstein, 1993)
Given that the market was dominated by physicians prior to
the 1980s, the studies suggest that hospitals competed on
the basis of hospital service capability. Therefore,
hospitals would be willing to cross subsidize inefficient
operations in order to attract physicians.
In a payer-dominated market, hospitals would be more
concerned with offering competitive prices to attract
health plan contracts and would have less incentive to
cross subsidize inefficient operations. Therefore, in a
payer-dominated market hospitals would be expected to
respond more like a profit-maximizer.
Utility Maximization
The objective of the hospital under this model is to
maximize the prestige, power and professional satisfaction
of the hospital. The objectives are to be achieved through
sophisticated inputs that will enhance either quality or
quantity. Higher quality is defined in terms of higher
quantity of inputs per patient day. To meet this
definition of quality, hospitals would be expected to
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
utilize high quantities of inputs, which would exacerbate
technical inefficiency. Moreover, the objective of
prestige may explain hospital behavior as suggested by
Feldstein:
It [prestige] does explain why hospitals might make
unprofitable investments or maintain unprofitable
services, as long as these services add prestige to
the institution. This model also suggests that
hospitals will invest in new technology as soon as
it becomes available, not necessarily because of
its effect on demand, but because of its effect on
the perceived image of the hospital. (Feldstein,
1993)
In a market dominated by physicians, the more prestigious a
hospital is, the more physicians it can expect to attract.
However, the objective of the hospital may change in a
market dominated by payers who control the choice of
hospitals and physicians of plan members. The objective of
the hospital in a payer-dominated market would be to offer
low prices by maximizing economic and technical efficiency.
In a physician-dominated market, regulation of
reimbursement rates may induce hospitals to respond more
like profit-maximizers. Therefore, the utility-maximizer
would be expected to respond to rate regulation as follows
(Rosko and Broyles, 1988; Feldstein, 1993):
1. given a fixed budget, the hospital will face a
trade off between quality and quantity
2. under per diem prospective payment hospitals will:
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
a. increase length of stay
b. increase admissions
c. pursue less complex cases if per diem rate is
not case-mix adjusted
d. raise charges to unregulated patients (cross
subsidization)
e. minimize costs of inputs
f. minimize volume of inputs
3. under a case based prospective payment hospitals
will:
a. decrease length of stay
b. increase admissions
c. reduce quality and increase number of cases
d. pursue less complex cases
e. raise charges to unregulated patients (cost
shifting)
f. minimize costs of inputs
g. minimize volume of inputs
Under PPS, hospitals would be less able to pass on high
costs of inefficient operations since the amount of
reimbursement is predetermined. Moreover, PPS may alter
the hospital's objective from utilization of high
quantities of inputs to utilization of only necessary
quantities of inputs. Therefore, PPS may induce hospitals
to produce services with greater technical efficiency. In
the absence of price competition or rate regulation, the
utility-maximizing model would predict that prices in a
competitive hospital market would be excessively high and
services would be duplicative (Rosko and Broyles, 1988) .
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Price Competition
Under traditional economic theory, price reflects a
balance between demand and supply. In a competitive
environment, price sensitive suppliers would have the
incentive to offer a product at the lowest price possible
or risk loosing market share to competitors. Moreover, if
another competitor can lower their price by finding a more
efficient way of producing a product, other competitors are
induced to adopt similar methods of production, or be
forced to leave the market. A competitor may raise their
price if they offer a higher quality product which in turn
may induce other competitors to produce higher quality
products. However, prices may again fall if a competitor
can find a more efficient way of producing the higher level
quality product. Therefore, under a competitive
environment, price reflects market efficiency which in turn
reflects economic and technical efficiency of individual
firms (Dorfman, 1978).
If competition based on price is to induce efficient
behavior among firms, then certain market conditions must
be present for competition to occur. These market
conditions include (Dorfman, 1978; Arrow, 1963):
1. consumer information to compare price and quality
of products or services
2. choice among similar products or services
3. no barriers to market entry by competing suppliers
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
When market conditions fail to exist in an industry then
price will not reflect efficiency. In cases of market
failure, prices may be higher than they would be under
normal competitive conditions. Under such circumstances,
the government may impose a cap on the price a firm can
receive for products or services in order to induce the
industry to behave more like a competitive market
(Feldstein, 1993). In the health care market, the presence
of cost-based FFS third-party payers creates market failure
by creating consumer and provider insensitivity to rising
prices. Hospitals that are insensitive to the high cost of
high utilization of services have less incentive to find
more efficient ways of providing services.
Hospital Competition
Hospitals receive patients from physicians and payers.
Therefore, how hospitals respond to competition depends on
whether the market is dominated by physicians or by payers.
Physician-dominated markets are markets were the payer does
not restrict the beneficiary's choice of provider.
Consequently, hospitals will attempt to attract physicians,
and therefore patients, by offering as many support
services as possible. This behavior gives rise to excess
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and duplicative services within a highly competitive
market. Hospital competition in physician-dominated
markets often results in increased hospital expenses
(Zwanziger, 1987; Porter, 1980).
Payer-dominated markets are markets where the payer is
allowed to restrict the choice of hospital or physician and
where there is an excess supply of hospitals and physicians
(Zwanziger, 1987; Porter, 1980). In payer-dominated
markets, payers are price-sensitive to medical services.
In markets where hospitals provide comparable services at
comparable costs, hospitals must be more price and quality
competitive. To be price competitive, hospitals must
minimize their costs. Therefore, in payer-dominated
markets hospitals will tend to minimize their cost and
quantity of inputs.
Berki (1972) points out that to determine if hospitals
are behaving efficiently, the analysis must consider the
hospital's objectives since the objectives act as
constraints on the degree of efficiency the hospital may
attain. In a physician-dominated market, where there are
no restrictions on the physician's utilization of hospital
services, hospital behavior tends to resemble the behavior
predicted by the utility-maximizing model. In a market
dominated by payers who can restrict the choice of
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
hospitals and physicians, the behavior of the hospital may
look more like the behavior predicted by the profit-
maximizing model. Thus, in a payer-dominated market, rate
regulation and price competition may alter the hospital's
objectives from cost increasing objectives under the
utility-maximizing model to cost decreasing objectives
under the profit-maximizing model.
Managed Care
If price competition is to induce hospitals to compete
based on price and quality, then the market must be
dominated by payers who are able to restrict the
beneficiary's choice of physicians and hospitals. Managed
care plans are plans that selectively contract with
hospitals and physicians to provide services to their plan
members. Managed care plans usually receive a fixed fee
each month for each enrollee (e.g., capitation). The
capitated fee creates an incentive for plans to contract
with those providers that can provide quality care at the
least expense. Managed care plans may reimburse hospitals
and physicians based on capitated amounts or negotiated
rates. The different forms of reimbursement to providers
give rise to different types of managed care plans such as
Health Maintenance Organizations (HMOs), Independent
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Provider Associations (IPAs) and Preferred Providers
Organizations (PPG).
In highly competitive markets dominated by payers,
hospitals have to win contracts by offering competitively
low prices. Given that labor is the primary input in
producing hospital ancillary and nursing services,
hospitals with a high percent of managed care contracts
would be expected to minimize the cost and quantity of
labor inputs in order to offer competitive prices.
Rate Regulation
The rational for rate regulation is based on the
assumption that competitive forces have failed to achieve
price equilibrium in the market place. Prices charged by
providers are prices that do not reflect efficiency since
many of the conditions for market competition have not been
present in the health care industry. Opponents of rate
regulation argue that rates under rate regulation may be
set higher or lower than prices would be under normal
market conditions. Rates that are set too high would not
correct incentives for inefficient behavior and rates that
are set too low may force suppliers to reduce the quality
of products or services provided (Rosko and Broyles, 1988) .
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Under normal market conditions, consumers may decrease
their demand for a product if prices increase. Health
insurance reduces consumer sensitivity to increases in the
price of physician and hospital services (Arrow, 1963).
Thus, in the absence of price controls or incentives,
consumers and providers have little incentive to minimize
the utilization of services. Similarly, consumers have
limited ability to assess the need for medical services and
to compare providers on the basis of quality and price.
Therefore, providers have less incentive to supply services
at a higher level of quality in order to maintain market
share (Feldstein, 1993; Arrow, 1963).
Prospective Payment
Prospective payment is "a generic term that refers to
financing mechanisms in which rates or levels of
compensation are determined prior to a future period and
the hospital receives the predetermined amount irrespective
of the costs that are incurred" (Rosko and Broyles, 1988) .
The unit of payment may be patient day or patient related
diagnosis. In 1983, Congress implemented a prospective
payment system based on patient related diagnoses or
Diagnostic Related Groups (DRGs). DRGs are a patient
classification system that groups patient diagnoses
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
according to expected resource expenditures. The advantage
of using case type over patient day as a basis for
reimbursement is that case type recognizes the multi
product nature of hospital outputs and accounts for
differences in costs of treating different types of
patients. However, some of the weaknesses of DRGs include
(Borden, 1986; Rosko, 1988; Horn and Schumacker, 1982):
1. they fail to account for differences in severity of
illness within a DRG since they fail to
differentiate patients in different stages of
illness
2. they are statistically meaningful only for those
populations from which they were derived
3. hospitals have the incentive to code patients into
higher DRGs to maximize reimbursement
Despite some of their short comings, DRGs were expected to
be a better alternative, than PPS based on patient days.
Therefore, DRGs were adopted by Medicare as the basis for
reimbursing hospital inpatient care for Medicare patients.
The rational for rate regulation based on DRGs was to slow
the rate of Medicare expenditures by inducing hospitals to
behave more efficiently. However, if prospective payment
is to be effective the following conditions should be met
(Borden, 1986; Rosko and Broyles, 1988):
1. mandatory participation by all hospitals
2. all payers must reimburse based on PPS
3. case-mix differences must be accounted for
4. PPS should contain incentives for hospitals to
behave more efficiently
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5. preserve the fiscal viability of providers
6. health needs should determine services provided to
patients versus fiscal incentives
The first condition was met by Medicare PPS since all
hospitals that treated Medicare patients were reimbursed by
DRGs. However, without participation by non-Medicare
payers, the Medicare Prospective Payment System, alone, may
not be effective at slowing the overall rate of health
expenditures since hospitals may theoretically charge
higher prices to non-Medicare payers. However, the rise of
capitated managed care plans may assist in slowing the rise
of health expenditures in the non-Medicare population.
In the absence of market determined prices, rate
regulation is expected to act as a price constraint on the
ability of hospital decision makers to achieve costly
objectives. Under this model hospitals are expected to
(Rosko and Broyles, 1988):
1. increase number of admissions
2. reduce the direct cost per case
3. reduce costs of inputs
4. reduce rate of growth of FTEs and FTE salaries
5. pursue more admissions of less complex cases (lower
case-mix)
6. shift inpatient care to outpatient settings
7. increase prices to non-regulated insurance plans
In anticipation of increased admissions under PPS, Congress
included in the Medicare legislation the establishment of
Physician Review Organizations (PROs) to review
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
appropriateness of admissions (Feldstein, 1993) . Thus,
under PROs, hospitals could potentially be denied Medicare
reimbursement for inappropriate admissions.
Hospitals were expected to respond to PPS by
minimizing the cost per input and the volume of inputs used
in providing services. Given that labor FTEs are a
significant input in the production of ancillary and
nursing services, hospitals would be expected to reduce
their labor inputs.
Efficiency
Efficiency is defined as 1) economic efficiency and 2)
technical efficiency. Economic efficiency refers to the
least expensive combination of inputs utilized in
maximizing output. Technical efficiency refers to the
least number of inputs utilized in maximizing output. Cost
minimization is achieved when a hospital is both
economically and technically efficient; or when the slope
of the isoquant line equals the slope of the isocost line
(Dorfman, 1978). Studies of California hospitals since
1983 have indicated that hospitals are operating more
efficiently since hospitals on average have reduced the
rate of hospital expenditures (Zwanziger et at., 1994a).
However, it is less clear if hospitals in California are
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
operating more efficiently because they have reduced the
cost of inputs, the number inputs or both.
Parametric Measures
Traditionally, the structure of hospital production
has been measured by parametric cost functions. Cost
functions measure cost minimization, which is the point of
tangency between the isoquant and the isocost lines. The
isocost represents the cost of input combinations and the
isoquant represents the quantity of input combinations. The
point of tangency represents the combination of fewest
inputs at the least cost to produce a certain level of
output, referred to as cost minimization. Most studies that
measure hospital efficiency use a cost function since it
specifies both economic efficiency (cost) and technical
efficiency (quantity). Cost functions are expected to
measure both economic and technical efficiency because "the
duality theory posits a relationship between production
functions and cost functions in which the specification of
one implies the specification of the other" (Rosko and
Broyles, 1988) . More specifically, "the dual to cost
minimization (subject to a given output) is output
maximization (subject to a budget constraint)" (Rosko and
Broyles, 1988). However the limitation in using the cost
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
function approach in determining technical efficiency is
that "if firms are not cost minimizers^ it becomes very
difficult to derive the duality relationship between cost
and production functions" (Rosko and Broyles, 1988). Given
that the objective of hospitals in physician-dominated
markets has been utility-maximization, hospitals would not
be cost minimizers especially under cost-based
reimbursement. Therefore, cost functions may not be
appropriate for measuring technical efficiency if hospitals
are not cost minimizers.
Given that the utility-maximizing objective function
of the hospital tends to violate the cost minimization
assumption of duality between production and cost
functions, an alternative to measuring technical efficiency
is a parametric production function. The most commonly used
production functions are the Cobb-Douglas and Translog
models (Feldstein, 1967; Rosko and Broyles, 1988). Rosko
and Broyles point out the significant differences between
the two approaches:
In the translog model the substitution elasticities
and the output elasticities are dependent on the
level of input utilization. In contrast, the Cobb-
Douglas model is based on a less realistic
assumption that output elasticities are independent
of the level of inputs. (Rosko and Broyles, 1988)
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
As the authors suggest, the translog model allows more
flexibility in the functional form of the production
function.
Another approach to measuring technical efficiency is
the stochastic production frontier. Efficiency is reflected
in the error term, which is composed of two parts. One
part reflects statistical "noise", which the firm has no
control over, and the other component measures technical
efficiency relative to the stochastic frontier. Wilson and
Jadlow write that:
The major weakness of this approach [stochastic
production frontier], however, is that it does not
provide a method for decomposing individual
residuals into their two components; thus, it is
not possible to estimate technical inefficiency for
each observation. The most one can do is to
estimate the average level of inefficiency for the
sançle. (Wilson and Jadlow, 1982)
The cost function and production function are parametric
approaches to measuring efficiency. One of the draw backs
of using parametric approaches in measuring efficiency is
that the coefficients are a reflection of average
performance rather than a reflection of the best
performance relative to some standard. The process of
averaging may require a weighting scheme that results in
various degrees of information loss. In addition, averaging
combines both efficient and inefficient firms together.
Therefore, parametric approaches "will only reflect
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
efficient relationships when all of the observations
themselves are efficient” (Sherman, 1984) . Moreover, the
averaging effect "provides little direct information
concerning the magnitudes of efficiency gains that are
possible at various decision-making units within the
sample" (Silkman, 1986). Therefore, cost functions "may be
a good predictor of what costs will be assuming a constant
level of inefficiency, but they say nothing about efficient
relationships" (Sherman, 1984).
Studies by Zwanziger et al. (1987, 1988, 1994a and
1994b) indicate that, on average, hospitals in California
had a decline in total expenditures between 1980 and 1990.
However, it is less clear if these hospitals were equally
efficient or if they were efficient relative to the best
standard of performance.
Data Envelopment Analysis
Data Envelopment Analysis (DEA) is an approach to
measuring technical efficiency that overcomes the
limitations of parametric approaches. Silkman writes that:
DEA. is a non-parametric approach for measuring
efficiency of decision making units (DMUs)
relative to an optimal level of performance. DEA
also identifies the efficiency gains that are
possible for each inefficient DMU. DEA permits
examination of particular production
characteristics such as efficiencies, returns to
scale and rates of transformation prevailing in
specific segments of the production possibility
set. (Silkman, 1986)
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In addition, DEA is useful in modeling multiple-output
multiple-input technologies (Banker et al., 1986). Ozcan
provides a straight forward and succinct explanation of the
mathematical formula behind DEA as follows:
Data Envelopment Analysis is a tool in which linear
programming is used to search for the optimal
combinations of inputs and outputs, based on the
actual performance of, in this case, hospitals. The
program evaluates the technical efficiency of each
hospital relative to "optimal" patterns of
production, which patterns are computed using the
performance of hospitals whose input/output
combinations are not bested by those of any other
comparison or peer hospital. The way in which DEA
program computes efficiency score is explained
briefly using mathematical notations (adapted from
Charnes and Cooper 1980). The efficiency scores
(Ej) for a group of peer hospitals (j=l,...n), are
confuted for the selected outputs (y, j, r=l,...S)
and inputs (Xt 3, 1=1, . . .m) using the following
linear programming formula:
S
sum U r Y r o
Maximize : Ep = r=l__________
m
sum Vi Xi
1=1
sum U r Yr J
Subject to: r=l________ < 1
m
sum Vi Xi j
1=1
Ur, Vi > 0 for all
r and I
In this formulation, the weights for the outputs
and inputs, respectively, are Ur and Vi, "o"
denotes a focal hospital (each hospital, in turn,
becomes a focal hospital when its efficiency score
is being computed) . Note that input and output
values as well as all weights are assumed by the
formulation to be greater than zero. The weights
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Ur and Vi for each hospital are determined entirely
from the output and input data of all hospitals in
the peer group. Therefore, the weights used for
each hospital are those that maximize its-the focal
hospital's-efficiency score. The program also
identifies a group of optimally performing
hospitals that are defined as efficient and assigns
them a score of one. These efficient hospitals are
then used to create an "efficiency frontier" or
"data envelop" against which all other hospitals
are compared. In sum, hospitals that require
relatively more weighted inputs to produce weighted
outputs or, alternatively, hospitals that produce
less weighted output per weighted inputs than do
hospitals defined by the program to be on the
efficiency frontier, are considered technically
inefficient. They also are given efficiency scores
of less than one, but greater than zero. (Ozcan,
1993)
An additional assumption of DEA is that "all points along
the efficient frontier, which is convex by definition, are
practically attainable production possibilities" (Ehreth,
1994). Given that the frontier is based on a specific set
of hospitals, it is assumed that each hospital has the
capacity to attain a position on the frontier. However, if
some hospitals are excluded from the study group, other
unknown capabilities may be possible but are not recognized
in the analysis. Thus, attainable levels of performance is
relative only to the hospitals included in the analysis.
The advantage of using DEA as an approach to measuring
technical efficiency relative to the parametric approaches
was demonstrated by Banker et al. (1986) . In this study,
the authors compared the results of DEA with a translog
flexible form regression model. Using the same set of data
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from hospitals in North Carolina, they found the presence
of increasing and decreasing returns to scale under DEA.
However, when applying the translog flexible form
regression model, these increasing and decreasing returns
to scale where not found because of the "averaging out
effect" of regression analysis (Banker et al., 1986).
Thus, DEA provides additional information regarding
technical efficiency that cost or production functions may
not provide.
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IV. LITERATURE REVIEW
Competition and Regulation
The utility-maximizing model of hospital behavior
predicts that the costs of producing services would be
higher than would be found in hospitals with an objective
function of profit-maximization. Moreover, given that
hospitals tend to compete on non-price factors in
physician-dominated markets, increasing competition among
hospitals would increase costs, since hospitals would
charge more to cover higher costs and quantities of inputs.
Prior to the 1980s, the health care market was considered a
physician-dominated market since the majority of health
plans did not restrict the choice of hospitals or
physicians of its beneficiaries. A number of studies prior
to the implementation of Medicare prospective payment, in
1983, tended to confirm the utility-maximizing behavior of
hospitals in competitive environments.
Joskow (1980) examined the relationship between
competition and hospital bed supply in 346 private, non
profit hospitals in 1976. A Hirschman-Herfindahl Index
(HHI) was used to measure competition within SMSA defined
market areas. Hospitals with lower HHI scores (e.g.,
greater competition) were found to have excess bed
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
capacity, which, functioned to accommodate the medical
staffs.
Robinson and Luft (1985) used 1972 data to examine the
relationship between hospital market and inpatient cost per
admission, inpatient cost per day, and the number of
inpatient admissions. Markets were defined as the area
within a 15 mile radius of the hospital and competition was
measured by the number of hospitals within the 15 mile
radius. They found that hospitals in more competitive
markets had higher cost per day, cost per admission and
lower volume than hospitals in less competitive markets.
Using this same approach to defining markets and measuring
competition, Luft and Robinson (1987) found that the
relationship between average cost per admission and degree
of competitiveness was about the same in 1982 as in 1972.
Using again this same approach to measuring
competition in 1972, Luft (1986) found that hospitals in a
more competitive area had a higher probability of offering
the same service as a competitor.
Wilson and Jadlow (1982) examined the relationship
between competition and technical efficiency in providing
Nuclear Medicine Services in 1973. Markets were defined by
patient catchment areas and competition was measured in
terms of the number of hospitals and population size. They
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
found that hospitals in more competitive areas were less
technically efficient and proprietary hospitals were more
efficient than non-profit hospitals.
Farley (1985) examined the effects of market structure
on hospital performance between 1970 and 1977. Market areas
were defined by county and competition was measured by the
HHI. The study showed that hospitals residing in highly
competitive counties performed more procedures for certain
diagnoses and used more labor and capital per patient.
The above studies suggest that under cost-based
reimbursement, in physician-dominated markets, hospitals
tended to compete on the basis of service volume rather
than on the basis of price. Under cost-based reimbursement
hospitals could simply increase prices as costs increased
without concern that demand would decline. The inflationary
dynamics of cost-based reimbursement and the utility
objective function of the hospital resulted in excess
health care cost inflation.
The inflationary dynamics of health care expenditures
under cost-based reimbursement have been considered to be
the outcome of market failure. Under economic theory,
price sensitive firms within competitive environments will
attempt to minimize the quantity and cost of inputs in
maximizing outputs given a market driven price constraint.
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Under cost-based reimbursement, the market lacked a market
driven price constraint. Rate regulation and price
competition have been two approaches to impose price
constraints on the market.
Hospitals in the health care market in California have
been operating under price constraints since the early
1980s when selective contracting and Medicare PPS
legislation were implemented. In 1982, the state of
California implemented legislation to promote price
competition. The legislation would allow public and
private health insurers to selectively choose providers to
contract with based upon agreed prices (e.g., managed
care). Plan beneficiaries would be required to utilize
only those physicians and hospitals contracted with the
plan. Under managed care contracting, if providers offered
rates that were higher than competitors and were not
justified by higher quality, then providers would be at
risk for loosing market share.
Similarly, in 1983, the federal government implemented
legislation to regulate hospital rates. This legislation
would allow Medicare to reimburse hospitals a fixed rate
based on a patient's diagnostic category (DRG) . The DRG
based prospective payment system (PPS) replaced cost-based
reimbursement of hospital services. Under both managed
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
care contracting and Medicare PPS, hospitals in competitive
markets would have the incentive to minimize the cost and
quantity of inputs in providing services to minimize the
price at which they could compete. The price constraints
imposed under PPS and managed care contracting created a
more price sensitive demand curve for medical services.
Analysis of California hospitals have provided
valuable insight into the effects of managed care
contracting and PPS on costs of hospitals in competitive
markets. By 1986, 60 percent of California beneficiaries
were enrolled in managed care contracts such as Preferred
Provider Organizations (PPOs) and Health Maintenance
Organizations (HMOs). By 1990, 80 percent of California
beneficiaries were enrolled (Zwanziger et al., 1994b).
Given that PPOs and HMOs selectively contract with
providers based on price and quality, hospitals in
competitive markets would be expected to compete based on
price for market share.
Zwanziger and Melnick (1988a) examined the effects of
PPS, Medi-Cal managed care contracting and market
competition on total hospital expenditures. All California
hospitals were included in the analysis. Market
competitiveness was measured by a Hirschmaui-Herfindahl
Index (HHI) which is an index of both the number of
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
competing firms in an industry and their relative market
share. PPS was measured by a PPS index based on the
percent of Medicare discharges. Managed care contracting
was measured by the percent of Medi-Cal discharges. A cost
function was used to measure the correlation between
Medicare PPS, Medi-Cal managed care contracting and market
competition (HHI), and total hospital expenditures from
1980 to 1985. The study confirmed that between 1980 and
1985, total hospital expenses declined significantly after
the implementation of PPS and managed care contracting.
Lower expenditures were statistically correlated with the
percent of Medicare discharges, Medi-Cal discharges, and
the HHI. Moreover, total expenses for hospitals in highly
competitive areas declined more significantly than
hospitals in less competitive areas. Thus, the authors
provide evidence that the utility-maximizing objective of
the hospital under cost-based reimbursement was cost
increasing. Conversely, under price constraints imposed by
PPS and managed care contracting, hospitals responded by
controlling expenditures. The reduction in costs may
suggest hospitals performed with greater economic and/or
technical efficiency.
In a second study, Zwanziger et al. (1994b) examined
the correlation between Medicare PPS, private and public
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
managed care contracting, and the rate of growth in
California hospital expenses in high and low areas of
competition from 1980 to 1990. Hospitals were divided into
areas of high and low competition based on the most and
least competitive quartiles of all California hospitals.
Market competition was measured by HHI. Medicare PPS was
measured by a PPS index based upon the percent of Medicare
discharges. Public managed care contracting was measured
by the percent of Medi-Cal discharges. Private managed
care contracting was measured by the percent of discharges
from managed care plans including HMOs and PPOs. The rate
of growth in hospital expenses was tested for correlation
with the percent of Medicare discharges, percent of HMO
discharges and percent of Medi-Cal discharges, controlling
for case-mix, hospital specialization, bed size, ownership,
location and input prices. The results indicated that,
between 1980 and 1982, hospital expenses of hospitals in
competitive markets were 17 percent higher than hospitals
in less competitive markets. In 1990, expenses of hospitals
in competitive markets were only 4 percent higher than
hospitals in less competitive markets. More specifically,
private and public managed care contracting "had induced
highly competitive hospitals to decrease their costs by
almost 13 percent relative to costs of hospitals in less
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
competitive markets" (Zwanziger et al., 1994a). Cost
savings were due primarily to reduced admissions rates,
lower LOS, growth in outpatient utilization, reduced cost
per discharge and cost per patient day. The lower cost per
discharge and cost per day may suggest that hospitals
became more economically and technically efficient in the
production of ancillary and nursing services.
Although prospective payment and managed care
contracting in areas of high competition have been
effective at reducing the rate of increase in hospital
expenses, there is still some question regarding whether
hospitals are becoming more efficient or simply shifting
the cost of care to other outpatient settings. Zwanziger
et al. (1994a) conducted a third study to examine hospital
efficiency. Efficiency was defined as the cost per standard
unit of measure (SUMS). The authors grouped 48 revenue cost
centers into 11 composite revenue centers and combined
output measures into standard units of measure (SUMS) by
various weighting schemes. Overall, they found that in
areas of high competition between 1982 and 1988, managed
care contracting and PPS reduced total expenses. The
reduction was due to a reduction in the cost per SUM. The
reduction in expenses appeared to be across cost centers
rather than concentrated in only a few. The lower cost
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
per unit of service may suggest that hospitals produced
services with greater economic and/or technical efficiency.
Another study by Phibbs and Robinson (1989) examined
the effects of Medi-Cal managed care contracting and
Medicare prospective payment on the rate of increases in
the average cost per admission for 262 private hospitals in
competitive environments in California between 1982 and
1986. Markets were defined by the zip codes that made up
75 percent of the hospital's discharges. Markets with 10 or
fewer hospitals within the market were considered areas of
low competition. Markets with more than 10 hospitals within
the market area were considered areas of high competition.
Managed care contracting was measured by the percent of
Medi-Cal discharges and the change in the percent of Medi-
Cal discharges for each hospital. PPS was measured by the
percent of Medicare discharges and the change in the
percent of Medicare discharges. Expenditures were defined
as the change in cost per admission. The study found that
hospitals in areas of high competition had lower rates of
increase in expenditures than hospitals in areas of low
competition. The study also found that hospitals with a
high percent of Medicare discharges experienced lower rates
of inflation in cost per admission than hospitals with a
lower percent of Medicare discharges.
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hospital Labor
As the previous studies indicate. Medicare PPS and
managed care contracting have had a greater effect on
reducing total costs and the rate of increase in costs for
hospitals in areas of high market competition, than for
hospitals in areas of low market competition. However, it
is less clear if this reduction in costs is due to a
reduction in labor hours used to produce hospital
discharges and outpatient visits. Only a few studies have
examined the changes in staffing levels due to PPS or
managed care contracting by public and private payers.
In 1994, the state of Tennessee converted its Medicaid
program to managed care contracts (Rudnick, 1995). St.
Joseph's Medical Center responded by cutting 22 managerial
jobs and 140 hourly jobs. St. Joseph's Medical Center
reduced expenditures by 8 million with 6 million due to
staffing reductions. These reductions suggest that
hospitals responded to price constraints under Medicaid
managed care contracting by reducing the number of labor
FTEs.
The National Hospital Rate-Setting Study of 1982
examined the effect of state prospective reimbursement on
reductions in staffing for 2,700 hospitals (Kidder and
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Sullivan, 1982). They found that hospitals responded by
reducing full-time-equivalent staff per adjusted inpatient
day. The study indicates that hospitals responded to
Medicaid managed care contracting by reducing labor hours
in their Nursing Service Areas.
Cromwell and Puskin (1989) examined the effects of
Medicare PPS on changes in labor hours for Ancillary
Service Departments and Nursing Service Departments for
1400 hospitals between 1980 and 1987. The measure of
analysis for Nursing Service Departments was labor hours
per adjusted discharge. The measure of analysis for the
Ancillary Service Departments were the hours per service.
In the Nursing Service Departments, the study found a
smaller annual percent increase in hours per discharge
between 1984 and 1987 than between 1981 and 1983.
Similarly, in the Ancillary Service Departments, the study
found a smaller annual percent increase in hours per
service between 1984 and 1987 than between 1981 and 1983.
The study indicates that hospitals responded to the
implementation of Medicare prospective payment, in 1983, by
reducing labor inputs in their Nursing Service Areas and in
their Ancillary Service Area between 1984 and 1987.
A study by Long et al. (1987) examined the effects of
Medicare prospective payment (PPS) on reductions in the use
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of ancillary services for 50 DRG categories for 729
hospitals between 1980 and 1984. Although the study did
not directly examine reduction in labor inputs, a
reduction in the number of ancillary services per DRG
category would imply a reduction in the need for labor
inputs. The study found a reduction in the use of lab
tests, x-rays and diagnostic tests across all DRG
categories between 1983 and 1984. The study implies that
hospitals responded to Medicare PPS by reducing the number
of services per DRG category, which further implies a
reduction in the need for labor inputs in the Ancillary
Service Areas.
Ozcan and Luke (1993) examined hospital
characteristics expected to influence technical efficiency
in 3,000 urban hospitals in 1987. Data Envelopment Analysis
(DEA) was used to generate scores of technical efficiency.
The DEA model input categories included non-physician full-
time-equivalents (FTEs) . The DEA scores were used as the
dependent variable in an Analysis of Covariance. Higher
technical efficiency was found to be associated with a
higher percent of managed care patients. An increase in
technical efficiency would suggest a reduction in non
physician FTEs. Therefore, the study indicates that
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
hospitals responded to managed care contracts by reducing
labor FTEs in their Ancillary and Nursing Service Areas.
Given the few studies that have examined changes in
labor hours or full-time-equivalents (FTEs), the results
indicate that hospitals have responded to Medicare PPS
(rate regulation) and public/private managed care
contracting (price competition) by reducing labor hours or
FTEs in their Ancillary or Nursing Service Areas.
Data Enveloixnent Analysis (DEA)
A number of studies have used DEA to measure technical
efficiency in the health care field.
Nyman and Bricker (1989) examined technical efficiency
in 184 Wisconsin nursing homes in 1979. Labor efficiency
scores were generated from DEA to be used as the dependent
variable in a regression model. The regression model
tested for correlation between labor efficiency and various
nursing home characteristics. The DEA input categories
consisted of labor hours for nursing, social service
workers, therapists and other workers. The DEA output
categories included the number of patients for SNF and IGF,
limited care requirements and residential care
requirements. The independent variables of the regression
model included ownership, percent Medicaid patients.
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
hospital affiliation, bed size and occupancy rate. Since
the DEA scores were not normally distributed, the model was
run with and then without those hospitals with a score of
1, eliminating skewness. The results were generally
consistent in both models. The results showed that for-
profit homes and homes with higher occupancy rates had
significantly higher efficiency scores. The percent of
Medicaid patients was weakly associated with efficiency and
neither size or hospital affiliation was significantly
related to efficiency.
Nyman, Bricker and Link (1990) examined the
determinants of efficiency for 296 Iowa Intermediate Care
Facilities (ICFs) in 1983. Labor efficiency scores were
generated by DEA that were then used as the dependent
variable in a regression model. The regression model
tested for correlation between labor efficiency and certain
nursing home characteristics. The DEA input categories
included labor hours for management, RNs, LPNs, aides,
social service and housekeeping. The DEA output category
was the number of IGF residents. The independent variables
of the regression model included ownership, beds, percent
Medicaid patients, activities-of-daily-living (ADL) scores,
occupancy rate and location. Since the DEA scores were not
normally distributed, the model was run with and then
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
without those hospitals with a score of 1, eliminating
skewness. The results were generally consistent in both
models. For-profit status, number of beds, supervisory
hours, and occupancy rate were found to be correlated with
higher labor efficiency. Confused patients and patients
over 85 were associated with decreased labor efficiency.
Ozcan and Luke (1993) examined hospital
characteristics expected to influence technical efficiency
in 3,000 urban hospitals in 1987. The DEA scores were used
as the dependent variable in an Analysis of Covariance
model to determine which hospital characteristics were
correlated with technical efficiency. The DEA input
categories included beds, number of diagnostic services,
non-physician FTEs and operational non-payroll expenses.
The DEA output categories included adjusted discharges,
outpatient visits and medical trainees. The independent
variables of the analysis included bed size, ownership,
membership in a multi-hospital system, percent of
discharges for Medicare, Medicaid and managed care. For-
profit hospitals and hospitals with a high percent of
Medicare patients were correlated with lower technical
efficiency. Managed care contracts, membership in a multi
hospital system, and bed size of hospital were all
correlated with higher hospital technical efficiency.
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
although not consistently significant across all MSAs. In
contrast/ Medicaid was not significantly related to
technical efficiency. Higher technical efficiency
associated with a higher percent of managed care patients
may suggest that managed care contracts induce hospitals to
produce services with greater technical efficiency.
Haung and McLaughlin (1989) examined the effects of
population size and organizational form of 77 rural primary
health care centers on technical efficiency. To test for
validity of the DEA efficiency scores, the scores were
compared with efficiency scores generated from ratio and
regression analysis. In both comparisons, the scores were
found to be consistent in classifying programs as efficient
or inefficient. The DEA model included controllable and
non-controllable inputs and outputs. The controllable input
categories were full-time-equivalents (FTEs) for
management, MDs, RNs, technicians, and physician's
assistants (PAs). The non-controllable input categories
included population size, age of program and percent of
users under age four. The controllable output categories
included the number of encounters by MDs, PAs and RNs and
the uncontrollable output category included encounters by
'other'. The results indicated that rural health programs
performed more efficiently in small population areas
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
regardless of organization form- The Organized Group
Practices (OGPs) were most efficient of the three
regardless of the population size. The Community Health
Centers (CHCs) were the least efficient, especially in
large service areas. The Primary Care Centers (PCCs)
performed more efficiently if located in small services
areas.
Valdmanis (1990) used Data Envelopment Analysis (DEA)
to examine technical efficiency between 41 public and
private non-profit hospitals with 200+ beds in Michigan in
1982. The DEA input categories included number of active
and associate physicians, full-time-equivalents (FTEs) for
RNs, medical trainees, non-physician labor and net plant
assets. The DEA output categories included inpatient days,
surgeries and outpatient visits. Since the DEA scores were
not normally distributed, a Man-Whitney non-parametric test
for statistical significance was used to measure
differences between DEA scores for public and private non
profit hospitals. The results indicated that public
hospitals were more technically efficient than private non
profit hospitals. More specifically, public hospitals used
fewer inputs and although they offered a smaller array of
services, the demand for these services justified the
expense. In keeping with their utility-maximizing objective
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
function, in 1982, private non-profit hospitals used a
greater number of inputs and offered a wider array of
sophisticated services that were not necessarily justified
given the demand.
Sexton et al. (1989) used DEA to examine the relative
efficiency of 159 Veterans Administration Medical Centers
(VAMCs) in 1985. The DEA input categories included full-
time-equivalents (FTEs) for RNs, physicians, residents,
technicians, equipment expenses and drugs and supply
expenses. The DEA output categories included weighted
DRGs for medical, surgical, psychiatry, intermediate care
days and outpatient visits. The DEA score was used as a
binary dependent variable in a probit model to determine if
technical efficiency was correlated with certain
characteristic of the VAMCs. The independent variables of
the probit model included university affiliation, age of
hospital, location, proportion of beds by hospital
department, number of administrative FTEs per physician and
number of trainees in the categories for medical students,
RNs, and assistants. The results indicated that one-third
of the VAMCs were more likely to be inefficient and larger
VAMCs and VAMCS associated with a university were more
likely to be inefficient. In addition, over 300 million
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
annually could be saved on personnel, equipment, drugs and
supplies if this inefficiency were eliminated.
Sherman (1984) used DEA to examine the relative
efficiency of Med/Surg units in 7 Massachusetts teaching
hospitals in 1976. The DEA input categories of the
Med/Surg unit included non-physician full-time-equivalents
(FTEs), number of beds, supplies and purchase service
expenses. The DEA output categories of the Med/Surg unit
included the number of patient days for patients over 65,
patient days for patients under 65, the number of medical
residents and the number of RN trainees. Two of the seven
Med/Surg units were identified as inefficient. A panel of
experts consisting of regulators, managers and hospital
management consultants, who were familiar with the seven
hospitals, concurred with the results of the DEA analysis.
Lynch and Ozcan (1994) examined causes to closures of
all non-governmental short term general hospitals in 1988.
The DEA input categories included total number of
diagnostic and special services, number of beds, non
physician full-time-equivalents (FTEs) and supply expenses.
The DEA output categories included Medicare case-mix
adjusted discharges, outpatient visits and number of
medical and dental trainees. The DEA scores were used as a
binary independent variable in a logistic regression model.
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The dependent variable was open or closed hospitals. The
independent variables included the number of beds,
competition (HHI), beds*competition, occupancy rate,
percent of Medicare and Medicaid days and the DEA score.
The results showed no relationship between closure and
percent of Medicare, Medicaid or competitiveness.
Conversely, hospitals with a high occupancy rate and a high
number of beds were less likely to close. However, the
interaction term, beds*competition, was negative and
significant, which indicated that the effect of size on
closure decreased as competition decreased. However,
inefficient hospitals were less likely to close.
Grosskopf and Valdmanis (1987) examined technical
efficiency in 22 public and 60 private non-profit hospitals
in California in 1982. The DEA input categories included
full-time-equivalents (FTEs) for physicians, non
physicians, number of admissions and net plant assets.
The DEA output categories included inpatient acute care
days, intensive care patient days, surgeries and outpatient
visits. Since the DEA scores were not normally distributed,
a non-parametric Mann-Whitney test was used to test for
statistical significance between the DEA scores for public
and private non-profit hospitals. The results indicated
that public hospitals utilized fewer resources than private
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
non-profit hospitals. On average, public hospitals had
more ambulatory and emergency room visits and utilized more
non-physician labor and less capital.
Nunamaker (1983) compared DEA with the Medicare cost
per patient day approach to identifying inefficiency. He
examined routine nursing services in 17 non-teaching
hospitals of similar size in Wisconsin between 1978 and
1979. Under Medicare's approach, hospitals were classified
as inefficient if they fell above the cost per patient day
ceiling of the their particular class. The DEA input
category was total inpatient routine expenses. The DEA
output categories included patient days for patients over
65, pediatric, maternity, and other routine days. DEA
identified that 60 percent of the hospitals were
inefficient in 1978 and 1979. Conversely, the Medicare
approach identified no hospitals as inefficient in 1978 and
only 7 percent in 1979. In addition, hospitals classified
as more efficient under DEA had corresponding lower cost
per day.
Ehreth (1994) evaluated various hospital performance
measures to determine those measures that represented
hospital behavior most accurately. Descriptive statistics
and factor analysis were used on a sample of hospitals to
test the reliability and validity of the measures.
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Medicare cost report and balance sheet information were
used for all hospitals between 1987 and 1989. The study
concluded that the measures that were most representative
of hospital performance included: DEA to measure hospital
technical efficiency; the current ratio to measure short
term financial performance; long-term debt to net-fixed
assets to measure long-term viability; total margin to
measure profitability; and Medicare margin to represent
Medicare's contribution to hospital financial position. The
factors most important in DEA analysis were the input
categories labor and capital and the output categories
case-mix adjusted discharges and outpatient visits.
Summary
Studies that examined the effect of market competition
on changes in hospital costs prior to the implementation of
Medicare PPS and selective contracting in 1983, indicated
that hospitals in areas of high competition had higher
costs than hospitals in areas of low competition. The
studies found that hospitals in areas of high competition:
a. had excess bed capacity, which functioned to
accommodate the medical staffs (Joskow, 1980)
b. had higher cost per day, cost per admission and
lower volume than hospitals in less competitive
markets (Robinson and Luft, 1985)
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
c. had a higher probability of offering the same
service as a competitor (Luft, 1986)
d. were less technically efficient although
proprietary hospitals were more efficient than
non-profit hospitals (Wilson and Jadlow, 1982)
e. performed more procedures for certain diagnoses and
used more labor and capital per patient (Farley,
1985)
The results of these studies suggest that hospitals in
areas of high competition tended to compete on the basis of
service rather than on the basis of price. Competing on
the basis of service to attract physicians contributed to
excessive costs and duplicative services in highly
competitive markets. This behavior would be congruent with
the utility maximizing behavior predicted by the Utility
Maximization Model of hospital behavior in physician-
dominated markets. Physician-dominated markets are markets
with a high number of physicians and where payers do not
control the choice of physician or hospital.
Studies in California, following 1983, indicated that
hospitals responded to Medicare PPS (rate regulation) and
public and private managed care contracting (price
competition) by reducing total hospital expenditures and
reducing the rate of increase in expenditures. Studies
following the implementation of Medicare and selective
contracting in 1983 indicated that:
a. between 1980 and 1985, total hospital expenses
declined 7 percent (Zwanziger and Melnick, 1988a)
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
b. hospital expenses were 17 percent higher than
hospitals in less competitive markets in 1982 and
only 4 percent higher in 1990 (Zwanziger et al.,
1994b)
c. between 1982 and 1988 hospitals reduced their cost
per standard unit of measure (SUM) across all cost
centers rather than in only a few cost centers
(Zwanziger et al., 1994a)
d. hospitals in areas of high competition had lower
rates of inflation in their cost per admission
than hospitals in areas of low competition (Phibbs
and Robinson, 1989)
e. hospitals with a high percent of Medicare or Medi
cal patients experienced a 3.3 or 9.7 percent
reduction in cost per admission inflation,
respectively (Phibbs and Robinson, 1989)
These findings suggest that rate regulation and price
competition were effective at inducing hospitals to
increase economic efficiency. Moreover, these studies
suggest that hospitals in areas of high competition tended
to compete on the basis of price rather than on the basis
of service. This behavior would be congruent with the
profit maximizing behavior predicted by the Profit
Maximization Model of hospital behavior in markets
dominated by payers. Payer-dominated markets are markets
where the payer can control the choice of hospital and
physician.
The studies on hospital costs following the
implementation of PPS and selective contracting in 1983
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
examined changes in economic efficiency but did not examine
changes in technical efficiency. A number of studies have
utilized DEA to examine technical efficiency in various
health care institutions. Of the studies, only one
examined the effect of Medicare PPS and managed care on
hospital technical efficiency (Ozcan and Luke, 1993) .
However, this study did not examine chauiges in technical
efficiency overtime nor did the study focus on only labor
inputs. Moreover, the study did not examine the effect of
market competition on changes in labor efficiency. In the
majority of the DEA studies, the most commonly used input
measures were FTE categories and net plant assets; and the
most commonly used output measures were case-mix adjusted
discharges and outpatient visits (Ozcan and Luke, 1993;
Valdmanis, 1990; Sexton et al., 1989; Sherman, 1984;
Valdmanis and Grosskopf, 1987; Nunamaker, 1983).
The few studies that have examined the correlation
between changes in FTEs and PPS and selective contracting
have found that hospitals tended to reduce their FTEs in
both hospital Ancillary and Nursing Service Areas. These
studies found that;
a. in 1994, St. Joseph's Medical Center in the state
of Tennessee responded to state Medicaid managed
care contracts by cutting 22 managerial jobs and
140 hourly jobs (Rudnick, 1995)
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
b. in 1982, 2,700 hospitals nation wide responded to
state prospective reimbursement by reducing full
time-equivalent staff per adjusted inpatient day
(Kidder and Sullivan, 1982)
c. in response to Medicare PPS, 1400 hospitals across
the nation experienced smaller average annual
percent increases in hours per discharge in
Nursing Service Areas; and smaller annual percent
increases in hours per service in Ancillary
Service Areas. The average annual percent
increases between 1984 and 1987 were 1.9 and 1.6
percent respectively; and between 1981 and 1983
were 3.6 and 2.8 percent respectively (Cromwell
and Puskin, 1989)
d. 729 hospitals across the nation responded to
Medicare PPS by reducing the use of lab tests,
x-rays and diagnostic tests across all DRG
categories between 1983 and 1984 (Long et al.,
1987)
e. 3,000 hospitals nation wide responded to managed
care contracts in 1987 by reducing non-physician
FTEs in their Ancillary and Nursing Service Areas
(Ozcan and Luke, 1993)
Although the studies above provide some evidence that
hospitals have responded to Medicare PPS and managed care
contracting by increasing labor efficiency, none of the
studies examined the effect on changes in labor efficiency
beyond 1987. Moreover, most of the studies did not use DEA
to measure labor efficiency. Therefore, research beyond
1987 which uses DEA to measure labor efficiency would
provide insight regarding the long term effects of rate
regulation and price competition on changes in labor
efficiency over time.
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
V. H Y P O T H E S E S
Until the 1980s, prices driven by cost-based fee-for-
service reimbursement was considered a significant cause of
market failure in the health care industry. Under economic
theory, firms within competitive environments will attempt
to minimize the quantity and cost of inputs in maximizing
outputs given a market driven price constraint. Under
retrospective cost-based reimbursement, the health care
market lacked a market driven price constraint. Rate
regulation and price competition have been two approaches
to impose price constraints on the health care market.
Studies prior to the implementation of Medicare PPS, in
1983, revealed that hospitals in areas of high competition
tended to have higher expenditures, than hospitals in areas
of low competition (Joskow, 1980; Robinson and Luft, 1985
and 1987; Luft, 1986; Wilson and Jadlow, 1982; and Farley,
1985) . Studies in California, following 1983, indicated
that hospitals responded to Medicare PPS (rate regulation)
and public and private managed care contracting (price
competition) by reducing total hospital expenditures and
reducing the rate of increase in expenditures. Moreover,
given payer-dominated markets, hospitals in areas of high
competition responded to a greater degree, than hospitals
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in areas of low competition (Zwanziger, 1987; Zwanziger et
al., 1988, 1994a, 1994b; Phibbs and Robinson, 1989).
However, the studies did not indicate if the reduction in
hospital expenditures was due, in part, to a reduction in
labor FTEs.
A few studies have indicated that hospitals have
responded to price constraints, under PPS and public and
private managed care contracting, by reducing labor hours
per discharge in Nursing Service Areas or by reducing hours
per service in Ancillary Service Areas (Cromwell and
Puskin, 1989; Ashby and Altman, 1992; Long et al., 1987 and
1990; Kidder and Sullivan, 1982; Rudnick, 1995) . The
reduction in the use of labor inputs per discharge or per
service indicates an increase in the efficient use of
labor.
A number of studies have applied Data Envelopment
Analysis (DEA) in the health care field to measure changes
in the use of labor and non-labor inputs in producing a
given level of hospital output (Ozcan and Luke, 1993;
Valdmanis, 1990; Sexton et al., 1989; Sherman, 1984;
Valdmanis and Grosskopf, 1987; Nunamaker, 1983). Using DEA
to measure the efficient use of labor FTEs in the Ancillary
Service Area and Nursing Service Area of hospitals, this
study will examine the following hypotheses:
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I. Hospitals with a high percent of Medicare
patients are more likely in areas of high market
competition, than in areas of low market
competition, to increase labor efficiency
overtime in their:
a. Ancillary Service Area
b. Nursing Service Area
II. Hospitals with a high percent of Medi-Cal
patients are more likely in areas of high market
competition, than in areas of low market
competition, to increase labor efficiency
overtime in their:
a. Ancillary Service Area
b. Nursing Service Area
III. Hospitals with a high percent of HMO patients are
more likely in areas of high market competition,
than in areas of low market competition, to
increase labor efficiency overtime in their:
a. Ancillary Service Area
b. Nursing Service Area
The percent of Medicare patients reflects the influence of
rate regulation on hospital labor efficiency. The percent
of HMO patients includes all patients that belonged to
private managed care plans. Given that hospitals are
expected to compete on the basis of price for contracts
with managed care plans, the percent of HMO patients or the
percent of Medi-Cal patients reflects the influence of
price competition on hospital labor efficiency.
Confirmation of the hypotheses would suggest that rate
regulation and price competition may create incentives for
hospitals to improve hospital labor efficiency.
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
VI. M E T H O D O L O G Y
Study Sample
This study examined 149 non-teaching, private for-
profit and non-profit hospitals. Table 1 contains the
percent distribution of hospitals by area of competition,
ownership and membership in a multi-hospital system.
TABLE 1
Percent Distribution of Study Hospitals
(N = 149)
Competition Ownership Multi-Hospital
Membership
Year High Low FP NP Members Non-Members
1983 63% 37% 39% 61% 47% 53%
1987 63% 37% 39% 61% 58% 42%
1991 63% 37% 39% 61% 61% 39%
In this study, hospitals were classified according to peer
groups defined by the office of Statewide Health Planning
and Development (OSHPD). Cluster analysis was used by
OSHPD to group acute care urban hospitals according to the
following criteria:
1. Number of licensed beds
2. Scope of Services
3. Case-mix
4. Percent of acute and intensive patient days
5. Percent of discharges that are mother and newborns
Cluster analysis is a mathematical approach for placing
similar hospitals together by minimizing the differences
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
between hospitals within each group and maximizing the
differences between groups (Bernstein, 1991). Considering
OSHPD classifications, the distribution of hospitals by bed
size for this study was as follows;
TABLE 2
Study Hospitals by Bed Size
Bed Size
HOSPITAL:
Number by Bee
3
Size
HOSPITALS
Percent by Bed Size
1983 1987 1991 1983 1987 1991
95-170 74 65 59 50% 44% 40%
171-270 46 50 54 31% 34% 36%
271-380 20 19 23 13% 13% 15%
281-500 9 15 13 6% 10% 9%
TOT HOS 149 149 149 100% 100% 100%
Dependent Variable
Changes in DEA Labor Efficiency Scores
The dependent variable of this study was the increase
or decrease in technical efficiency of labor FTEs overtime
in the Ancillary Service Area and Nursing Service Area of
the hospitals in the study. Data Envelopment Analysis (DEA)
was used to generate labor efficiency scores for the
Ancillary Service Area and Nursing Service Area.
DEA is a non-parametric, linear programming approach
for measuring technical efficiency in organizations of
similar characteristics. Given defined inputs and outputs.
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
DEA finds the most efficient combination of inputs to
outputs (the lowest number of inputs per output or the
greatest number of outputs given the level of inputs).
From these efficient combinations^ DEA creates a reference
set. All hospitals in the data set are compared to this
reference set and those hospitals that match the reference
set are deemed efficient and are assigned a score of 1.
Those hospitals that do not match the reference set are
deemed inefficient and are assigned a score less than 1 but
greater than 0. The lower the score, the more the hospital
uses excessive inputs to produce a given level of output.
DEA also provides information regarding the reduction
of inputs or increases in outputs that would be necessary
for the hospital to be deemed efficient (Silkman, 1986) .
There are a growing number of studies that have utilized
DEA to analyze technical efficiency in the health care
field (Ozcan, 1993; Valdmanis, 1987; Sherman, 1984; Sexton,
1989; Nunamalcer, 1983) .
Data Envelopment Analysis (DEA) was used to generate a
measure of labor efficiency for the Ancillary Service Area
and for the Nursing Service Area for the years 1983, 1987
and 1991. The DEA data set contained three years of input
and output data for 149 hospitals. DEA scores were
generated from this one data set containing 447 records
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(e.g., 149 x 3). Therefore, three scores were generated for
each hospital (e.g., 1983, 1987, 1991). The study will
examine changes in DEA scores during three time periods:
1983-1987
1987-1991
1983-1991
To determine an increase or decrease in DEA scores during
each time period, the DEA score in the later year was
subtracted from the DEA score in the previous year. DEA
scores with a negative difference were coded as a decrease
in efficiency. DEA scores with a positive difference, or
if the DEA score was 1 in both years, were coded as an
increase in efficiency.
DEA Inputs and Outputs
The input/output variables for each DEA model were as
follows :
1. Ancillary Service Area
a. Inputs: (full-time-equivalents (FTEs))
1. Management
2. Technicians
3. RNs
4. Aides
5. Clerks
6. Other/LVNs
b. Outputs :
1. Case-mix adjusted inpatient discharges
2. Outpatient visits
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2. Nursing Service Area (Routine Nursing Departments)
a. Inputs: (full-time-equivalents (FTEs))
1. Management
2. RNs
3. LVNs
4. Aides
5. Clerks
6. Other/Technicians
b. Outputs :
1. Case-mix adjusted inpatient discharges
The operation of a hospital consists primarily of Nursing
Routine Services, Ancillary Services, Ambulatory Services
and Administrative Services. The most expensive and labor
intensive areas of the hospital are the Ancillary and
Nursing Service Areas. To identify changes in the use of
labor FTEs due to prospective payment (rate regulation) and
public/private managed care contracting (price
competition), this analysis will determine a separate DEA
score for the Ancillary Service Area and the Nursing
Service Area of the hospitals in the study.
Ehreth et al. (1994) applied factor analysis to
determine the best measures of hospital inputs and outputs
in calculating DEA scores. They found that the input
category labor and the output categories case-mix adjusted
discharges and outpatient visits were significant
determinants of DEA efficiency scores.
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Although a number of studies have used patient days
as the output measure in determining DEA efficiency scores,
Dittman et al. (1991) point out that under a prospective
payment system, patient days are not an appropriate output
objective. The objective under PPS is to minimize rather
than maximize patient days. Therefore, maximizing adjusted
discharges is a more appropriate output variable for DEA.
Data Source: The Annual Hospital Financial Disclosure
Reports from the Office of Statewide Health Planning and
Development (OSHPD) for 1983, 1987 and 1991.
Case-Mix Adjusted DEA Outputs
The most critical assumption in DEA analysis is that
the decision making units (DMUs) to be compared are of
similar characteristics. To adjust for differences among
hospital service-mix and patient illness acuity, patient
discharges were case-mix adjusted, which would create a
level field of comparison. In this way, hospitals caring
for resource intensive patients would be credited with more
discharges, while those hospitals caring for less resource
intensive patients would be debited with fewer discharges.
Case-mix indices were generated separately for each
year. The calculation was based on a population of 243
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
private for-profit and non-profit, teaching and non
teaching California hospitals with 95 to 500 beds.
A number of health care related studies have found
that case-mix adjusted discharges are statistically
significant in calculating DEA efficiency scores (Ozcan,
1992 and 1993; Ehreth, 1994). The calculation for case-mix
was as follows:
1. DRG weight =
(average DRG charge)/(average total charges)
2. Hosp. total weight =
sum of(# hosp. DRGs x DRG weight)
3. Hosp. case-mix =
(hosp. tot. weight)/(hosp. tot. # patients)
The calculation for adjusted discharges was as follows:
Adjusted Discharges = (case-mix index)x(# discharges)
Among analysts in the health care field, there is
controversy over the best method for calculating case-mix
indices. This controversy centers upon the best method for
calculating DRG weights (Horn and Schumacker, 1982). DRG
weights may be based upon what a hospital charges or on
what it cost the hospital to provide care for a patient in
a particular DRG. A study by the Federal Health Care
Financing Administration (HCFA) compared both cost and
charge methods for calculating DRG weights (Cotterill et
al., 1986). The study found little difference in the case-
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
mix indices since the index is a reflection of whether a
hospital treats a greater or lesser number of intensive or
less intensive DRGs than another hospital. As long as the
same measures are applied to all hospitals in the analysis,
this relativity does not appear to be significantly
effected.
Data Source: The Patient Discharge Data File from
the Office of Statewide Health Planning and Development
(OSHPD) for 1983, 1987 and 1991.
DEA Model Assumptions
The computer software used to calculate the DEA
efficiency scores for this study was the Integrated Data
Envelopment Analysis System (IDEAS). IDEAS was developed
by Agha Iqbal Ali, Ph.D., Professor of Operations
Management in the School of Management at the University of
Massachusetts at Amherst. He serves as Associate Editor
for the Journal of Operations Research and has written a
number of articles on Data Envelopment Analysis (Ali et
al., 1993a, 1993b, 1993c, 1994 and 1995).
IDEAS provides the ability to perform DEA using a
number of different models. Each of the DEA models
evaluates efficiency on the basis of three essential
components: orientation, envelopment surface, and relative
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
trade off implicit in the multiplier lower bounds. IDEIAS
allows for up to 12 models based on different combinations
of these components. The envelopment surface may reflect
constant returns to scale (CRS) or variable returns-to-
scale (VRS) . In addition, a DEA model may have an input
orientation, an output orientation or no orientation. The
evaluation system that governs the implicit pricing
mechanism in the model may be either standard or units-
invariant.
The components of the DEA models in this study
included a variable returns to scale (VRS) envelopment
surface with an input orientation and an units invariant
evaluation pricing mechanism. It should be noted that the
term "price" in DEA does not refer to dollars but rather to
the weights of inputs and/or outputs generated by the
model. Weights determine the amount of inputs and/or
outputs that must be decreased or increased in order for a
firm to become efficient. IDEAS allows input or output
weights (called price ratios) to be imposed on the model to
alter substitution assumptions regarding inputs and/or
outputs used in the production process under investigation.
Imposing these substitution ratios in essence customizes
the models to more accurately reflect the internal workings
of the organization or department.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Envelopment Surface
A VRS envelopment surface was selected for the DEA
models of this study. The envelopment surface is
equivalent to the reference set against which all hospitals
will be compared and determined to be efficient or
inefficient. As Ali writes:
Each of the various models for data envelopment
analysis (DEA) seeks to determine which of the n
decision-making units (DMUs) determine an
envelopment surface. This envelopment surface
is referred to as the eirpirical production
function or the efficiency frontier. DEA
provides a coit^rehensive analysis of relative
efficiency for multiple-input multiple-output
situations by evaluating each DMU and measuring
its performance relative to an envelopment
surface composed of other DMUs. Units that lie
on (determine) the surface are deemed efficient
in DEA terminology. Units that do not lie on
the surface are termed inefficient and the
analysis provides a measure of their relative
efficiency. (Ali, 1993c)
The two basic types of envelopment surfaces in DEA are
Constant Returns to Scale (CRS) and Variable Returns to
Scale (VRS). Which surface is most appropriate is
"frequently determined (dictated) by economic and other
assumptions regarding the data set to be analyzed" (Ali et
al.f 1993c). For this study, the VRS model was chosen
since a variable rate of return is a more realistic
representation of production reality. In contrast, a
constant rate of return (CRS) would assume that an
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
additional input produces the same amount of output. For
example, a CRS surface would assume that if 10 RNs can
produce 40 discharges, then 100 RNs should produce 400
discharges. In reality, however, 100 RNs may produce only
200 discharges or perhaps 600 discharges. In DEA, a
variable rate of return (VRS) would allow for either an
increasing or decreasing returns to scale. This type of
variability in return is a more accurate reflection of
operational realities. Therefore a VRS envelopment surface
was selected for the models of this study.
Input/Output Orientation
An input-orientation was chosen for the DEA models of
this study. Measures of efficiency for DMUs address the
discrepancy between the observed point and the projected
point on the envelopment surface. For any DMU, a projected
point for each input/output combination lies on the
envelopment surface and is the optimal (minimum) value for
that DMU. The observed point is the actual point at which
that input/output combination lies. For inefficient DMUs,
the distance between the projected point and the observed
point determines the degree of inefficiency. This
discrepancy between the observed and projected points
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
provides information regarding how much the DMU must
decrease its input, increase its output or both.
One of the options in building a DEA model is to
choose among an input orientation, an output orientation or
a non-orientation component. Any of the three orientations
can be used with either a CRS or VRS envelopment surface.
An oriented model (versus a non-oriented model) does not
alter whether a DMU is efficient or not efficient.
However, it does alter the information regarding
inefficient DMUs. An oriented model changes the distance
between the observed point and the projected point. This
distance provides information regarding how much a DMU must
increase or decrease its inputs and/or outputs in order to
become efficient (Ali, 1993c).
In this study, it is assumed that the objective of
hospitals under price competition and rate regulation is to
reduce inputs rather than increase outputs. Therefore, the
orientation component of the DEA models of this study was
an input-orientation.
Evaluation System
An units-invariant evaluation system was selected for
the DEA models of this study. Another component of the DEA
model is the relative value of the marginal worth of the
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
excess inputs or output slacks. The standard for
determining the marginal value of excess input and/or
output slack is a function of the evaluation system assumed
by the model. The two basic evaluation systems are the
standard or units-invariant.
The standard evaluation system assumes that the
marginal worth of each of the excess inputs and output
slack for a DMU is the same. Moreover, it assumes that the
marginal worth is the same across all DMUs (Ali, 1993c).
In regard to the units-invariant model, Ali writes
that:
The units-invariant models allow each DMU the
additional flexibility to select a base of
evaluation that is determined by its own particular
relative mix of inputs and outputs. In particular,
these models implicitly assume that the marginal
values of non-zero output slack and excess input
variables are not identical. The relative worth of
each output slack or input excess is gauged with
respect to the level of the output or input. This
not only distinguishes the relative worth among the
inputs and outputs for a particular decision making
unit but also distinguishes the relative worth of a
particular output or input across decision-making
units. (Ali, 1993c)
Given a units-invariant evaluation system, a hospital with
only two RNs would value the 2nd RN more than a hospital,
with 100 RNs, would value the 100th RN. In short, a
hospital with fewer excess inputs would value or weight
them more than a hospital with greater excess inputs.
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The evaluation system forms the basis for how a model
will determine the amount of inputs to be reduced or the
amount of outputs to be increased. However, "the
envelopment surface and the efficient/inefficient
classification remain the same for both the standard and
units-invariant formulations" (Ali, 1993).
The evaluation component of the DEA model chosen for
this study is the units-invariant. It is assumed that
hospitals do not put the same marginal weight or value on
their excess inputs and outputs as other hospitals.
Moreover, hospitals are more likely to weight or value
their excess inputs or output slacks relative to their own
volumes.
Categorical Variables
Bed size was defined as a categorical variable in the
DEA models of this study. The efficiency scores that the
DEA model generates are a function of the peer group
against which all DMUs are compared (e.g., the production
frontier) . A component of the DEA model that influences
the composition of the peer group is the presence of
categorical variables. Data may be coded into categories
according to some characteristic that improves the
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
comparability of DMUs by improving similarity among the
DMUs.
Studies have found that the use of categorical
variables can effect the outcome of efficiency scores.
Banker et al. (1986) compared the use of continuous versus
categorical treatment of market size in examining technical
and scale efficiency in 69 pharmacies. They found that
when market size was defined as a continuous variable, 52
pharmacies were found to have technical inefficiencies.
When market size was defined as a categorical variable, the
number of hospitals found to have technical inefficiencies
were reduced to 43. The difference appears to be due to
refining the similarity of the comparison or reference
group. The authors conclude that appropriate use of
categorical variables "strengthens the credibility of the
insights obtained about the improvements possible because
the referent composite DMUs consist only of DMUs which have
been matched more carefully with the DMU being evaluated"
(Banker and Morey, 1986). Similarly, studies by Ozcan
(1993) and Sexton et al. (1989) found that hospital bed
size was significantly related to DEA efficiency scores.
In this study, hospitals were classified according to
four bed size categories. In DEA, coding hospitals by bed
size assumes an implicit hierarchy of the categories. The
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
result is that each hospital will be compared against
hospitals in its own bed size category and the bed size
category below it. In this way smaller hospitals are not
expected to produce at the same level that larger hospitals
may be capable of producing.
Discretionary and Non-Discretionary Variables
All input and output variables were defined as
controllable (discretionary) variables in the DEA models of
this study. Discretionary or non-discretionary input or
output variables can alter the production frontier and
therefore efficiency scores. Discretionary variables are
controllable by the department manager while non-
discretionary variables are not controllable. In DEA,
variables can be defined as discretionary or non-
discretionary. By identify controllable and non-
controllable variables in the DEA model, "the marginal
worth of non-discretionary input excesses (and similarly
output slacks) should not enter into the evaluation of
inefficiency" (Ali, 1993c). Therefore, the output from
the DEA model will reflect "the most productive scale size
for the discretionary inputs given the fixed level of the
non-discretionary inputs" (Banker and Morey, 1985). Given
this information, managers can determine how much to reduce
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
inputs or augment outputs for those variable over which
they have control.
In this study, the inputs of the DEA model are full-
time-equivalent s (FTEs) . Given that the DEA model is an
input-oriented model, it is assumed that the department
managers have control over the staffing levels of their
departments. Therefore, all inputs and outputs are defined
as discretionary variables.
Substitution Ratios
The DEA models of this study did not assume any
substitution (price) ratios. One outcome of a DEA model is
information regarding substitution among inputs or among
outputs. Substitution constraints can be imposed on the
model that will alter the rates of substitutability. The
rates of substitutability are defined under Production
Theory as follows:
Production theory refers to these rates as marginal
rates of technical substitution (MRTS) and marginal
rates of production transformation (MRPT) where,
respectively, MRTSpg, is the rate at which input p
is substituted for input q while still producing
the same levels of outputs and keeping other inputs
constant; and MRPTvw represents the rate at which
output V must be sacrificed to obtain more of
output w while keeping the consumption of all
inputs and the level of all other outputs constant.
(Ali and Lerme, 1994)
Under DEA, the marginal rates of technical substitution or
product transformation can be determined and imposed for
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
existing production technology. However, input substitution
rates were not imposed on the DEA models for this study.
To impose substitution ratios would imply that FTEs in
hospital departments are substitutable for each other. For
example, given that technicians have very different
training and different tasks from RNs, substitutability
between RNS and technicians would not be feasible.
Similarly, it is possible that RNs may be able to do the
work of LVNs, aides or even clerks. However, given that RNs
are a more expensive input it would be contrary to economic
efficiency to impose such substitution ratios. It is
conceivable that certain tasks of one FTE category may be
allocated to another FTE category. However, determining
proportional substitution ratios for such tasks is beyond
the scope of this analysis. Therefore, input substitution
ratios were not imposed on the DEA models for this study.
Model Degeneracy
To reduce model degeneracy, LVNs were combined with
the category 'Other' in the DEA model for Ancillary Service
Areas. Similarly, technicians were combined with the
category 'Other' in the DEA model for Nursing Service
Areas.
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The number of input categories in the DEA model can
alter the efficiency frontier. By altering the frontier,
DEA scores for inefficient hospitals may increase.
Moreover, hospitals that were classified as inefficient may
be classified as efficient. However, model degeneracy would
not reclassify an efficient hospital as inefficient.
The degree of inefficiency is a function of the
distance between the projected point and the observed point
of the efficiency frontier. The effect of adding variables
to the model draws the frontier closer to the inefficient
DMUs and further increases their efficiency scores.
Therefore, "the more variables considered, the greater
chance some inefficient DMU will dominate on the added
dimension, and thus become efficient" (Nunamaker, 1983b).
The choice of variables for the DELA model should
reflect those inputs that most significantly impact the
production process under investigation. In the Ancillary
Service Area, LVNs were only 2 to 3 percent of the total
FTEs. Therefore, the 'LVN' FTE category was combined with
the 'Other' FTE category in the DEA model for the Ancillary
Service Area. Similarly, in the Nursing Service Area,
technicians were only 2 to 3 percent of the total FTEs.
Therefore, the 'Technician' FTE category was combined with
the 'Other' FTE category in the DEA model for the Nursing
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Service Area. When, the LVN FTE and technician FTE
categories were added as separate FTE categories to the
ancillary and nursing models respectively, the number of
hospitals that were classified as efficient increased.
Given that the LVN and technician categories contribute
such a small amount to the output, it was appropriate not
to include them as separate input categories in the DEA
models.
Control Variables
Ownership
Ownership was included in the regression models as a
control variable. Studies that have examined the
differences in cost and efficiency between for-profit and
non-profit hospitals tend to be inconclusive and
contradictory. Given their profit incentive, for-profits
are expected to exercise greater control over budgets and
expenditures and enforce more stringent controls over
personnel and facilities than non-profits (Larson, 1983).
However, Lewin et al. (1981) found that total operating
expenses per admission were 4 percent higher in for-
profits. Conversely, Sloan and Vraicu (1983) found that
for-profits had operating expenses 3 percent lower than
non-profits. Moreover, Friedman and Shortell (1988) did not
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
find any statistically significant relationship between
ownership type and hospital cost. In regard to labor
inputs. Sear (1991) found that for-profit hospitals had
fewer FTEs per bed and fewer manhours per adjusted patient
day. As these studies indicate, for-profit hospitals may or
may not consistently conform to theoretical expectations.
Given these inconsistencies between theory versus
study findings, ownership will be included as a control
variable in examining labor efficiency.
Data Source: OSHPD Annual Hospital Financial
Disclosure Report for 1983, 1987 and 1991.
Multi-Hospital Membership
Membership in a multi-hospital system was included as
a control variable in the regression models. Between 1981
and 1990, approximately 33 percent of the total non-
federal, acute care hospitals belonged to a multi-hospital
system (Feldstein, 1993). The total number of multi
hospital systems (172) remained relatively constant during
this time period. The objective of the multi-hospital
system is to improve operational efficiency by minimizing
transaction costs related to the production of services.
Feldstein defines transaction costs as follows:
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A hospital's transaction costs, or the costs of
controlling and coordinating the delivery of
medical services, involves the collection of
information about the different suppliers of each
of the components of medical services, the
negotiation of separate contracts with each set of
suppliers, monitoring compliance with the contract,
and the hospital's ability to enforce such
contracts. The necessity for frequent negotiations
with suppliers, uncertainty as to the supplier's
quality and the difficulty of monitoring a
supplier's behavior increase transaction costs to
the firm. (Feldstein, 1993)
Reduction of transaction costs is expected to occur through
horizontal and vertical integration of services.
Horizontal integration is the merging of a number of
hospitals into a single system. The extent of the
advantages to both the individual hospital and the larger
system depends on the degree of affiliation. At minimum,
the affiliation allows for economies of scale through joint
purchasing arrangements. A stronger affiliation may also
provide economies of scale such as lower interest costs,
access to capital markets, improved cash management and
lower malpractice premiums (Feldstein, 1993).
The multi-hospital system may also reduce transaction
costs by controlling and coordinating the delivery of
services through vertical integration. Under cost-based
reimbursement each service was reimbursed separately.
Therefore, providers would not be at financial risk for the
ordering practices of physicians. However, with the
implementation of Medicare prospective payment, hospitals
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
would be at risk for excess services ordered by physicians.
Similarly, hospitals became more dependent on nursing homes
and home health agencies to accept patients discharged
"quicker and sicker." Moreover, under price competition,
providers have been under pressure to minimize production
costs to stay price competitive when negotiating with
managed care plans. Multi-hospital systems have attempted
to lower transaction costs through ownership of, or
exclusive contracting with providers, suppliers and
insurers. Given the financial pressures under prospective
payment from both public and private payers, hospitals have
looked to multi-hospital systems to vertically integrate
services that are both complementary and substitutable to
their own inpatient services (Feldstein, 1993) .
Although there is much agreement in the literature
regarding the expected effects of multi-hospital systems,
there appear to be few empirical studies that test these
expectations. However, one study by Lynch and McCue found
that:
For-profit multi-hospital systems were able to
improve many of the financial and operating
problems of acquired facilities. In comparisons to
independent not-for-profit hospitals, acquired
hospitals were found to increase access to long
term debt, make inç>rovements to plant and
equipment, improve profitability and increase
efficiency to a greater extent. (Lynch and McCue,
1990)
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Given that membership in a multi-hospital system may result
in more efficient production of hospital services,
membership in a multi-hospital system was included as a
control variable.
Data Source: The American Hospital Association (AHA)
Hospital Statistics for 1983, 1987 and 1991.
Independent Variables
Market Competition
Hospital markets were defined as a 15 mile radius
around each hospital. Market competitiveness was defined
as the number of hospitals within a 15 mile radius from
each hospital. Hospitals were classified as either in
areas of high competition or in areas of low competition.
High competition was defined as more than 10 hospitals
within a 15 mile radius from the hospital and low
competition was defined as 10 or fewer hospitals within a
15 mile radius from the hospital.
A study by Phibbs (1995) at the Stanford University
Center for Health Care Evaluation examined different
definitions of hospital markets. The Hirschman-Herfindahl
index was calculated using different definitions of
hospital markets. Definitions of hospital markets are
generally based upon geographic boundaries or patient zip
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
codes. Phibbs applied each HHI, based on different
definitions of markets, to the same cost function. The
following market definitions of the Hirschman-Herfindahl
Index (HHI) were compared:
1. County
2. Fixed Radius: the number of hospitals within 15
miles of the hospital
3. Patient Zip Code Overlap: number of hospitals which
share at least 5 percent of patient zip codes
4. Patient Zip Code Specific: a HHI is calculated for
each zip code. Competitiveness is the weighted
average of the zip code HHIs, weighted by the
number of patients the hospital received from each
zip code
The comparative analysis found all HHIs to be highly
correlated (85 percent) regardless of how the market was
defined, except for the county definition of market.
Hospital markets defined as county were less sensitive to
differences in competitiveness. The highly correlated HHIs
produced similar results in the same cost function. Thus,
the study found that markets defined by zip code or by a 15
mile radius should provide similar results.
Studies by Luft et al. (1986a and 1986b) determined
statistically significant definitions of high and low areas
of competition. Areas of high competition were found to be
hospitals with more than 10 competitors within a 15 mile
radius of the hospital. Areas of low competition were
105
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
found to be hospitals with 10 or fewer competitors within a
15 mile radius. It is assumed that physicians generally do
not travel more than 15 miles between home, office and
hospitals in treating patients. Similarly, Phibbs and
Robinson (1989 and 1993) found that high and low
competition was statistically significant at above and
below 10 competitors within a 15 mile radius of the
hospital.
For this study, data on the number of hospitals within
a 15 mile radius was provided by the Institute for Health
Policy Studies. The most recent year available was 1988.
In this study, the 1988 figures were assumed for all three
years: 1983, 1987 and 1991. Given closings or openings of
hospitals within market areas, it is possible that some
hospitals may be misclassified in the time period before
and/or after 1988. However, given that this study assumes
that the point between high and low competition is 10
hospitals, only those hospitals that border on this cut-off
point would be effected by possible misclassification.
As indicated in Table A.l in Appendix A, the number of
hospitals that bordered on the cut-off point of 10
competitors (e.g., 9 to 12 competitors) was approximately
10 or 6.7 percent of the total sample. Hospitals in markets
with 9 or 12 competitors (5 or 3.4 percent) would have to
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
experience 2 closings or 2 openings to change their
classification of high or low, than would be assumed in the
1988 data. Similarly, hospitals in markets with 10 or 11
competitors (5 or 3.4 percent) would have to experience 1
closing or opening to change their classification of high
or low, than would be assumed in the 1988 data. Assuming
that not all 10 hospitals are likely to experience closings
or openings in their market area, the possible number of
hospitals potentially misclassified would be less than 6.7
percent of the sample size.
Given that data was not available beyond 1988 and
given that a very small percentage of hospitals of the
sample might be misclassified, 1988 data was assumed for
all three years of this study.
Data Source: California hospital neighbor data from
the Institute for Health Policy Studies for 1988.
Rate Regulation
Rate regulation was measured by the hospital's percent
of Medicare discharges in the base year of the time period
and the increase in the percent of Medicare discharges
during the time period.
Zwanziger et al. (1987, 1988, 1994a and 1994b)
examined the effects of hospital market competition and
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Medicare prospective payment on hospital cost behavior.
The effect of PPS was measured by an index based upon the
hospital's percent of Medicare discharges. Similarly,
Ozcan and Luke (1993) examined the relationship between
technical efficiency and various hospital characteristics.
The effect of PPS was measured by the hospital's percent of
Medicare discharges.
Data Source; OSHPD Patient Discharge Data File for
1983, 1987 and 1991.
Price Competition
Price competition was measured by four variables: the
hospital's percent of Medi-Cal discharges in the base year
of the time period and the increase in the percent of Medi-
Cal discharges during the time period; and the hospital's
percent of HMO discharges in the base year of the time
period and the increase in the percent of HMO discharges
during the time period. The category of HMO included all
discharges from private managed care plans of the hospital.
Phibbs and Robinson (1989) examined the effects of
price competition on the rate of cost inflation for
inpatient services of California's Medi-Cal program. Price
competition was measured by the hospital's percent of Medi-
Cal discharges. In other studies, Zwanziger et al. (1988
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and 1994b) examined the relationship between price
competition and changes in hospital expenses. Again, the
percent of Medi-Cal and HMO discharges represented price
competition.
Data Source: OSHPD Patient Discharge Data File for
1983, 1987 and 1991.
Interaction Terms
The hypotheses of the study were measured by the
interaction terms between the variable for hospital market
competition and the variables for payer. Controlling for
ownership and membership in a multi-hospital system, the
study hypothesizes that hospitals with a high percent of
Medicare, Medi-Cal or HMO patients are more likely in areas
of high market competition, than in areas of low market
competition, to have an increase in labor efficiency,
overtime, in their a)Ancillary Service Area and b)Nursing
Service Area. To measure the effect of payer in areas of
high versus low market competition, interaction terms
between hospital market competition and payer were added to
the regression models. The likelihood of increases in labor
efficiency was analyzed for three time periods:
1983-1987
1987-1991
1983-1991
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
For each time period, each model contained two types of
interaction terms. The interaction terms differed in how
the payer variable was defined. In one interaction term,
the payer variable (HMO, Medi-Cal, Medicare) was defined as
the percent of discharges in the base year of the time
period. In the second interaction term, the payer variable
was defined as the increase in the percent of discharges
during the time period. The measure of market competition
was defined as the number of hospitals within a 15 mile
radius of the hospital. The interaction terms were defined
as follows:
Interaction Term I:
(Competition) x (Percent of discharges in base year of
time period for each payer)
Interaction Term II:
(Competition) x (Increase in the percent of discharges
during time period for each payer)
The first interaction term measured whether hospitals that
had a high percent of discharges in a particular payer
category in the base year were more likely in areas of high
market competition, than in areas of low market
competition, to have had an increase in labor efficiency
during the time period. The second interaction term
measured whether hospitals that had an increase in the
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
percent of discharges in a particular payer category during
the time period were more likely in areas of market high
competition, than in areas of low market competition, to
have had an increase in labor efficiency during the time
period.
Logistic Regression Models
Logistic regression was chosen to test the
relationship between rate regulation and labor efficiency,
and price competition and labor efficiency. SPSS computer
software was used to test each variable for linearity and
normality and to run the logistic regression models.
Multiple regression was first attempted but the data
grossly violated the assumptions of normal distribution of
the dependent and independent variables. In addition,
there was little significant linear relationship between
the dependent variable and each independent variable.
Logistic regression predicts the likelihood of an
event occurring (change in DEA labor efficiency score)
given the presence of selected independent variables
(percent discharges by Medicare, Medi-Cal and HMO) and
control variables (ownership and multi-hospital
membership).
Ill
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Model Variables and Coding
Logistic regression models for the Ancillary Service
Area and Nursing Service Area were run for each time
period, which resulted in six models indicated in Table 3:
TABLE 3
Logistic Regression Models
Ancillary Service
Area
Nursing Service
Area
1983-87 A1 N1
1987-91 A2 N2
1983-91 A3 N3
The regression models were as follows:
A1 = f (A83, MCR83, MCL83, HM083, MEM83, OWN, MCRl, MCLl,
HMOl, MEMl, COMPT, COMPTxMCR83, COMPTxMCL83,
C0MPTXHM083, COMPTxMCRl, COMPTxMCLl, COMPTxHMOl)
A2 = f(A87, MCR87, MCL87, HM087, MEM87, OWN, MCR2, MCL2,
HM02, MEM2, COMPT, COMPTxMCR87, COMPTxMCL87,
C0MPTXHM087, C0MPTxMCR2, C0MPTxMCL2, C0MPTxHM02)
A3 = f (A83, MCR83, MCL83, HM083, MEM83, OWN, MCR3, MCL3,
HM03, MEM3, COMPT, COMPTxMCR83, COMPTxMCL83,
C0MPTXHM083, C0MPTxMCR3, C0MPTxMCL3, C0MPTxHM03)
N1 = f(N83, MCR83, MCL83, HM083, MEM83, OWN, MCRl, MCLl,
HMOl, MEMl, COMPT, COMPTxMCR83, COMPTxMCL83,
C0MPTXHM083, COMPTxMCRl, COMPTxMCLl, COMPTxHMOl)
N2 = f(N87, MCR87, MCL87, HM087, MEM87, OWN, MCR2, MCL2,
HM02, MEM2, COMPT, COMPTxMCR87, C0MPTxMCL87,
C0MPTXHM087, C0MPTxMCR2, C0MPTxMCL2, C0MPTxHM02)
N3 = f(N83, MCR83, MCL83, HM083, MEM83, OWN, MCR3, MCL3,
HM03, MEM3, COMPT, COMPTxMCRS3, COMPTxMCL83,
C0MPTxHM083, C0MPTxMCR3, C0MPTxMCL3, C0MPTxHM03)
The dependent and independent variables for the regression
models were coded as indicated in Table 4 :
112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CD
■ D
O
Q .
C
g
Q .
■ D
CD
C/)
C/)
8
( O '
TABLE 4
LOGISTIC REGRESSION MODEL VARIABLES AND CODING
3
3"
CD
CD
■ D
O
Q .
C
a
o
3
" O
o
CD
Q .
■ D
CD
C/)
C/)
VARIABLE DEFINITION 1983-87 1987-91 1983-91 CODE
Chanae in DEA Score A1 A2 A3 1 = Increase or no change in
Dependent for Ancillary and Nursing N1 N2 N3 DEA score of 1
Service Areas 0 = Decrease in DEA Score
Base Year DEA Score A83 A87 A83
Independent Ancillary & Nursing N83 N87 N83
Percent of discharges in MCR83 MCR87 MCR83
Independent base year of time period MCL83 MCL87 MCL83
HM083 HM087 HM083
Increase in oercent of MCRl MCR2 MCR3 1 = Increase in % discharges
Independent discharges during time MCLl MCL2 MCL3 0 = decrease in % discharges
period HMOl HM02 HM03
1 = 1 1 or more competitors
Independent Market Comoetition COMPT COMPT COMPT within 15 miles
0 = 10 or less competitors
within 15 miles
Control Ownership OWN OWN OWN 1 = for-profit
0 = non-profit
Multi-hospital system
Control membershio in base vear of MEM83 MEM87 MEM83 1 = member
time period 0 = non-member
Hospitals that became 1 = became members
Control members of a multi-hospital MEMl MEM2 MEM3 0 = no membership change
system during time period during time period
Interaction Comoetition variable times COMPTxMCR83 COMPTXMCR87 COMPTxMCR83
Terms the oercent of discharges In COMPTXMCL83 COMPTXMCL87 COMPTXMCL83
Base Year base year of time period C0MPTXHM083 COMPTXHM087 C0MPTXHM083
Interaction Comoetition variable times
Terms increase or decrease in COMPTxMCRl COMPTxMCR2 COMPTXMCR3
Change in oercent of discharges during COMPTxMCLl COMPTXMCL2 COMPTXMCL3
Percent time period COMPTxHMOl C0MPTxHM02 C0MPTXHM03
113
VII. R E S U L T S
Regression Model Diagnostics
The variables in each Logistic Regression Model were
tested for collinearity. Table B.l in Appendix B contains
the tolerance scores for each variable in each model. A
variable with a tolerance score less than (.1) indicates
that the variable is highly correlated with another
variable in the model. Included in Appendix B are
correlation matrices for each regression model. The
correlation matrices identifies which variables are highly
correlated with those variables that have a tolerance score
less than (.1). In the correlation matrices, a score close
to (1 or -1) indicates that the variables are highly
correlated.
When variables are highly correlated, the presence or
absence of one of the highly correlated variables in the
model may change the significance of the other variable
that it is highly correlated with. However, when
interaction terms are included in the regression model, it
is not uncommon and is usually expected to find high
correlation between the interaction term and one of its
main effects. In each model in this study, some
collinearity was found between interaction terms and one of
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
its main effects. The pairs of terms found to be highly
correlated for each model were as follows;
TABLE 5
MODEL
HIGHLY CORRELATED TERMS
OF REGRESSION MODELS
A1 COMPT/COMPTxMCRS 3HM08 3/COMPTxHMOS 3
A2 COMPT/COMPTxMCRS 7HMO87/COMPTxHMO83
A3 COMPT/COMPTxMCRS 3 C0MPT/C0MPTXHM03 HMO83/COMPTxHMO83
N1 COMPT/COMPTxMCRS 3HMOS3/COMPTxHMOS3
N2 COMPT/COMPTxMCRS 7HMO87/COMPTxHMO83
N3 COMPT/COMPTxMCRS3 COMPT/C0MPTXHM03 HM083/COMPTxHMOS3
As indicated in Table 5, the interaction terms are highly
correlated with one of their main effects. When both terms
are in the same model, each term will exhibit a certain
level of significance. However, if one of the terms of the
highly correlated pair is dropped from the model, the
significance of the term that it is highly correlated with
may change. Therefore, to test if the significance of each
term in the highly correlated pair changes when the other
term is dropped, several iterations of each model were run.
Each iteration dropped one of the terms found to be highly
correlated with another term. If the significance of one
highly correlated term did not change when the term it was
correlated with was dropped, then both terms where kept in
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the final model- If one of the correlated terms became
significant in the absence of the term it was highly
correlated with, then the term that was significant was
kept in the final model while the non-significant term was
dropped. If each term of the highly correlated pair are
significant in the absence of the other but both terms are
not significant when in the model together, the terms are
interpreted as if significant when in the model together.
In this study, only model N2 required omitting an
interaction term (COMPTxHMOS?) since its main effect
(HM087) was significant in the absence of the interaction
term.
When the interaction term and one of its main effects
are both statistically significant, the interpretation of
the interaction term is a function of the sum of the beta
scores for the interaction term and its significant main
effect.
If the interaction term is significant, but both of
its main effects are not significant, then the
interpretation of the interaction term pertains only to the
portion of the term coded as 1. More specifically, there
is no effect on the portion of the term coded as 0.
Therefore, since high competition was coded as 1 in the
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
term COMPT, the interpretation of the interaction term
would pertain only to areas of high competition.
DEA Scores
The average DEA score for all hospitals combined is
shown in Figures 1 and 2 in Appendix C. In both the
Ancillary and Nursing Service Areas the average DEA score
declined during each time period. The slope of the decline
is greatest for the Ancillary Service Area between 1983 and
1987 (.89 to .82). The slope of the decline for the
Nursing Service Area between 1983 and 1991 is relatively
flat (.82 to .80).
As indicated in Table 6, the percent of hospitals with
a DEA score of 1 in their Ancillary Service Area decreased
between 1983 and 1987 then rose between 1987 and 1991.
TABLE 6
DEA EFFICIENCY SCORES
Hospitals with a DEA Score of 1
ANCILLARY NURSING
Number Percent Number Percent
1983 69 46% 51 34%
1987 38 26% 48 32%
1991 47 32% 48 32%
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overall, however, the percent of hospitals with a DEA score
of 1 in their Ancillary Service Area decreased a total of
14 percent between 1983 and 1991.
The percent of hospitals with a DEA score of 1 in
their Nursing Service Area slightly decreased between 1983
and 1987 then stayed constant between 1987 and 1991.
Overall, the percent of hospitals with a DEA score of 1 in
their Nursing Service Area declined a total of 2 percent
between 1983 and 1991.
As indicated in Table 7, the percent of hospitals with
an increase in their DEA score in their Ancillary Service
Area increased between 1983 and 1987 but then decreased
between 1987 and 1991.
TABLE 7
INCREASE IN DEA EFFICIENCY SCORES
Hospitals with an Increase in DEA Scores
During Time Periods
ANCILLARY NURSING
Number Percent Number Percent
1983-87 58 39% 80 54%
1987-91 83 56% 85 57%
1983-91 64 43% 79 53%
Overall, the percent of hospitals with an increase in their
DEA score for their Ancillary Service Area increased a
total of 4 percent between 1983 and 1991.
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The percent of hospitals that had an increase in their
DEA score in their Nursing Service Area rose slightly
between 1983 and 1987 and declined slightly between 1987
and 1991. Overall, the percent of hospitals that had an
increase in their DEA score for their Nursing Service Area
declined a total of 1 percent between 1983 and 1991.
The average DEA score by bed size category is shown in
Figures 3 and 4 of Appendix C. In the Ancillary Service
Area, small hospitals (95-170 beds) had a decline in the
average DEA score in each year. The average DEA scores for
medium (171-170 beds) and large (271-380 beds) hospitals
declined between 1983 and 1987 but then rose in 1991.
Overall, however, the average DEA score for large hospitals
declined between 1983 and 1991. The average DEA score for
extra-large hospitals (381-500 beds) was relatively
constant in 1983 and 1987 then rose in 1991.
In the Nursing Service Area, small size hospitals had
a relatively constant average DEA score in each year. The
average DEA score for medium size hospitals slightly
declined each year. Similarly, the average DEA score for
large hospitals steadily declined in each year. The average
DEA score for extra-large hospitals increased each year.
Figures 5 and 6 in Appendix C show the percent of
hospitals by bed size that had an increase in their DEA
119
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
scores out of the total number of hospitals that had an
increase in DEA scores during each time period. In the
Ancillary Service Area, small hospitals had the highest
percent of hospitals to have an increase in their DEA score
between 1983 and 1987. However, between 1987 and 1991
medium size hospitals had the highest percent of hospitals
to have an increase in their DEA score. Overall, medium
size hospitals had the highest percent of hospitals to have
an increase in their DEA score between 1983 and 1991. The
percent of large hospitals to have an increase in their DEA
score was relatively constant during the time periods 1983-
1987 and 1987-1991 but overall the percent slightly
increased between 1983 and 1991. The percent of extra-
large hospitals to have an increase in their DEA score rose
during each time period.
In the Nursing Service Area, small hospitals had the
largest percent of hospitals to have an increase in their
DEA score during each time period. The percent of medium
size hospitals to have an increase in their DEA score was
relatively constant during the time periods 1983-1987 and
1987-1991 but overall the percent increased between 1983
and 1991. The percent of large hospitals to have an
increase in their DEA score was relatively constant during
the time periods 1983-1987 and 1987-1991 but overall the
120
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
percent decreased between 1983 and 1991. The percent of
extra-large hospitals to have an increase in their DEA
score increased during the time periods 1983-1987 and 1987-
1991 but overall the percent of hospitals slightly
decreased between 1983 and 1991.
Tables 8a and 8b contain the average percent that
inefficient hospitals would have to reduce their various
labor categories to obtain an efficiency score of 1 in
their Ancillary and Nursing Service Areas.
TABLE 8a
ANCILLARY SERVICE AREA
Average Percent Reduction of FTEs
in Inefficient Hospitals
N Hosp Mangmnt Techs RNs Aides Clerks Other Total
1983 80 54% 29% 28% 23% 30% 28% 26% 29%
1987 111 74% 33% 32% 27% 33% 34% 30% 33%
1991 102 68% 35% 34% 34% 38% 35% 34% 37%
TABLE 8b
NURSING SERVICE AREA
Average Percent Reduction of FTEs
in Inefficient Hospitals
N Hosp Mangmnt RNs LVNs Aides Clerks Other Total
1983 98 66% 42% 28% 34% 36% 48% 31% 36%
1987 101 68% 41% 29% 33% 34% 46% 35% 35%
1991 101 68% 38% 31% 33% 36% 41% 43% 36%
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Between 1983 and 1991, inefficient hospitals would have to
reduce, on average, between 29 to 37 percent of total labor
to obtain a DEA score of 1 for their Ancillary Service
Area. Similarly, between 1983 and 1991, inefficient
hospitals would have to reduce, on average, between 35 to
36 percent of total labor to obtain a DEA score of 1 in
their Nursing Service Area. In the Ancillary Service
Areas, the labor categories were relatively equal in the
required average percent reduction (28-35 percent) over the
three time periods. In the Nursing Service Areas, the labor
categories with the greatest required average percent
reduction appears to be in the management (38-42 percent)
and clerk (41-48 percent) categories. The smallest required
average percent reduction is in the RN category (28-31
percent).
Summary
In sum, the average DEA score for all hospitals
combined, declined in each of the three years for both
Ancillary and Nursing Service Areas. For both the Ancillary
and Nursing Service Areas, the average DEA score for small,
medium and large hospitals declined between 1983 and 1991.
However, the average DEA score for hospitals with 381-500
beds increased between 1983 and 1991. In addition, between
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1983 and 1991 the percent of hospitals that had an increase
in their DEA scores increased 4 percent for Ancillary
Service Areas and decreased 1 percent for Nursing Service
Areas. On average, hospitals would have to reduce their
total FTEs by approximately 36 percent in their Ancillary
Service Area and 36 percent in their Nursing Service Area
to obtain a DEA score of 1. Tables D.l and D.2 in Appendix
D contain the percent reduction in FTEs by percentile for
each model for each time period.
Regression Results
The results of each logistic regression model are
contained in Appendix E. The value under the significance
column (SIG) identifies a statistically significant
relationship between the dependent variable and the
independent variable. If the (SIG) value is less than .05,
then the variable was found to be significantly correlated
with the dependent variable. The beta coefficients
represent the direction and magnitude of the likelihood of
an increase or decrease in the dependent variable given an
increase in the independent variable. For example, if the
beta is positive, then an increase in the independent
variable (a value of 1) would increase the likelihood of an
increase in labor efficiency. If the beta is negative.
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
then an increase in the independent variable (a value of 1)
would increase the likelihood of a decrease in labor
efficiency.
The odds ratio indicates an increase or decrease in
the odds of the dependent variable occurring. An odds ratio
greater than 1 would mean that the odds are increased and
an odds ratio less than 1 would mean that the odds are
decreased. For example, an odds ratio of 1.5 would mean
that given an unit increase in the independent variable,
the odds that hospitals had an increase in labor efficiency
would be increased by 50 percent. An odds ratio of .80
would mean that given an unit increase in the independent
variable, the odds that hospitals had an increase in labor
efficiency would be decreased 20 percent (e.g., 1-.8).
Goodness of Fit
The Goodness-of-Fit value indicates how well the
observed outcome compares to the predicted outcome of the
model. The percentage represents the observed outcome as a
percent of the predicted outcome. For example, a score of
100 percent would indicate that the observed outcome and
the predicted outcome were identical. Table 9 contains the
Goodness-of-Fit values for each of the regression models.
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE 9
Regression Models
Goodness of Fit Scores
REGRESSION MODELS GOODNESS OF FIT
A1 71.14%
A2 68.46%
A3 68.46%
N1 72.48%
N2 68.46%
N3 73.83%
The goodness of fit values indicate that the observed
outcome was approximately 70 percent of the predicted
outcome. This would suggest that the models fit the data
relatively well.
Hypothesis I
The first hypothesis of the study was as follows:
I. Hospitals with a high percent of Medicare patients
are more likely in areas of high market
competition, than in areas of low market
competition, to increase labor efficiency overtime
in their:
a. Ancillary Service Area
b. Nursing Service Area
The variables that tested these hypotheses were the two
interaction terms in the Tables 10a and 10b. The
interaction term COMPTxMCR(83,87,91) measured whether
hospitals with a high percent of Medicare patients in the
base year of the time period were more likely in areas of
125
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
high market competition, than in areas of low market
competition, to have had an increase in labor efficiency
during the time period. The interaction term
COMPTxMCRd, 2,3) measured whether hospitals with an
increase in the percent of Medicare patients during the
time period where more likely in areas of high market
competition, than in areas of low market competition, to
have had an increase in labor efficiency during the time
period. The scores in the tables were taken from the
regression models in Appendix E.
The scores in Table 10a indicate that in the Ancillary
Service Area there were no statistically significant
relationships between Medicare and labor efficiency for any
of the three time periods. Therefore, the results did not
confirm the hypothesis.
The scores in Table 10b indicate that in the Nursing
Service Area, the interaction term C0MPxMCR2 was
statistically significant between 1987 and 1991 (Model N2).
Given that neither of the main effects were significant,
the sum of the beta scores would be as follows:
BETAS
MAIN EFFECTS -0-
C0MPTXMCR2 + -2.4543
C0MPTXMCR2 = -2.4543
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
since the main effects were not significant, there was no
relationship between an increase in the percent of Medicare
patients and an increase in labor efficiency in areas of
low market competition. However, in areas of high market
competition, there was a relationship. The beta of the
interaction term was negative, which indicates that in
areas of high competition, hospitals that had an increase
in the percent of Medicare patients between 1987 and 1991
were more likely to have had a decrease in labor efficiency
during that time period, than hospitals that had a decrease
or no change in their percent of Medicare patients. This
was in the opposite direction of the hypothesis and
therefore did not confirm the hypothesis.
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CD
■ D
O
Q.
C
g
Q.
■ D
CD
I
C / )
TABLE 10a
c/)
o '
3
ANCILLARY - REGRESSION COEFFICIENTS
O
3
CD
MEDICARE
8
■ D
ë '
MODEL Al
1983-87
MODEL A2
1987-91
MODEL A3
1983-91
i
TERM BETA ODDS
RATIO
TERM BETA ODDS
RATIO
TERM BETA ODDS
RATIO
3
CD
"n
c
3 .
3"
CD
COMPT
MCR83
MCRl
COMPTXMCR83
COMPTxMCRl
.1249
- .0234
.3078
- .0121
- .3196
1.1330
.9768
1.3604
.9879
.7264
COMPT
MCR87
MCR2
COMPTXMCR87
C0MPTXMCR2
1.4474
- .0432
- .3768
- .0059
.1281
4.2519
.9578
.6860
.9941
1.1366
COMPT
MCR83
MCR3
COMPTXMCR83
C0MPTXMCR3
.5711
- .0511
- .0113
.0180
- .3159
1.7703
.9502
.9887
1.0182
.7291
CD
■ D
O
Q.
C
a
TABLE 10b
3
■ D
NURSING - REGRESSION COEFFICIENTS
O
3"
C T
1 —H
MEDICARE
CD
Q.
g
1 —H
3"
O
MODEL NI
1983-87
MODEL N2
1987-91
MODEL N3
1983-91
■O
TERM BETA ODDS
RATIO
TERM BETA ODDS
RATIO
TERM BETA ODDS
RATIO
3
C / )
c /)
o "
3
COMPT
MCR83
MCRl
COMPTXMCR83
COMPTxMCRl
- .7449
- .0140
1.0571
.0167
-1.1587
.4748
.9861
2.8782
1.0169
.3139
COMPT
MCR87
MCR2
COMPTXMCR87
C0MPTXMCR2
- .2827
.0084
.9532
.0215
-2.4543
.7537
1.0084
2.5941
1.0217
* .0859
COMPT
MCR83
MCR3
COMPTXMCR83
C0MPTXMCR3
.7062
- .0639
1.2270
.0685
-1.7607
2.0262
.9381
3.4108
1.0709
.1719
* Significance Level < .05
128
Hypothesis II
The second hypothesis of the study was as follows:
II. Hospitals with a high percent of Medi-Cal patients
are more likely in areas of high market
competition, than in areas of low market
competition, to increase labor efficiency
overtime in their:
a. Ancillary Service Area
b. Nursing Service Area
The variables that tested these hypotheses were the two
interaction terms in the Tables 11a and lib. The
interaction term COMPTxMCL(83, 87, 91) measured whether
hospitals with a high percent of Medi-Cal patients in the
base year of the time period were more likely in areas of
high market competition, than in areas of low market
competition, to have had an increase in labor efficiency
during that time period. The interaction term
COMPTxMCL(1,2,3) measured whether hospitals with an
increase in the percent of Medi-Cal patients during the
time period where more likely in areas of high market
competition, than in areas of low market competition, to
have had an increase in labor efficiency during the time
period. The scores in the tables were taken from the
regression models in Appendix E.
The scores in Table 11a indicate that in the Ancillary
Service Area, there were no statistically significant
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
relationships between Medi-Cal and labor efficiency for any
of the three time periods. Therefore, these results did
not confirm the hypothesis.
The scores in Table 11b indicate that in the Nursing
Service Area the terms MCL3, COMPTxMCL83, and C0MPTxMCL3
were found to be statistically significant for the time
period 1983 to 1991.
Given that neither of the main effects were
significant for the interaction term COMPTxMCL83, the sum
of the beta scores would be as follows:
BETAS
MAIN EFFECTS -0-
COMPTXMCL83 + .1144
COMPTxMCL83 = .1144
Since the main effects were not significant, there was no
relationship between the percent of Medi-Cal patients in
1983 and labor efficiency in areas of low market
competition. However, there was a relationship in areas of
high market competition. The beta of the interaction term
is positive which indicates that in areas of high
competition, hospitals that had a high percent of Medi-Cal
patients in 1983 were more likely to have had an increase
in labor efficiency between 1983 and 1991, than hospitals
with a low percent of Medi-Cal patients in 1983. The
results indicate that Medi-Cal had an effect in areas of
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
high competition and in the hypothesized direction while
having no effect in areas of low competition. Although,
the results are in the hypothesized direction, the results
did not confirm the hypothesis that hospitals are more
likely in areas of high market competition, than in areas
of low market competition, to have had an increase in labor
efficiency.
The interpretation of MCL3 and C0MPTxMCL3 is a
function of the sum of their beta scores as follows:
BETAS
MCL3 2.2957
C0MPTXMCL3 + (-2.0941)
MCL3 = .2016
The sum of the beta scores indicates that in areas of both
high and low market competition, hospitals that had an
increase in the percent of Medi-Cal patients between 1983
and 1991 were more likely to have had an increase in labor
efficiency during that time period, than hospitals that had
a decrease or no change in their percent of Medi-Cal
patients. However, hospitals in areas of low competition
had a slightly greater increase in efficiency, than
hospitals in areas of high competition. Although these
results were in the hypothesized direction, they were not
congruent with the hypothesis and therefore did not confirm
the hypothesis.
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CD
■ D
O
Q .
C
g
Q .
■ D
CD
C/)
C/)
8
TABLE lia
3
3"
CD
CD
■ D
O
Q .
C
a
O
3
" O
O
CD
Q .
■ D
CD
C/)
C/)
ANCILLARY - REGRESSION
MEDI-CAL
COEFFICIENTS
MODEL Al MODEL A2 MODEL A3
1983-87 1987-91 1983-91
TERM BETA ODDS TERM BETA ODDS TERM BETA ODDS
RATIO RATIO RATIO
COMPT .1249 1.1330 COMPT 1.4474 4.2519 COMPT .5711 1.7703
MCL83 - .0810 .9222 MCL87 .0274 1.0277 MCL83 -.0342 .9663
MCLl 1.4950 4.4594 MCL2 .2641 1.3022 MCL3 .5514 1.7357
COMPTXMCL83 .0860 1.0898 COMPTxMCL87 - .0110 .9891 COMPTXMCL83 .0720 1.0747
COMPTxMCLl - .6871 .5030 C0MPTXMCL2 - .2597 .7712 C0MPTXMCL3 -.7658 .4650
TABLE 11b
* Significance Level < .05
NURSING - REGRESSION COEFFICIENTS
MEDI-CAL
MODEL NI MODEL N2 MODEL N3
1983-87 1987-91 1983-91
TERM BETA ODDS TERM BETA ODDS TERM BETA ODDS
RATIO RATIO RATIO
COMPT - .7449 .4748 COMPT - .2827 .7537 COMPT .7062 2.0262
MCL83 - .0103 .9898 MCL87 - .0013 .9987 MCL83 - .0602 .9416
MCLl -1.0722 .3423 MCL2 1.1777 3.2470 MCL3 2.2957 *9.9311
COMPTXMCL83 - .0067 .9933 COMPTXMCL87 .0047 1.0047 COMPTXMCL83 .1144 *1.1212
COMPTxMCLl 1.7021 5.4852 C0MPTXMCL2 - .3515 .7036 C0MPTXMCL3 -2.0941 * .1232
132
Hypothesis III;
The third hypothesis of the study was as follows:
III. Hospitals with a high percent of HMO patients are
more likely in areas of high market competition,
than in areas of low market competition, to
increase labor efficiency overtime in their:
a. Ancillary Service Area
b. Nursing Service Area
The variables that tested these hypotheses were the two
interaction terms in the Tables 12a and 12b. The
interaction term COMPTxHMO(83,87,91) measured whether
hospitals with a high percent of HMO (managed care)
patients in the base year of the time period were more
likely in areas of high market competition, than in areas
of low market competition, to have had an increase in labor
efficiency during that time period. The interaction term
COMPTxHMO(1,2,3 ) measured whether hospitals with an
increase in the percent of HMO patients during the time
period where more likely in areas of high market
competition, than in areas of low market competition, to
have had an increase in labor efficiency during the time
period. The scores in the tables were taken from the
regression models in Appendix E.
The scores in Table 12a indicate that in the Ancillary
Service Area there were no statistically significant
relationships between HMO and labor efficiency for any of
133
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the three time periods. Therefore, the results did not
confirm the hypothesis.
In Model N2 (Nursing 1987-91), the interaction term
COMPTxHMOS? was highly correlated with its main effect
HMOS7, and when in the model together, neither were found
to be significant. However, HMOS 7 was found to be
significant when COMPTxHMOS? was dropped from the model.
Therefore, Model N2 did not contain the term COMPTxHMOS?.
The beta score for HMOS? was positive which indicates
that in areas of both high and low market competition,
hospitals with a high percent of HMO patients in 19S? were
more likely to have had an increase in labor efficiency
between 19S? and 1991, than hospitals that did not have a
high percent of HMO patients in 19S7. Although these
results were in the hypothesized direction, they did not
confirm the hypothesis that a high percent of HMO patients
would have more of an effect in areas of high market
competition, than in areas of low market competition.
The scores in Table 12b indicate that in the Nursing
Service Area the term HM03 was statistically significant.
Given that the beta score was positive, this indicates that
in both high and low areas of market competition, hospitals
that had an increase in the percent of HMO patients between
1983 and 1991 were more likely to have had an increase in
134
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
labor efficiency during that time period, than hospitals
that had a decrease or no change in their percent of HMO
patients. Although these results were in the hypothesized
direction, they did not confirm the hypothesis that an
increase in HMO patients would have more of an effect in
areas of high market competition, than in areas of low
market competition.
135
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CD
■ D
O
Q .
C
g
Q .
■ D
CD
C/)
C/)
8
( O '
TABLE 12a
3
3"
CD
CD
■ D
O
Q .
C
a
o
3
" O
o
CD
Q .
■ D
CD
C/)
C/)
ANCILLARY - REGRESSION COEFFICIENTS
HMOs
MODEL A1 MODEL A2 MODEL A3
1983-87 1987-91 1983-91
TERM BETA ODDS TERM BETA ODDS TERM BETA ODDS
RATIO RATIO RATIO
COMPT .1249 1.1330 COMPT 1.4474 4.2519 COMPT .5711 1.7703
HM083 - .4027 . 6685 HM087 - .0082 .9918 HM083 - .0236 .9767
HMOl .9597 2.6109 HM02 .8351 2.3050 HM03 1.2667 3.5491
C0MPTXHM083 .4255 1.5304 C0MPTXHM087 .0220 1.0222 C0MPTXHM083 .0141 1.0142
COMPTxHMOl -1.2123 .2975 C0MPTXHM02 -1.4325 .2387 C0MPTXHM03 -1.6178 .1983
TABLE 12b
* Significance Level < .05
NURSING - REGRESSION COEFFICIENTS
HMOs
MODEL N1 MODEL N2 MODEL N3
1983-87 1987-91 1983-91
TERM BETA ODDS TERM BETA ODDS TERM BETA ODDS
RATIO RATIO RATIO
COMPT -.7449 .4748 COMPT -.2827 .7537 COMPT .7062 2.0262
HM083 -.0812 .9220 HM087 .0443 *1.0453 HM083 .0951 1.0998
HMOl .3271 1.3870 HM02 .1660 1.1806 HM03 2.5335 *12.5975
C0MPTXHM083 .0813 1.0847 C0MPTXHM087 N/A N/A C0MPTXHM083 -.0890 .9149
COMPTxHMOl -.1705 .8432 C0MPTXHM02 .4290 1.5357 C0MPTXHM03 -2.0038 .1348
136
Control Variables
The scores in Tables 13a and 13b indicate that the
regression models also identified statistically significant
relationships between membership, ownership and changes in
labor efficiency scores.
The base year efficiency scores in the regression
models were represented by A83, A87, A91, N83, N87 and N91.
With the exception of A87, all of the variables were
significant and negative. The scores in Table 13a and 13b
indicate that in areas of both high and low market
competition, hospitals with a high efficiency score in the
base year of the time period were more likely to have had
a decrease in their efficiency score during the time
period, than hospitals that had low efficiency scores in
the base year. Therefore, hospitals that had high DEA
scores were more likely to have had a decrease in their DEA
scores during the time period.
Ownership was found to be statistically significant
only in the Nursing Service Area between 1983 and 1987.
The beta score of the term OWN in Table 13b is positive.
This indicates that in areas of both high and low market
competition, for-profit hospitals were more likely, than
non-profit hospitals, to have had an increase in labor
efficiency in their Nursing Service Area from 1983 to 1987.
137
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Membership in a multi-hospital system was found to be
related to labor efficiency in the Ancillary Service Area
but not in the Nursing Service Area. The scores in Table
13a indicate that the terms MEM87, MEM2 and MEM3 were
statistically significant and the beta scores were
positive. MEM87 indicates that in areas of both high and
low market competition, hospitals that were a member of a
multi-hospital system in 1987 were more likely to have had
an increase in labor efficiency in their Ancillary Service
Area between 1987 and 1991, than hospitals that were not a
member of a multi-hospital system in 1987. Similarly, MEM2
indicates that in areas of both high and low market
competition, hospitals that became a member of a multi
hospital system between 1987 and 1991 were more likely to
have had an increase in labor efficiency in their Ancillary
Service Area, than hospitals that did not change membership
between 1987 and 1991. The term MEM3 in Table 13a was
statistically significant and the beta score was positive.
This indicates that in areas of both high and low market
competition, hospitals that became a member of a multi
hospital system between 1983 and 1991 were more likely to
have had an increase in labor efficiency in their Ancillary
Service Area, than hospitals that did not change their
membership between 1983 and 1991.
138
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CD
■ D
O
Q .
C
g
Q .
■ D
CD
TABLE 13a
c/)
c/)
8
CD
3 .
3"
CD
ANCILLARY - REGRESSION COEFFICIENTS
CONTROL VARIABLES
MODEL Al MODEL A2 MODEL A3
1983-87 1987-91 1983-91
TERM BETA ODDS TERM BETA ODDS TERM BETA ODDS
RATIO RATIO RATIO
A83 -4.2232 * .0147 A87 -2.0204 .1326 A83 -4.1605 * .0156
OWN .7148 2.0438 OWN - .5038 .6042 OWN - .8961 .4081
MEM83 .1771 1.1938 MEM87 1.0573 *2.8785 MEM83 .2205 1.2467
MEMl - .0089 .9911 MEM2 1.5445 *4.6855 MEM3 1.5722 *4.8171
CD
" O
O
Q .
C
a
O
3
" O
O
CD
Q .
■ D
CD
C /)
C /)
TABLE 13b
NURSING - REGRESSION COEFFICIENTS
CONTROL VARIABLES
MODEL N1 MODEL N2 MODEL N3
1983-87 1987-91 1983-91
TERM BETA ODDS TERM BETA ODDS TERM BETA ODDS
RATIO RATIO RATIO
N83 -3.2284 * .0396 N87 -3.1752 * .0418 N83 -5.0696 * .0063
OWN 1.3900 *4.0149 OWN - .0733 .9293 OWN .0532 1.0547
MEM83 - .2949 .7446 MEM87 .4209 1.5233 MEM83 - .1315 .8768
MEMl - .2993 .7414 MEM2 - .8014 .4487 MEM3 - .3036 .7381
* Sign]Lficance Level < .05
139
Summary
The logistic regression models did not confirm any of
the hypotheses that hospitals in areas of high market
competition were more likely, than hospitals in areas of
low market competition, to have had an increase in labor
efficiency in either their Ancillary Service Area or
Nursing Service Area in any of the time periods. Moreover,
with the exception of Model A2 (Ancillary 1987-1991), all
of the models found that in areas of both high and low
market competition, hospitals with a high efficiency score
in the base year of the time period were more likely to
have had a decrease in their efficiency score during each
time period.
In areas of both high and low market competition,
hospitals that had an increase in the percent of Medi-Cal
and private managed care patients (price competition)
between 1983 and 1991 were more likely to have had an
increase in labor efficiency in their Nursing Service Area
during that time period, than hospitals that had a decrease
or no change in the percent of managed care patients.
However, in areas of high market competition, hospitals
that had an increase in the percent of Medicare patients
(rate regulation) between 1987 and 1991 were more likely to
have had a decrease in labor efficiency in their Nursing
140
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Service Area during that time period, than hospitals that
had a decrease or no change in the percent of Medicare
patients.
In both areas of market competition, hospitals that
joined a multi-hospital system between 1983 and 1991 were
more likely to have had an increase in labor efficiency in
their Ancillary Service Area, than hospitals that did not
change membership during that time period. Similarly, in
both areas of market competition, for-profits were more
likely than non-profits to have had an increase in labor
efficiency in their Nursing Service Areas between 1983 and
1987.
141
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
VIII. D I S C U S S I O N
The results of this study did not confirm the
hypotheses that hospitals in areas of high market
competition were more likely, than hospitals in areas of
low market competition, to have had an increase in labor
efficiency during any of the time periods analyzed. The
results of the study may raise some question regarding the
reliability of the DEA efficiency scores. Table 14 below
is based on the raw data which shows the trend in total
average FTEs per total average discharge and total average
outpatient visit in both areas of high and low competition.
In the Ancillary Service Area, the raw data indicates
that, overall, the average ETE per average discharge
increased slightly in both areas of high and low
competition. Therefore, the overall increase in average
FTE per average discharge and average outpatient visit,
combined, is congruent with the decline in the average DEA
scores for the Ancillary Service Area found in Appendix C,
Figure 1. Between 1983 and 1991, the average FTE per
average discharge and outpatient visit, combined, declined
approximately 3 percent in areas of high market competition
and increased 4 percent in areas of low market competition.
Given that these changes were small, the raw data would be
142
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
congruent with the study which found no statistically
significant changes in labor efficiency in the Ancillary
Service Area for each time period.
In the Nursing Service Area, the raw data indicates
that, overall, the average FTE per average discharge
changed very little during each time period and in both
areas of high and low competition. This would be
consistent with the little change in average DEA efficiency
scores found in Appendix C, Figure 2. Moreover, given that
little change was found in the average FTE per average
discharge in both areas of high and low competition, the
raw data is congruent with the study which found that
hospitals in areas of high market competition were not more
likely, than hospitals in areas of low market competition,
to have had an increase in labor efficiency.
TABLE 14
Total Average FTEs per
Total Average Discharge and Total Average Visit
(N = 149)
ANCILLARY
HIGH COMPETITION LOW COMPETITION
1983 1987 1991 1983 1987 1991
FTE/UDIS
FTE/ADIS
FTE/VIS
FTE/UTOT
FTE/ATOT
.021
.022
.0100
. 0068
.0068
.023
.023
.0089
.0064
.0065
.024
.025
.0090
.0065
.0066
.019
.020
.0074
.0053
.0054
.023
.023
.0077
.0057
.0058
.025
.026
.0074
.0057
.0057
NURSING
FTE/UDIS
FTE/ADIS
.022
.022
.020
.021
.020
.021
.019
.020
.020
.021
.020
.021
UDIS = Unadjusted Discharge/ ADIS = Case-mix Adjusted
Discharge
143
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The FTE data for this study was provided by the Office
of Statewide Health Planning and Development. The state
requires all hospitals to report data by hospital cost
center according to definitions applicable to all
hospitals. Assuming that all hospitals comply with these
uniform definitions, the data for the study should be
reasonably uniform.
The results may also raise some questions regarding
the reliability of the data used for defining areas of high
and low market competition. The Institute for Health
Policy Studies provided the number of competitors within a
15 mile radius for hospitals in California. The data
source used by the Institute was data published by the
American Hospital Association. Assuming that the data
reported to the American Hospital Association was accurate,
the definitions for areas of high and low competition in
this study should be reliable.
Given that the study did not confirm the hypotheses,
the results may raise questions regarding the reliability
of the statistical models. Each logistic regression model
was tested for collinearity and if high correlation between
variables was found, the models were tested for changes in
statistical significance by running each model with and
144
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
without the highly correlated pairs. The results of the
models reported in this study were not effected by
collinearity present in the models. Therefore, the
statistical models should be reasonably reliable.
Assuming that the data, the DEA scores and the
statistical models were reliable, the results of the study
may be reflecting changes in industry behavior that has not
resulted in increased labor efficiency. The statistical
models for the Ancillary Service Areas did not find any
statistically significant relationships between the
variable for increased labor efficiency and the variables
for payer. Zwanziger et al. (1994a) found that the decline
in the rate of hospital costs between 1980 and 1990 was
due, in part, to shifting inpatient services to outpatient
settings. An increase in outpatient activity would require
an increase in outpatient labor. However, as inpatient
patients decline, the average ancillary FTE per inpatient
discharge may increase and, as outpatient visits increase,
the average FTE per outpatient visit may decline.
Consequently, the overall average FTE per discharge and
visit, combined, may remain relatively unchanged as is
reflected in the raw data and is indicated by the findings
of this study. Therefore, the offset of decreasing
inpatient activity by increasing outpatient activity may
145
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
explain why this study did not find statistically
significant overall changes in labor efficiency in the
Ancillary Service Areas in any of the time periods.
The study found that hospitals that joined a multi
hospital system between 1983 and 1991 were more likely to
have had an increase in labor efficiency in their Ancillary
Service Area. Under multi-hospital membership, hospitals
may have been able to reduce their ancillary FTEs if
certain diagnostic services were shifted to providers that
had been consolidated by the multi-hospital corporation.
Redirecting patients to consolidated providers would allow
member hospitals to reduce their own in-house ancillary
personnel. Thus, the findings of this study regarding
multi-hospital membership may be congruent with the
consolidation practices of multi-hospital systems in the
industry.
The study found that hospitals that had an increase in
the percent of Medicare patients between 1987 and 1991 were
more likely in areas of high market competition, than in
areas of low market competition, to have had a decrease in
labor efficiency in their Nursing Service Area. From a
theoretical point of view, these results may suggest that
reimbursement rates set under rate regulation may be higher
than would be found under normal market conditions.
146
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Therefore, hospitals in areas of high competition might
have less incentive to increase efficiency. However, given
the shift of inpatient care to outpatient settings, a
higher proportion of Medicare patients admitted to the
hospital may have high levels of illness and therefore
require higher use of resources. Although this study case-
mixed adjusted inpatient patients, the proportion of sicker
patients within a particular DRG may have increased over
the time period of the study. For example, between 1983
and 1987, 10 out of 20 patients admitted under DRG 50 may
have had high severity of illness. With the increase in
outpatient care during the 1980s, 15 out of 20 patients
admitted under DRG 50 between 1987 and 1991 may have had
high levels of illness. Consequently, a higher percent of
Medicare patients admitted for inpatient care would require
a higher level of care from nursing service personnel.
Therefore, a higher proportion of Medicare patients with
high severity levels of illness may explain why hospitals
with an increase in the percent of Medicare patients
between 1987 and 1991 were more likely to have had a
decrease in labor efficiency in their Nursing Service
Areas.
The study found that hospitals with a high percent of
managed care patients or an increase in the percent of
147
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
managed care patients were more likely to have had an
increase in labor efficiency in their Nursing Service Area
regardless of the degree of market competition. From a
theoretical point of view, these results may suggest that
markets that are dominated by payers who can control the
beneficiary's choice of hospital and physician are more
effective at inducing efficient hospital behavior than the
presence of other competitors in the market. However,
given that hospitalization of managed care patients
requires the permission of the managed care plan prior to
admission, managed care plans may be able to minimize
unnecessary hospital admissions. Consequently, hospitals
with a high percent of managed care patients may require
fewer nursing personnel to support fewer inpatient
admissions. Therefore, this study's findings of increased
labor efficiency in nursing departments of hospitals with a
high percent of managed care patients may be congruent with
the monitoring practices of managed care plans.
The studies by Zwanziger (1988a, 1994a, 1994b) and
Phibbs (1989, 1993) found that the rate of hospital costs
in California declined in the decade following the
implementation of Medicare BPS and selective contracting in
1983. The findings of this study may suggest that the
reduction in costs found in the studies by Zwanziger and
148
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Phibbs were not due to an overall reduction in FTEs in
Ancillary and Nursing Service Areas. More likely, the
reduction in costs may have been due to minimizing hospital
admissions and shifting care from expensive inpatient
settings to less expensive outpatient settings. The
implication of these studies combined is that price
constraints imposed on the market under rate regulation and
price competition may result in cost minimization by
appropriately allocating services according to the level of
care required.
149
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
IX. IMPLICATIONS FOR PUBLIC POLICY
Rate Regulation
Rate regulation is one policy approach to induce
providers to be more sensitive to the cost of producing
services. The DRG based Prospective Payment System for
hospitals was one means to impose price constraints on
hospitals. Given that profits would be more difficult to
maintain under price constraints, hospitals in areas of
high market competition would be expected to find ways to
reduce the cost and volume of inputs in order to maintain
profit. However, the theory of rate regulation assumes
that the prices set under rate regulation are prices that
would be found in naturally occurring competitive markets.
If prices are set to high, then hospitals may have less
incentive to produce services as efficiently as they would
under lower prices that might occur under normal market
conditions. Given that DRG rates are not based on prices
that occur naturally in competitive markets, the DRG rates
may be higher than would be under normal competitive
conditions. If rates are set to high then hospitals may
have less incentive to change their utility-maximizing
behavior. This study did not find that hospitals with a
high percent of Medicare patients were more likely to
150
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
increase labor efficiency in their Ancillary or Nursing
Service Areas. Conversely, the study found that in areas
of high market competition, hospitals that had an increase
in the percent of Medicare patients between 1987 and 1991
were more likely to have had a decrease in labor efficiency
in their Nursing Service Area during that time period.
Therefore, these findings may suggest that DRG rates are
not low enough to induce hospitals to increase their
efficient use of labor. Moreover, this may suggest that
rate regulation is marginally effective at changing
utility-maximizing behavior of hospitals.
Price Con^etition
The capitated premium that managed care plans receive
on behalf of beneficiaries creates the incentive to
selectively contract with hospitals and physicians that can
provide quality care with the least expense. Private and
public payers in California have been selectively
contracting with hospitals since 1982. By 1986, 60 percent
of California beneficiaries were enrolled in managed care
plans and by 1990, 80 percent were enrolled (Zwanziger,
1994b). Consequently, California became a payer-dominated
market during the 1980s.
151
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Under a payer-dominated market, hospitals must
minimize the cost and volume of inputs in order to offer
competitively low prices to managed care payers. The
theory of price competition would predict that hospitals in
areas with a large number of competitors would be more
likely to provide medical services with greater economic
and technical efficiency than hospitals in areas with fewer
competitors. However, the results of this study found that
in both high and low areas of market competition, hospitals
that had an increase in the percent of patients from
publicly or privately financed managed care plans between
1983 and 1991 were more likely to have had an increase in
labor efficiency in their Nursing Service Area, than
hospitals that did not have an increase in the percent of
managed care patients. Given that hospitals with an
increase in managed care patients increased labor
efficiency even in areas with relatively few competitors,
may suggest that prices that arise out of payer-dominated
markets may be effective at changing hospital utility-
maximizing behavior.
Quality Care
The results of this study may have implications for
quality of care. Since the implementation of Medicare PPS
152
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and the rise of managed care, policy makers have been
concerned that, given market price constraints, hospitals
would have the incentive to withhold necessary services in
order to maximize profits. The results of this study
suggest that under rate regulation or price competition,
hospitals have not reduced their labor FTEs in their
Ancillary Service Areas, which would suggest that necessary
services are not being withheld. However, the study did
find that hospitals that joined a multi-hospital system
between 1983 and 1991 were more likely to have reduced
their Ancillary labor FTEs, than hospitals that did not
change membership. Although a reduction of personnel may
suggest that services are being withheld, it is also
probable that hospitals reduced their Ancillary labor FTEs
if certain diagnostic services were shifted to other
hospitals or providers that had been consolidated under
multi-hospital membership. Such consolidation would suggest
allocative efficiency in the delivery of medical services.
In the Nursing Service Area, the results indicate that
hospitals have reduced their labor FTEs in response to
price competition. In the Nursing Service Areas, hospitals
are required to staff the patient care units of the Nursing
Service Areas with the nursing mix appropriate to the level
of illness of patients on the floors. If hospitals reduce
153
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
their nursing-mix below that level which is appropriate,
then patients may not receive necessary care. Therefore,
regulatory agencies, such as the Joint Commission of
Accreditation of Health Organizations (JCAHO), may need to
closely monitor reductions in labor FTEs in Nursing Service
Areas of hospitals that have a high percent of managed care
contracts.
154
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
X. STUDY LIMITATIONS
Data Envelopment Analysis (DEA)
When using DEA to define a hospital as efficient or
inefficient, it should be kept in mind that the score is
only a reflection of efficiency relative to other hospitals
in the analysis and not relative to a standard of
efficiency. Consequently, adding or deleting hospitals to
the data set may change whether a hospital is found
efficient or inefficient. Similarly, changing the number of
input and/or output categories of the model can change
whether a hospital is determined to be efficient or not
efficient. The number of input and output categories can
change where the hospital sits relative to the efficiency
frontier. Therefore, the accuracy of DEA in classifying
hospitals as efficient or inefficient is significantly
dependent on the similarity of the hospitals selected for
comparison as well as the selection of input and output
categories assumed to appropriately reflect the process of
production.
Data
The data used to determine the DEA scores was provided
by the California Office of Statewide Health Planning and
155
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Development. Between 1980 and 1992 hospitals were required
to separately report costs but not hours for Registry
nurses. Therefore, this study was not able to include
hours for Registry nurses in the calculation of the DEA
efficiency scores. Registry nurses are primarily utilized
in the Nursing Service Area. If some hospitals utilize a
high proportion of Registry hours in their Nursing Service
Area, these hospitals might have appeared to use fewer
FTEs. However, if all hospitals utilized Registry hours to
similar proportions then the relativity of the DEA scores
may not have been effected.
Logistic Regression
This study was unable to use multiple regression since
the data violated the assumptions of normality and
linearity. In turning to logistic regression, a certain
amount of information may have been lost regarding the
dependent variable. In multiple regression, the dependent
variable is a continuous variable which takes into account
a full range of values. Logistic regression requires
defining the dependent variable as efficient or not
efficient thereby losing the varying degrees of efficiency
less than 1 but greater than 0. If multiple regression
156
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
could have been used, this information would not have been
lost and the model may have produced different results.
Case Mix
Although the Health Care Financing Administration has
adopted the use of charge adjusted case-mix indices as a
measure of hospital patient illness acuity, some analysts
argue that there is still too much variance within each DRG
category regarding the utilization of resources to be
reliable. Those analysts and organizations that continue
to calculate case-mix have turned to a method known as
staging. In staging, each patient within a DRG is weighted
by a score of 1 to 4. Each score represents a different
level of illness. Therefore, stage-mix is believed to be a
more refined reflection of the degree of illness of
patients and therefore more reflective of hospital resource
requirements. Stage-mix requires a special software
program that was not available for this study. Had stage-
mix been available, it is conceivable that this analysis
may have produced different results.
157
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
XI. F U R T H E R R E S E A R C H
The Health Security Act under the Clinton
Administration is based on principles of managed
competition. A central principle of managed competition
is that the health care market consists of a number of
managed care plans that compete on a price and quality
basis to attract health plan members. To stay competitive,
managed care plans must minimize cost to minimize price and
still make a profit. One way to minimize cost is to
minimize the amount of reimbursement the plan offers
providers. Further research which used DEA to analyze
changes in labor inputs for individual nursing departments,
might identify if reimbursement policies effect departments
differently. Moreover, analysis at the departmental level
would be useful to hospital administrators.
The DEA inputs for this study were labor FTE
categories for all ancillary or nursing departments
combined. Further research which used the number of
ancillary tests or procedures as DEA inputs would provide
information on changes in the actual number of tests or
procedures provided per discharge. Changes might suggest an
influence of reimbursement policies on physician ordering
practices.
158
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The results from DEA are only as good as the
assumptions made regarding the similarity of the
organizations being compared. This analysis attempted to
meet this criteria by case-mix adjusting the discharges.
However, greater similarity may have been achieved by
categorizing hospitals according to similar case-mix, size,
ownership and membership in a multi-hospital system. This
approach would require determining statistically
significant case-mix categories and obtaining a large
enough sample of hospitals for each category. The sample
size of 149 hospitals in this study would not have provided
enough hospitals per case-mix category.
This study examined technical efficiency of labor
inputs in only Ancillary and Nursing Service Areas. A
study which examined administrative departments would
provide insight into the effects of reimbursement policies
on administrative labor efficiency. However, such changes
might be correlated with membership in a multi-hospital
system.
This study examined changes in labor efficiency
between 1983 and 1991. Applying later data to the design
of this study may identify further long term effects of
rate regulation or price competition on labor efficiency.
159
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BIBLIOGRAPHY
Ali, A., Lerme, C. and Seiford, L. (1995), "Components of
Efficiency Evaluation in Data Envelopment Analysis,"
European Journal of Operational Research, Vol. 80, pp. 462-
473.
All, A. and Seiford, L. (1993a), "Computational Accuracy
and Infinitesimals in Data Envelopment Analysis," INFOR,
Vol. 31, pp. 290-297.
All, A., Lerme, C. and Nakosteen, R. (1993b), "The
Assessment of Intergovernmental Revenue Transfers," Socio-
Economic Planning Sciences, Vol. 27, pp. 109-118.
All, A., and Lerme, C. (1994), "Inefficiency as Comparative
Disadvantage in DEA," Working Paper, The University of
Massachusetts at Amherst, pp. 1-17.
Ali, A. and Seiford, L. (1993c), "The Mathematical
Programming Approach to Efficiency Analysis," in The
Measurement of Productive Efficiency by Fried, H., Lovell,
C. and Schmidt, S., Oxford University Press, New York, New
York.
Anderson, G., Heyssel, R. and Dickler, R. (1993),
"Competition Versus Regulation: Its Effect on Hospitals,"
Health Affairs, Vol. 12, pp. 70-80.
Arrow, K. (1963), "Uncertainty and the Welfare Economics of
Medical Care," The American Economic Review, Vol. 58, pp.
941-973.
Ashby, J. and Altman, S. (1992), "The Trend in Hospital
Output and Labor Productivity, 1980-1989," Inquiry, Vol.
29, pp. 80-91.
Banker, R. and Morey, R. (1985), "Efficiency Analysis for
Exogenously Fixed Inputs and Outputs," Operations Research,
Vol. 34, pp. 513-521.
Banker, R. and Morey, R. (1986), "The Use of Categorical
Variables in Data Envelopment Analysis," Management
Science, Vol. 32, pp. 1613-1627.
160
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Banker, R., Charnes, A. and Cooper, W. (1984), "Some Models
for Estimating Technical and Scale Inefficiencies in Data
Envelopment Analysis," Management Science, Vol. 30, pp.
1078-1092.
Banker, R., Conrad, R. and Strauss, R. (1986), "A
Comparative Application of Data Envelopment Analysis and
Translog Methods: An Illustrative Study of Hospital
Production," Management Science, Vol. 32, pp. 30-44.
Berki, S. (1972), Hospital Economics, Lexington Books,
Lexington, Massachusetts.
Bernstein, R. (1991), Hospital Peer Grouping, State of
California, Department of Health Services, Rate Development
Branch.
Bordon, J. (1986), An Assessment of the Impact of
Diagnostic Related Group(DRG)-Based Reimbursement on the
Technical Efficiency of New Jersey Hospitals, Dissertation,
Drexel University.
Broyles, R.W. (1990), "Efficiency, Costs and Quality: The
New Jersey Experience Revisited," Inquiry, Vol. 27, pp. 86-
96.
Charnes, A., Cooper, W. and Rhodes, E. (1981), "Evaluating
Program and Managerial Efficiency: An Application of Data
Envelopment Analysis to Program Follow Through, " Management
Science, Vol. 27, pp. 668-697.
Chesney, J. (1990), "Utilization Trends Before and After
PPS," Jnguiry, Vol. 27, pp. 376-381.
Cohen, W.J., (1985), "Reflections on the Enactment of
Medicare and Medicaid," Health Care Financing Review,
Annual Supplement, pp. 3-11.
Conrad, R. and Strauss, R. (1983), "A Multiple-Output
Multiple-Input Model of the Hospital Industry in North
Carolina," Applied Economics, Vol. 15, pp. 341-352.
Cotterill, P., Bobula, J. & Connerton, R. (1986), "A
Comparison of Alternative DRG Relative Weights for the
Medicare Prospective Payment System," Health Care Financing
Review, Vol. 7, pp. 37-51.
161
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Coulam, R. and Gaumer, G. (1991), "Medicare's Prospective
Payment System: A Critical Appraisal," Health Care
Financing Review, Annual Supplement, pp. 45-77.
Cromwell, J. and Puskin, D. (1989), "Hospital Productivity
and Intensity Trends: 1980-87," Inquiry, Vol. 26, pp. 366-
380.
Davis, K. (1972), Economic Theories of Behavior in
Nonprofit, Private Hospitals, The Brookings Institute,
Washington D.C.
Dittman, R., Capettini, R. and Morey, R. (1991), "Measuring
Efficiency in Acute Care Hospitals: An Application of Data
Envelopment Analysis," Journal of Health and Human
Resources Administration, Vol. 14, pp. 89-108.
Donaldson, C. and Magnussen, J., "DRGS: The Road to
Hospital Efficiency," Health Policy, Vol. 21, pp. 47-64.
Dorfman, R. (1978) Prices and Markets, Prentice Hall, Inc,
Englewood Cliffs, New Jersey, 3rd Edition.
Durenberger, D. (1993), "Government and the Competitive
Marketplace," Health Affairs, Vol. 12, pp. 81-84.
Ehreth, J. (1994), "The Development and Evaluation of
Hospital Performance Measures for Policy Analysis," Medical
Care, Vol. 32, pp. 568-587.
Enthoven, A. (1994), "Why Not the Clinton Health Plan?,"
Inquiry, Vol. 31, pp. 129-135.
Enthoven, A. and Singer, J. (1994), "A Single-Payer System
in Jackson Hole Clothing," Health Affairs, Vol. 13, pp. 81-
95.
Enthoven, A.C. (1988), Theory and Practice of Managed
Competition in Health Care Finance, New York: North
Holland.
Farley, D. (1985), Competition Among Hospitals: Market
Structure and Its Relation to Utilization, Costs and
Financial Position, Research Note 7, Hospital Studies
Program, National Center for Health Services Research and
Health Care Technology Assessment.
162
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Farrell/ M.J. (1957)/ "The Measurement of Productive
Efficiency/" Journal of the Royal Statistical Society,
Series A, p. 253-290.
Farrell/ M.J. and Fieldhouse/ M. (1962)/ "Estimating
Efficient Production Functions Under Increasing Returns to
Scale/" Journal of Royal Statistical Society, Series A,
pp. 252-267.
FeldsteiU/ M. (1967)/ Economic Analysis for Health Service
Efficiency, North-Holland Publishing Company/ Amsterdam.
Feldstein/ P.J. (1993)/ Health Care Economics, Delmar
Publishers/ Inc./ Albany/ New York.
Finkler/ M. and Wirtschafter/ D. (1993)/ "Cost-
Effectiveness and Data Envelopment Analysis/" Health Care
Management Review, Vol. 18/ pp. 81-88.
Friedman/ B. and Shortell/ S. (1988)/ "The Financial
Performance of Selected Investor-Owned and Not-For-Profit
System Hospitals Before and After Medicare Prospective
Payment/" Health Services Research, Vol. 23, pp. 237-267.
Gillies, R. (1993), "Conceptualizing and Measuring
Integration; Findings from the Health Systems Integration
Study," Hospital & Health Services Administration, Vol. 38,
pp. 467-489.
Greenberg, W. (1991), Competition,. Regulation and Rationing
of Health Care, Health Administration Press, Ann Arbor,
Michigan.
Grosskoph, S. and Valdmanis, V. (1993), "Evaluating
Hospital Performance with Case-mix Adjusted Outputs,"
Medical Care, Vol. 31, pp. 525-532.
Horn, S. and Schumacher, D. (1982), "Comparing
Classification Methods: Measurement of Variation in
Charges, Length of Stay and Mortality," Medical Care, Vol.
20/ pp. 489-500.
Huang, Y. and McLaughlin, C. (1989), "Relative Efficiency
in Rural Primary Health Care: An Application of Data
Envelopment Analysis," Health Services Research, Vol. 24,
pp. 143-158.
163
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Huskamp, H. and Newhouse, J. (1994), "Is Health Spending
Slowing Down," Health Affairs, Vol. 13, pp. 32-38.
Johnson, A., Dowdm, B., Morris, N. and Lurie, N. (1989),
"Differences in Inpatient Resource Use by Type of Health
Plan," Inquiry, Vol. 26, pp. 388-398.
Johnson, A.N. and Aquilina, D. (1986), "The Impact of
Health Maintenance Organizations and Competition on
Hospitals in Minneapolis/St. Paul," Journal of Health
Politicsr Policy and Law, Vol. 10, pp. 659-674.
Joskow, P.L. (1980), "The Effects of Competition and
Regulation on Hospital Bed supply and the Reservation
Quality of the Hospital, " The Bell Journal of Economics,
Vol. 11, pp. 421-447.
Kahn, K., Rogers, W., Draper, D., Kosecoff, J., Keeler, E.,
Rubenstein, L., Reinisch, E., Sherwood, M., and Brook, R.
(1990), "Measuring Quality of Care with Explicit Process
Criteria Before and After Implementation of the DRG-Based
Prospective Payment System," JAMA, Vol. 264, pp. 1969-1973.
Kahn, K., Rogers, W., Draper, D., Kosecoff, J., Keeler, E.,
Rubenstein, L., Bentow, S., Reinisch, E., Sherwood, M., and
Brook, R. (1990), "Comparing Outcomes of Care Before and
After Implementation of the DRG-Based Prospective Payment
System," JAMA, Vol. 264, pp. 1984-1988.
Kahn, K., Rubenstein, L., Draper, D., Kosecoff, J., Rogers,
W., Keeler, E. and Brook, R. (1990), "The Effects of the
DRG-Based Prospective Payment System on Quality of Care for
Hospitalized Medicare Patients," JAMA, Vol. 264, pp. 1953-
1955.
Keeler, E., Kahn, K., Draper, D., Kosecoff, J., Rubenstein,
L., Reinisch, E., Sherwood, M., and Brook, R. (1990),
"Changes in Sickness at Admission Following the
Introduction of the Prospective Payment System," JAMA, Vol.
264, pp. 1962-1968.
Kidder, D. and Sullivan, D. (1982), "Hospital Payroll
Costs, Productivity and Employment Under Prospective
Reimbursement," Health Care Financing Review, Vol. 4, pp.
89-100.
164
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Koch, A.L. (1988), "Financing Health Services," in Williams
and Torrens, Introduction to Health Services, 3rd Edition,
New York: John Wiley and Sons.
Kosecoff, J., Kahn, K., Rogers, W., Draper, D., Rubenstein,
L., Reinisch, E., Sherwood, M., Roth, C., Chew, C. and
Brook, R. (1990), "Prospective Payment System and
Impairment at Discharge: The 'Quicker-and-Sicker' Story
Revisited," JAMA, Vol. 264, pp. 1980-1983.
Larson, J. (1983), "Factors in the Success of the Investor-
Owned Hospitals: Implications for the Not-For-Profits."
Hospitals and Health Services Administration, Vol. 28, pp.
43-44.
Levit, K., Cowan, C., Lazenby, H., McDonnell, P., Sensenig,
A., Stiller, J. and Won, D. (1994), "National Health
Spending Trends, 1960-1993," Health Affairs, Vol. 13, pp.
14-31.
Lewin, L., Derzon, R. and Margulies, R. (1981), "Investor-
Owned and Non-Profits Differ in Economic Performance,"
Hospitals, Vol. 55, pp. 52-58.
Long, M., Chesney, J. and Fleming, S. (1990), "A
Reassessment of Hospital Product and Productivity Changes
Over Time," Health Care Financing Review, Vol. 11, pp. 69-
77.
Long, M., Chesney, J., Ament, R., DesHarnais, S., Fleming,
S., Kobrinski, E. and Marshall, B. (1987), "The Effect of
PPS on Hospital Product and Productivity, " Medical Care,
Vol. 25, pp. 528-538.
Luft, H. (1978), "How do Health Maintenance Organizations
Achieve their Savings," New England Journal of Medicine,
Vol. 298, pp. 1336-1343.
Luft, H., Robinson, J., Garnick, D., Maerki, S. and McPhee,
S. (1986a), "The Role of Specialized Clinical Services in
Competition Among Hospitals," Inquiry, Vol. 23, pp. 83-94.
Luft, H. (1986b), "The Competitive Effects of Health
Maintenance Organizations: Another Look at the Evidence
from Hawaii, Rochester, Minneapolis/St. Paul," Journal of
Health Politics, Policy and Law, Vol. 10, pp. 625-658.
165
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Lynch/ J. and McCue, M. (1990) , "The Effects of For-Profit
Multi-Hospital System Ownership on Hospital Financial and
Operating Performance/" Health Services Management
Research, Vol. 3/ pp. 182-192.
Lynch/ J. and Ozcan/ Y. (1994)/ "Hospital Closure; An
Efficiency Analysis/" Hospital & Health Services
Administration, Vol. 39/ pp. 205-220.
Manning/ W. (1984)/ "A Controlled Trial of the Effect of a
Prepaid Group Practice on Use of Services/" New England
Journal of Medicine, Vol. 310/ pp. 1505-1510.
Manning/ W./ Newhouse/ P./ Duau/ N./ Keeler/ E./ LeibowitZ/
A. & Marquis/ M. (1987)/ "Health Insurance and the Demand
for Medical Care: Evidence from a Random Experiment/"
American Economic Review, Vol. 11, pp. 251-277.
Manton/ K. (1993)/ "Use of Medicare Services before and
after Introduction of the Prospective Payment System/"
Health Services Research, Vol. 28/ pp. 269-292.
Melnick/ G., Zwanziger/ J. & Patterson/ R. (1992)/ "The
Effects of MarJcet Structure and Bargaining Position on
Hospital PriceS/" Journal of Health Care Economics, Vol.
11/ pp. 217-233.
McGuire/ A. (1987)/ "The Measurement of Hospital
Efficiency/" Social Science and Medicine, Vol. 24/ pp. 719-
724.
Mendelson/ D. and SchwartZ/ W. (1993)/ "The Effects of
Aging and Population Growth on Health Care CostS/" Health
Affairs, Vol. 12/ pp. 119-125.
Newhouse/ J. and Huskamp/ H. (1994) Is Health Spending
Slowing Down?/" Health Affairs, Vol. 13/ pp. 32-38.
Nippert/ R.P. (1992)/ "Competition in Health Care Delivery
of the U.S.: Panacea or Poison? A Critical View," Health
Policy, Vol. 20/ pp. 301-308.
NorusiS/ M.J./ (1994) SPSS Guide to Data Analysis, Prentice
Hall/ Englewood CliffS/ New Jersey.
166
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Nunamaker, T. (1983a), "Measuring Routine Nursing Services:
A Comparison of Cost per Patient Day and Data Envelopment
Analysis," Health Services Research, Vol. 18, pp. 183-205.
Nunamaker, T. (1983b), Efficiency Measurement and Medicare
Reimbursement in Nonprofit Hospitals: An Investigation of
the Usefulness of Data Envelopment Analysis, Doctoral
Dissertation, University of Wisconsin - Madison.
Nyman, J., Bricker, D. (1989), "Profit Incentives and
Technical Efficiency in the Production of nursing home
Care," The Review of Economics and Statistics, Vol. 71, pp.
586-594.
Nyman, J., Bricker, D. and Link, D. (1990), "Technical
Efficiency in nursing homes," Medical Care, Vol. 28, pp.
541-551.
Ozcan, Y. (1993), "Sensitivity Analysis of Hospital
Efficiency Under Alternative Output/Input and Peer Groups:
A Review," Knowledge and Policy, The International Journal
of Knowledge Transfer and Utilization, Vol. 5, pp. 1-29.
Ozcan, Y. and Luke, R. (1993), "A National Study of the
Efficiency of Hospitals in Urban Markets," Health Services
Research, Vol. 27, pp. 719-739.
Pattison, R. and Katz, H. (1983, "Investor-Owned and Not-
For-Profit Hospitals: A Comparison Based on California
Data," New England Journal of Medicine, Vol. 309, pp. 347-
353.
Phibbs, C. and Robinson, J. (1989), "An Evaluation of
Medicaid Selective Contracting in California," Journal of
Health Economics, Vol. 8, pp. 437-455.
Phibbs, C. and Robinson, J. (1993), "A Variable-Radius
Measure of Local Hospital Market Structure," Health
Services Research, Vol. 28, pp. 313-324.
Phibbs, C., Zwanziger, J., Weinberg, S., Garnick, D., Luft,
H., Melnick, G., and Robinson, J. (1995), Alternative
Measures of the Competitiveness of Hospital Markets,
Working Paper, VA Medical Center, Department of Health
Research and Policy, Stanford University, Draft Paper,
September.
167
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Porter, M. (1980), Competitive Strategy, New York, The Free
Press.
Renn, S.C. (1987), "The Structure and Financing of the
Health Care Delivery System of the 1980's," in Schramm
Health Care and Its Costs, New York: W.W. Norton and
Company.
Robinson, J.C. and Luft, H. (1985), "The Impact of Market
Structure on Patient Volume, Average Length of Stay, and
the Cost of Care," Journal of Health Economics, Vol. 4, pp.
333-356.
Rogers, W., Kahn, K., Draper, D., Kosecoff, J., Keeler, E.,
Rubenstein, L., and Brook, R. (1990), "Quality of Care
Before and After Implementation of the DRG-Based
Prospective Payment System: A Summary of Effects," JAMA,
Vol. 264, pp. 1989-1994.
Rosenberg, D. (1991), "Productivity Analysis: DEA with
Fixed Inputs," Journal of Health and Human Resources
Administration, Vol. 14, pp. 84-88.
Rosendberg, D. (1991), "DEA or Dubious Efficiency
Assessment?," Journal of Health and Human Resources
Administration, Vol. 14, pp. 65-74.
Rosendberg, D. (1991), "DEA Preview," Journal of Health and
Human Resources Administration, Vol. 14, pp. 77-83.
Rosendberg, D. (1991), "Forget CEA. Use DEA," Journal of
Health and Human Resources Administration, Vol. 14, pp.
109-112.
Rosendberg, D. (1991), "Practical Productivity Analysis for
Health Care Providers: A Mini-Symposium," Journal of Health
and Human Resources Administration, Vol. 14, pp. 62-64.
Rosendberg, D. (1991), "Productivity Analysis: DEA with
Fixed Inputs," Journal of Health and Human Resources
Administration, Vol. 14, pp. 84-88.
Rosko, M. (1984), "The Impact of Prospective Payment: A
Multidimensional Analysis of New Jersey's SHARE Program,"
Journal of Health Politics, Policy and Law, Vol. 9, pp. 81-
102.
168
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Rosko, M. (1988), "DRGs and Severity of Illness Measures:
An Analysis of Patient Classification Systems," Journal of
Medical Systems, Vol. 12, pp. 257-274.
Rosko, M. (1989), "A Comparison of Hospital Performance
Under the Partial-Payer Medicare PPS and State All-Payer
Rate-Setting Systems," Inquiry, Vol. 26, pp. 48-61.
Rosko, M. (1990), "Measuring Technical Efficiency in Health
Care Organizations," Journal of Medical Systems, Vol. 14,
pp. 307-322.
Rosko, M. and Broyles, R. (1988), The Economics of Health
Care, Greenwood Press, Westport, Connecticut.
Rudnick, J. (1995), "Hospital Layoffs. One Facility's
Experience With Workforce Reduction," Health Progress, Vol.
76, pp. 26-29.
Sear, A. (1991), "Comparison of Efficiency and
Profitability of investor owned multi-hospital systems with
Not-For-Profit Hospitals," Health Care Management Review,
Vol. 16, pp. 31-37.
Sexton, T., Leiken, A., Nolan, A., Liss, S., Hogan, A. and
SilJonan, R. (1989), "Evaluating Managerial Efficiency of
Veterans Administration Medical Centers Using Data
Envelopment Analysis," Medical Care, Vol. 27, pp. 1175-
1188.
Sherman, H.D. (1984), "Hospital Efficiency Measurement and
Evaluation, Empirical Test of a New Technicjue," Medical
Care, Vol. 22, pp. 922-935.
Silkman, R.H. (1986), "Measuring Efficiency: An Assessment
of Data Envelopment Analysis," New Directions For Program
Evaluation, No. 32, Jossey-Bass, Inc. Publishers, San
Francisco, California.
Sloan, F. and Vraciu, R. (1983), "Investor-Owned and Not-
For-Profit Hospitals: Addressing Some Issues," Health
Affairs, Spring, pp. 25-37.
Sorkin, A. (1986), Healthcare and the Changing Economic
Environment, Lexington, Massachusettes: Lexington Books.
169
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Starr, P. (1982), The Social Transformation of American
Medicine, New York: Basic Books.
Starr, Paul (1994), "Why the Clinton. Plan is Not the
Enthoven Plan," Inquiry, Vol. 31, pp. 136-140.
Steinwald, B. and Dummit, L.(1989), "Hospital Case-Mix
Change: Sicker Patients or DRG Creep?," Health Affairs,
Vol. 8, pp. 35-47.
Torrens, P.R. (1988), "Historical Evolution and Overview of
Heath Services in the United States," in Williams and
Torrens, Introduction to Health Services, 3rd Edition.
Valdmanis, V. (1990), "Ownership and Technical Efficiency
of Hospitals," Medical Care, Vol. 28, pp. 552-561.
Valdmanis, V. and S. Grosskopf, S. (1987), "Measuring
Hospital Performance, A Non-Parametric Approach," Journal
of Health Economics, Vol. 6, pp. 89-107.
Wagstaff, A. (1989), "The Measurement of Hospital
Efficiency: A Comment," Social Science and Medicine, Vol.
28, pp. 13-17.
Watt, J. (1986), "The Comparative Economic Performance of
Investor-Owned Chain and Not-For-Profit Hospitals," The New
England Journal of Medicine, Vol. 314, pp. 89-96.
Welch, B. (1987), "The RAND Health Insurance Study: A
Summary Critique," Medical Care, Vol. 25, pp. 148-156.
Wilkinson, A.M. (1989), A Preliminary Analysis of
Beneficiary Discharge Status and Post Hospital Placement
Before and After the Implementation of Medicare's
Prospective Payment System, Doctoral Dissertation, Portland
State University.
Wilson, F.A. and Neuhauser, D. (1982), Health Services In
the United States, 2nd Edition, Ballinger Publishing
Company Cambridge, Massachusetts.
Wilson, G. and Jadlow, J. (1982), "Competition, Profit
Incentives and Technical Efficiency in the Provision of
Nuclear Medicine Services," The Bell Journal of Economics,
Vol. 13, pp. 472-482.
170
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Young, W. (1982), "The Measurement of Hospital Case-Mix,"
Medical Care, Vol. 20, pp. 501-513.
Young, W., Swinkola, R. and Zorn, D. (1982), "The
Measurement of Hospital Case Mix," Medical Care, Vol. 20,
pp. 501-512.
Zelman, W. (1994), "The Rationale Behind the Clinton Health
Care Reform Plan," Health Affairs, Vol. 13, pp. 9-29.
Zwanziger, J. (1987), Competition In the California
Hospital Industry 1980-1985, Doctoral Dissertation, RAND
Graduate School.
Zwanziger, J. and Melnick, G. (1988a), "The Effects of
Hospital Competition and the Medicare PPS Program on
Hospital Cost Behavior in California," Journal of Health
Economics, Vol. 7, pp. 301-320.
Zwanziger, J., Melnick, G, Mann, J. and Simonson, L.
(1994a), "How Hospitals Practice Cost Containment with
Selective Contracting and the Medicare Prospective Payment
System," Medical Care, Vol. 32, pp. 1153-1162.
Zwanziger, J., Melnick, G. and Bamezai, A. (1994b), "Costs
and price competition in California Hospitals, 1980-1990,"
Health Affairs, Vol. 13, pp. 118-126.
Zwanziger, J., Melnick, G. and Mann, J. (1990), "Measures
of Hospital Market Structure: a Review of the Alternatives
and a Proposed Approach," Socio-Economic Planning Sciences,
Vol. 24, pp. 81-95.
Zwanziger, J. and Melnick, G. (1988b) "Hospital Behavior
Under Competition and Cost Containment Policies, " JAMA,
Vol. 260, pp. 2669-2675.
Zwanziger, J., Melnick, G. and Bradley, T. (1989),
"Competition and Cost Containment in California: 1980-
1987," Health Affairs, Vol. 8, pp. 129-136.
171
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX A
MARKET COMPETITORS
172
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE A.l
Number of Competitors Within a 15 Mile Radius of Hospital
HOSPITAL 1988 HOSPITAL 1988 HOSPITAL 1988 HOSPITAL 1988
1 1 49 8 97 41 145 81
2 1 50 8 98 42 146 82
3 1 51 8 99 42 147 82
4 1 52 9 100 42 148 89
5 1 53 9 101 42 149 89
6 1 54 9 102 46
7 1 55 10 103 47
8 1 56 11 104 48
9 2 57 11 105 48
10 2 58 11 106 48
11 2 59 11 107 49
12 2 60 12 108 50
13 2 61 12 109 50
14 2 62 13 110 51
15 2 63 15 111 51
16 2 64 15 112 51
17 2 65 15 113 51
18 2 66 16 114 51
19 2 67 16 115 52
20 2 68 16 116 54
21 3 69 18 117 55
22 3 70 18 118 55
23 3 71 19 119 57
24 3 72 19 120 57
25 3 73 19 121 58
26 4 74 20 122 59
27 4 75 20 123 61
28 4 76 20 124 61
29 4 77 20 125 61
30 4 78 20 126 62
31 4 79 21 127 63
32 5 80 21 128 63
33 5 81 22 129 66
34 5 82 23 130 66
35 5 83 24 131 66
36 6 84 26 132 67
37 6 85 28 133 68
38 6 86 28 134 69
39 6 87 29 135 70
40 7 88 29 136 70
41 7 89 31 137 70
42 7 90 32 138 70
43 7 91 33 139 74
44 7 92 34 140 74
45 7 93 35 141 75
46 7 94 36 142 75
47 7 95 38 143 77
48 7 96 39 144 78
173
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX B
LOGISTIC REGRESSION
COLLINEARITY AND CORRELATION DIAGNOSTICS
174
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
m
H
i
o CO O CO a\ VO CM VO VO CO fH CM VO o\
H m CM 00 fH o 00 VO o fH VO VO
VO m VO CO iH o m m m CO VO in CO 00 o
1 - 4 CM CM CO VO CM VO CM CO o 00 p* CO VO P- CM VO
( * 1
QD o V o CM CM CM CM p* VO VO o o r4 fH CM o
z
• • • • • • • • • • • • • • • •
p
u 4 ) - K * ♦ ♦ *
o
r —i
o X CO CO CO
£
10 CO CO CO CO CO 00 CO 00 CO 00
•H a* CO 00 CO 00 CO 00 CO 00 o o i - a oc cc
U M X o Q i - a oc z Z z Z u y y y
00 O u o u u u s S Z z z z
> Z U S X z z z Z z 5 X X X X X X
r- o m f - 4 CO CO m CM m CM oo o p- o>
VO m r4 m CO f - 4 o CO VO CO CM fH 00 CM CO p* m
< N J r* CM m CM m CO o r- VO m V 00 CM m o
m CM CO CO VO CM VO CM CO o> 00 p- CO VO P* CM VO
00 o o CM CM CM CM t f ) VO o o fH fH CM o
< • • • • • • • • • • • • • • •
P
H 0) 4 t ♦ $ * *
Q iH
O CO CO CO
£ l O CO CO CO CO CO 00 CO 00 CO 00
•H On CO 00 CO 00 CO 00 CO 00 o o fQ cc cc
W CO
X
o o HP X cc z u U y y
(0 00 o u u u u s z Z z
> < u X z X z Z Z o X X X X X
r4 o CO 00 o a\ m o VO CO CM o m VO V
VO r4 r- a\ o r- CO r- VO o> CM VO CO CM i f ) p^
r* CM co rH 1 - 4 V V a CM 00 m a 00 r4 p- CO r4
8 CM CM vn CM m m 00 O in CM CO a V m m
OT
00 O CO 1 - 4 CM CM CM fH 00 VO VO fH o fH r4 CM o
z • • • • • • • • • • • • • • • •
p
u a * * ♦
Q 1 - 4
Q p- P- C^
£
( 0 r* r- CM 00 CM 00 00
•H 04 CM 00 CM 00 CM 00 CM 00 o O i - a d oc
U
X O Q i - q a d cc Z y y y y
00 o U u u o Q
> z u X z z X
*
X o X X
1 - 4 00 V o r- 00 to o r* p* o VO CM m
VO CM iH a 00 CM 00 r- CM in VO VO CM VO VO 00
CM V CO m o fH 00 r* rH CM Oi 00 00 CO o
8 CM CM VO CM 00 VO 00 r* o m VO Oï CO m if )
(S
OT
00 O CO iH CM CM CM fH 00 VO VO r4 o rH fH CM o
<
• • • • • • • • • • • • • • •
p
M 0 ) 4 t ♦ *
Q 1 - 4
Q p- P- p^
£
(0 r- r- r- p- CM 00 CM 00 CM 00
•H 04 CM 00 CM 00 CM 00 CM 00 O O tA oc cc
M
r- X
o Q hQ Hi d a: Z U O y y
00 o U u u o s
>
< u z Z Z Z z z o X X
m CM m CO CM VO p- r* VO V o CM m C\ rH VO
00 o o tn CM VO VO CO V p- VO o m VO fH CM CO
CM VO o 00 r4 00 fH CO CO p- VO fH
8
CM CO a\ CO CM o I f ) CM CO CO V m CO o VO CM VO
t H
w
00 o CM o CM CM CM CM VO VO fH o CM fH CM o
Z • • • • • * • • • • • * •
P
u 0 ) + * * *
a 1 - 4
Q ua CO CO CO
£ E - * CO CO CO CO fH 00 fH 00 r4 00
i4 04 r4 00 r4 00 r4 00 fH 00 o o i - a c c oc
M
CO X
O Q hQ a oc cc z z u o Ü y
Jg 00 o U o u u S m z z z
> z Ü X X z Z z X o X X X X X
m CM VO CO r~ m f - 4 CM 00 CM oo CO o 00 CM o m
CO t - 4 00 CM V r- V CO CO P- CM fH CO 00 m rH V
VO t r > V o CO r- C^ V O CM CO CO o in VO r4
r- CO 00 CO CO o I f ) CM CO CO in CO fH VO CM VO
rj 00 o CM o CM CM CM CM r* VO VO fH o CM fH CM O
<
• • • • • • * • • • • • • • • • •
P
M o; * * * *
O r4
o
-2
CO CO CO
£
( 0 CO CO CO CO fH 00 r4 00 r4 00
•H 04 r4 00 r4 oo r4 00 fH 00 o o >a » - a CC oc
U
CO X O Q i - a cc cc z Ü u Ü u
Jg 00 o u o u u S
>
< a X zzZzz o X X
in
r*
m
e
•H
V
Xi
■P
< P
o
0 )
c
o
x:
+ j
T J
0 )
-M
< d
I —I
< u
p
p
o
o
Xi
o »
H
Xi
O )
u
n >
O C
*
<
04
X a
E h >1 to
04 4 - »
s
•H x i
o U 4 - 1
o ( d •H
0 ) 5
d
•
•H C O
O' rH 6
•
iH U
0 ) 0 0 )
U -M
6 I I a
u o
0 ) fH ■H
&4 #
4 - 1
U
c a ( d
o n S p
•H x i
0 ) .
4J ■ p 4J C O
Ü C ! 4J
( d C O -H O
u C O ( U
V 0 ) 0 ) (P
-M r4 ‘ W
d 8 - 1 < a
M ( U
p
1 1 o
o M
C O E h
X O
< Z
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
l a
C M
( Q
M
s
K g
OC «-
3
@8
r
8
I'
g s
u 8
§8
X
i
8
8.
8
? K
I § I I I
i I i I I
I I i g i
3 8
8 8 8
P « I
III
^ i i
8
8 # 8
I I I
^ 1 *
I I i
P N
8 §
8
B §
Ml 1 1 1
1 1 1
s
M
s
8 §
R X
C O a
8
P
S
5
; I g
1 1
Î
1 1 1
8 8 8
? ‘ “
8
? T ?
I i I 8
: § N 8 8
S ; K 3 8
r I * I *
8 5 8 8 8
II U
I g I §
I
s S
8 5
i g I I s
I §
I i I
IS *
8 5 ? 5
5 % ^ o
to to
8 ^
u
O'
«
> 0
r»
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CD
■D
O
Q .
C
g
Q .
■ D
CD
C/)
W
o"
3
0
3
CD
8
■ D
( O '
3 "
1
3
CD
"n
c
3 .
3"
CD
CD
■ D
O
Q .
C
a
O
3
■ D
O
CD
Q .
■ D
CD
( / )
( / )
TABLE B.2b
MODEL A2
CORRELATION COEFFICIENTS
A87 COMPT HMO
2
HMO
87
MOL
2
MOL
87
MGR
2
MGR
87
MEM
2
MEM
87
OWN XHMO
2
XHMO
87
XMGL
2
XMCL8
7
XMGR
2
XMGR
87
A87 1.0000 - . 1 1 7 3 . 0 2 3 5 . 1 6 8 6 - . 1 2 9 1 . 1 1 9 4 - . 0 9 6 9 - . 0 1 9 8 . 0 0 2 8 . 1 4 5 8 . 1 3 3 1 - . 0 9 8 8 . 0 4 8 7 - . 1 1 1 1 . 0 03 2 - . 0 6 6 7 - . 0 7 6 1
COMPT - . 1 1 7 3 1. 00 00 . 0 6 0 9 . 0 7 9 8 - . 0 5 3 4 - . 1 6 / 8 - . 1 2 2 9 0066 - . 0 5 9 2 . 0 8 7 9 . 3 25 4 . 7 7 0 1 . 4 0 2 0 . 7 1 0 4 .4 853 . 4 3 1 7 . 8 3 3 7
HM02 . 02 3 5 . 0 6 0 9 1. 00 00 - . 1 4 8 5 - . 0 8 2 1 - . 1 5 1 3 - . 0 9 2 3 . 0 61 2 . 0 5 0 4 - . 0 5 6 8 . 0 9 4 3 .5 3 7 0 - . 1 6 9 1 . 0 4 1 6 - . 0 7 4 2 - . 0 7 6 5 . 0 9 4 0
HM087 .1 68 6 . 0 7 9 8 - . 1 4 8 5 1 .0 0 0 0 - . 2 2 4 0 - . 1 0 9 4 - . 0 5 3 3 - . 4 2 9 4 - . 0 2 0 5 - . 0 2 4 2 . 0 4 2 7 - . 0 9 7 8 . 8 6 2 8 - . 1 1 2 9 - . 0 6 8 0 .0 9 5 7 - . 1 6 6 6
MCL2 - . 1 2 9 1 - . 0 5 3 4 - . 0 8 2 1 - . 2 2 4 0 1. 00 00 - . 0 4 5 3 - . 2 2 3 0 . 0 2 8 0 . 0 1 2 6 - . 0 7 5 5 - . 1 7 8 3 - . 0 4 2 7 - . 2 1 8 9 . 5 3 3 8 - . 0 6 4 1 - . 2 1 9 9 - . 0 2 7 1
MCL87 .1194 - . 1 6 7 8 - . 1 5 1 3 - . 1 0 9 4 - . 0 4 5 3 1 . 0 0 0 0 - . 1 1 7 6 - . 2 8 2 1 . 0 2 5 5 . 0 3 6 4 - . 0 4 8 5 - . 2 0 8 4 - . 1 6 6 4 - . 1 4 9 6 .62 0 7 - . 1 4 5 7 - . 2 6 0 2
MCR2 - . 0 9 6 9 - . 1 2 2 9 - . 0 9 2 3 - . 0 5 3 3 - . 2 2 3 0 - . 1 1 7 6 1. 00 00 . 0 4 8 5 - . 0 2 8 1 . 0 1 7 4 . 0 5 8 0 - . 1 6 8 5 . 0 0 4 1 - . 2 3 4 9 - . 1 2 8 6 . 6 5 0 5 - . 1 0 4 1
MCR87 - . 0 1 9 8 . 0 0 6 6 . 0 6 1 2 - . 4 2 9 4 . 0 2 8 0 - . 2 8 2 1 . 0 4 8 5 1. 0000 . 1 2 5 1 . 0 4 8 3 - . 0 0 6 4 .0 6 2 2 - . 3 7 1 4 . 0 2 8 7 - . 2 0 5 1 . 0 0 0 0 . 4 9 0 4
MEM2 .0 0 2 8 - . 0 5 9 2 . 0 5 0 4 - . 0 2 0 5 . 0 1 2 6 . 0 2 5 5 - . 0 2 8 1 . 1 2 5 1 1 . 0 0 0 0 - . 3 6 6 2 - . 1 9 8 1 . 0 2 1 7 - . 0 3 3 7 - . 0 0 1 0 - . 0 8 6 7 - . 0 6 3 4 . 0 4 0 1
MEM87 .1 4 5 9 . 0 8 7 9 - . 0 5 6 8 - . 0 2 4 2 - . 0 7 5 5 .0 3 6 4 . 0 1 7 4 . 0 4 8 3 - . 3 6 6 2 1 . 0 0 0 0 . 4 5 0 6 . 0 3 2 9 - . 0 1 8 4 - . 0 0 7 9 .1 6 1 0 . 1 2 6 8 . 0 6 5 3
OWN . 1 3 3 1 . 3 2 5 4 . 0 9 4 3 . 0 4 2 7 - . 1 7 8 3 - . 0 4 8 5 . 0 5 8 0 - . 0 0 6 4 - . 1 9 8 1 . 4 5 0 6 1. 0 0 0 0 . 2 4 2 4 . 1 3 5 3 . 1 4 1 9 . 2 2 1 1 . 2 5 6 8 . 2 5 8 3
XHM02 - . 0 9 8 8 . 7 7 0 1 . 5 3 7 0 - . 0 9 7 8 - . 0 4 2 7 - . 2 0 8 4 - . 1 6 8 5 . 0 6 2 2 . 0 2 1 7 . 0 3 2 9 . 2 4 2 4 1. 0 0 0 0 . 1 4 8 9 . 5 4 5 6 .28 7 6 . 2 4 7 1 . 6 7 7 9
XHM087 .0 4 8 7 . 4 0 2 0 - . 1 8 9 1 "8628 - . 2 1 8 9 - . 1 6 6 4 . 0 0 4 1 - . 3 7 1 4 - . 0 3 3 7 - . 0 1 8 4 . 1 3 5 3 . 1 4 8 9 1 . 0 0 0 0 . 1 1 4 5 .06 7 4 .23 5 4 . 1 0 0 0
XMCL2 -.1111 . 7 1 0 4 . 0 4 1 6 - . 1 1 2 9 . 5 3 3 8 - . 1 4 9 6 - . 2 3 4 9 .0 2 8 7 - . 0 0 1 0 - . 0 0 7 9 . 1 4 1 9 . 5 4 5 6 . 1 1 4 5 1. 00 00 . 3 1 1 7 . 1 3 6 1 . 6 0 7 4
XMCLB7 . 0 0 3 2 . 4 8 5 3 - . 0 7 4 2 - . 0 6 8 0 - . 0 6 4 1 .6 2 0 7 - . 1 2 8 6 - . 2 0 5 1 - . 0 8 6 7 . 1 6 1 0 . 2 2 1 1 . 2 8 7 6 . 0 8 7 4 . 3 1 1 7 1. 00 00 . 1 2 9 8 . 2 7 3 7
XMCR2 - . 0 6 8 7 . 4 3 1 7 - . 0 7 6 5 . 0 9 5 7 - . 2 1 9 9 - . 1 4 5 7 . 6 5 0 5 .0 0 0 0 - . 0 6 3 4 . 1 2 6 6 . 2 5 6 8 . 2 4 7 1 . 2 3 5 4 . 1 3 6 1 . 1 2 9 8 1. 0000 . 3 5 8 1
XMCR87 - . 0 7 6 1 "8337 . 0 9 4 0 - . 1 8 6 6 - . 0 2 7 1 - . 2 6 0 2 - . 1 0 4 1 .49 0 4 . 0 4 0 1 . 0 6 5 3 . 2 5 8 3 . 6 7 7 9 . 1 0 0 0 . 6 0 7 4 .2 73 7 . 3 5 8 1 1. 00 00
** A Score of 1 or -1 = Correlation 'X' = Interaction Term (e.g., COMPTxPAYER)
177
CD
■D
O
Q .
C
g
Q .
■ D
CD
C/)
W
o"
3
0
5
CD
8
c i '
3 "
1
3
CD
3 .
3"
CD
CD
■ D
O
Q .
C
a
O
3
■ D
O
CD
Q .
■ D
CD
C/)
C/)
TABLE B.2c
MODEL A3
CORRELATION COEFFICIENTS
A83 COMPT HMO
3
HMO
63
MCL
3
MCL
63
MCR
3
MCR
63
MEM
3
MEM
63
OWN XHMO
3
XHMO
63
XMCL
3
XMCL8 XMCR
3
XMCR
83
A83 1 . 00 00 - . 1 5 7 1 - . 0 9 6 1 - . 1 4 3 6 - . 0 7 3 4 - . 0 3 7 0 . 0 7 4 0 . 00 14 . 0 5 3 5 . 1 0 4 7 . 1 3 7 3 - . 1 9 3 0 - . 1 3 9 0 - . 1 6 0 6 - . 1 5 2 4 - . 0 5 0 4 - . 1 0 7 1
COMPT - . 1 5 7 1 1 . 0 00 0 . 0 8 7 3 . 21 73 - . 0 0 5 7 - . 1 8 2 2 - . 1 6 1 3 . 0 0 1 2 - . 0 3 7 3 . 0 7 9 1 . 3 2 5 4 . 9 0 6 1 . 2 8 1 6 . 6 3 6 9 . 4 9 3 0 . 4 5 6 6 . 8 5 2 6
HM03 - . 0 9 8 1 . 0 6 7 3 1. 00 00 - . 0 9 4 0 - . 1 3 9 3 - . 1 0 6 5 . 0 3 0 9 . 0 5 3 1 . 0 3 2 1 . 0 7 2 7 . 0 6 6 4 . 3 6 1 5 - . 1 1 0 1 - . 0 1 2 6 - . 0 3 7 9 . 1 39 4 . 0 8 3 3
HMOB3 - . 1 4 3 6 . 2 1 7 3 - . 0 9 4 0 1.00 00 - . 1 1 6 9 - . 0 3 6 2 - . 0 7 5 2 - . 2 1 3 2 - . 1 3 7 3 - . 0 0 6 2 . 0 6 5 0 . 11 76 . 9 7 6 2 . 0 6 0 0 . 09 34 . 0 27 3 . 0 6 5 6
MCL3 - . 0 7 3 4 - . 0 0 5 7 - . 1 3 9 3 - . 1 1 6 9 1.000 0 . 14 06 - . 2 6 0 0 . 1 6 5 6 . 0 5 2 0 - . 1 0 0 7 - . 1 0 6 5 - . 0 4 6 6 - . 0 9 3 3 . 6 0 9 6 . 0 4 2 0 - . 1 4 0 6 . 0 8 6 1
MCL83 - . 0 3 7 0 - . 1 6 2 2 - . 1 0 6 5 - . 0 3 6 2 . 1 4 0 6 1.000 0 - . 0 6 6 7 - . 3 9 0 3 . 0 6 5 1 - . 0 1 6 6 - . 1 1 2 4 - . 2 0 9 9 - . 0 6 4 2 - . 0 7 5 4 . 6 1 8 2 - . 1 2 9 9 - . 3 3 9 0
MCR3 . 0 7 4 0 - . 1 6 1 3 . 0 30 9 - . 0 7 5 2 - . 2 6 0 0 - . 0 6 6 7 1. 00 00 - . 1 6 2 4 - . 0 6 5 1 . 0 7 1 2 . 0 6 5 1 - . 0 9 4 2 - . 1 0 6 6 - . 2 2 0 5 - . 1 2 3 7 . 6 2 4 1 - . 1 6 1 3
MCR83 . 0 0 1 4 . 0 0 1 2 . 0 5 3 1 - . 2 1 3 2 . 1 6 5 6 - . 3 9 0 3 - . 1 6 2 4 1 . 00 00 . 0 6 2 9 . 0 6 3 0 - . 0 2 6 7 . 00 94 - . 1 9 1 4 . 1 3 9 6 - . 3 1 2 6 - . 0 7 6 6 . 4 3 6 8
MEM3 . 0 5 3 5 - . 0 3 7 3 . 0 3 2 1 - . 1 3 7 3 . 0 5 2 0 . 0 6 5 1 - . 0 6 5 1 . 0 6 2 9 1 . 0 00 0 - . 4 4 2 6 - . 0 8 9 7 - . 0 2 7 0 - . 1 3 1 5 . 0 3 3 5 . 0 1 5 2 - . 0 4 2 3 - . 0 1 8 1
MEM83 . 1 0 4 7 . 0 7 9 1 . 0 7 2 7 - . 0 0 6 2 - . 1 0 0 7 - . 0 1 6 6 . 0 7 1 2 . 0 6 3 0 - . 4 4 2 6 1. 00 00 . 3 7 9 3 . 1 1 2 6 - . 0 2 4 6 - . 0 4 5 3 . 06 94 . 1 4 3 1 . 0 8 0 3
OWN . 1 3 7 3 . 3 2 5 4 . 0 66 4 . 0 65 0 - . 1 0 6 5 - . 1 1 2 4 . 0 6 5 1 - . 0 2 6 7 - . 0 6 9 7 . 3 7 9 3 1 . 0 00 0 . 2 6 3 0 . 1 0 3 9 . 1 4 7 1 . 1 7 7 6 . 2 7 6 1 . 2 5 6 0
XHM03 - . 1 9 3 0 * * . 9 0 6 1 . 3 6 1 5 . 11 76 - . 0 4 6 6 - . 2 0 9 9 - . 0 9 4 2 . 0 0 9 4 - . 0 2 7 0 . 1 1 2 6 . 2 8 3 0 1.00 00 . 1 7 5 4 . 5 3 6 7 . 3 9 6 8 . 4 7 1 7 . 7 7 7 7
XHM083 - . 1 3 9 0 . 2 6 1 6 - . 1 1 0 1 * * . 9 7 6 2 - . 0 9 3 3 - . 0 6 4 2 - . 1 0 6 6 - . 1 9 1 4 - . 1 3 1 5 - . 0 2 4 6 . 1 0 3 9 . 1 7 5 4 1. 00 00 . 0 90 4 . 1 2 5 0 . 05 62 . 1 1 9 9
XMCL3 - . 1 6 0 6 . 6 3 6 9 - . 0 1 2 6 . 05 00 . 6 0 9 6 - . 0 7 5 4 - . 2 2 0 5 . 1 3 9 6 . 0 3 3 5 - . 0 4 5 3 . 1 4 7 1 . 5 3 6 7 . 0 9 0 4 1 .0 000 . 3 5 7 4 . 1 6 6 3 . 6 3 0 1
XMCL83 - . 1 5 2 4 . 4 9 3 0 - . 0 3 7 9 . 09 34 . 0 4 2 0 . 6 1 6 2 - . 1 2 3 7 - . 3 1 2 8 . 0 1 5 2 . 0 6 9 4 . 1 7 7 6 . 3 9 6 6 . 1 2 5 0 . 35 74 1 .0 000 . 1 7 4 3 . 2 2 3 6
XMCR3 - . 0 5 0 4 . 4 5 5 5 . 13 94 . 02 73 - . 1 4 0 6 - . 1 2 9 9 . 6 2 4 1 - . 0 7 6 6 - . 0 4 2 3 . 1 4 3 1 . 2 7 6 1 . 4 7 1 7 . 0 5 6 2 . 1 5 6 3 . 1 7 4 3 1. 00 00 . 3 3 8 5
XMCR83 - . 1 0 7 1 * * . 6 5 2 5 . 0 6 3 3 . 06 56 . 0 6 5 1 - . 3 3 9 0 - . 1 6 1 3 4366 - . 0 1 6 1 . 0 8 0 3 . 2 5 6 0 . 7 7 7 7 . 1 1 9 9 . 6 3 0 1 . 2 2 3 8 . 3 3 8 6 1 . 0 00 0
** A Score of 1 or -1 = Correlation 'X' = Interaction Term (e.g., COMPTxPAYER)
178
CD
■D
O
Q .
C
g
Q .
■ D
CD
C/)
W
O *
3
O
8
c i '
3 "
i
3
CD
3 .
3"
CD
CD
■ D
O
Q .
C
a
O
3
■ D
O
CD
Q .
■ D
CD
C/)
C/)
TABLE B.3a
MODEL NI
CORRELATION COEFFICIENTS
N83 COMPT HMO
1
HMO
83
MCL
1
MCL
83
MCR
1
MCR
83
MEM
1
MEM
83
OWN XHMO
1
XHMO
83
XMCL
1
XMCL8
3
XMCR
1
XMCR
83
N83 1.00 00 - . 2 4 5 3 - . 1 0 7 9 . 0 5 4 3 . 0 2 8 5 - . 0 9 5 1 - . 0 3 8 8 . 0 7 5 0 - . 0 9 6 2 . 09 43 . 08 59 - . 2 6 0 1 . 0 4 2 5 - . 1 3 5 6 - . 2 2 6 4 - . 1 9 2 9 - . 1 5 8 9
COMPT - . 2 4 5 3 1 . 0 0 0 0 - . 0 1 5 6 . 2 1 7 3 - . 1 0 1 8 - . 1 8 2 2 - . 0 0 3 2 . 0 0 1 2 . 0 1 5 6 . 0 7 9 1 . 32 54 . 6 9 1 4 . 2 8 1 6 . 44 76 . 4 9 3 0 . 4 5 5 5 . 8 5 2 5
HM01 - . 1 0 7 9 - . 0 1 5 6 1.0 00 0 . 04 36 - . 0 1 1 9 - . 0 9 3 8 - . 1 3 6 9 - . 0 0 2 6 . 0 2 7 9 - . 0 6 7 8 - . 1 4 2 3 . 5 6 6 3 . 0 1 3 7 - . 1 1 2 5 - . 1 1 4 7 - . 0 6 8 1 - . 0 0 9 6
HM083 . 0 5 4 3 . 2 1 7 3 . 0 4 3 6 1.0 00 0 - . 0 6 0 0 - . 0 3 8 2 - . 0 8 2 8 - . 2 1 3 2 - . 1 1 6 1 - . 0 0 8 2 . 0 6 5 0 . 1 6 6 5 . 9 7 6 2 . 0 7 9 0 . 0 9 3 4 - . 0 0 8 2 . 0 6 5 8
MCL1 . 0 2 8 5 - . 1 0 1 8 - . 0 1 1 9 - . 0 6 0 0 1. 0 00 0 . 13 44 - . 2 5 9 4 . 14 24 . 0 8 4 9 - . 1 3 5 5 - . 0 4 6 9 - . 1 5 4 2 - . 0 4 4 8 . 6 5 6 1 - . 0 7 3 7 - . 1 6 2 1 - . 0 0 7 4
MCLB3 - . 0 9 5 1 - . 1 8 2 2 - . 0 9 3 8 - . 0 3 8 2 . 13 44 1. 00 00 - . 0 4 8 4 - . 3 9 0 3 . 03 58 - . 0 1 6 8 - . 1 1 2 4 - . 2 1 6 5 - . 0 6 4 2 - . 1 0 6 6 . 6 1 8 2 - . 0 9 5 2 - . 3 3 9 0
MCR1 - . 0 3 8 8 - . 0 0 3 2 - . 1 3 6 9 - . 0 8 2 8 - . 2 5 9 4 - . 0 4 8 4 1 . 0 0 0 0 - . 1 4 0 6 . 0 6 7 0 . 078 4 . 0 5 2 1 - . 0 5 1 4 - . 0 9 5 2 - . 1 1 9 1 - . 0 1 3 3 . 7 0 5 3 - . 0 2 8 3
MCR83 . 0 7 5 0 . 0 0 1 2 - . 0 0 2 6 - . 2 1 3 2 . 14 24 - . 3 9 0 3 - . 1 4 0 6 1 . 00 00 - . 0 6 5 2 . 06 30 - . 0 2 6 7 . 0 0 6 1 - . 1 9 1 4 . 1 4 4 9 - . 3 1 2 8 - . 0 4 5 2 . 4 3 6 8
MEM1 - . 0 9 6 2 . 0 1 5 6 . 02 79 - . 1 1 6 1 . 0 8 4 9 . 0 3 5 8 . 0 6 7 0 - . 0 6 5 2 1.0 00 0 - . 3 7 0 6 . 08 94 - . 0 3 9 3 - . 1 0 5 1 . 0 85 8 . 0 6 0 3 . 0 3 4 3 - . 0 2 3 4
MEM83 . 09 43 . 0 7 9 1 - . 0 6 7 8 - . 0 0 8 2 - . 1 3 5 5 - . 0 1 6 8 . 0 7 8 4 . 0 6 3 0 - . 3 7 0 6 1.0 000 . 37 93 . 0 1 4 2 - . 0 2 4 8 - . 0 5 7 1 . 08 94 . 0 8 1 9 . 0 8 0 3
OWN . 0 8 5 9 . 3 2 5 4 - . 1 4 2 3 . 0 6 5 0 - . 0 4 6 9 - . 1 1 2 4 . 0 5 2 1 - . 0 2 6 7 . 08 94 . 37 93 1.00 00 . 0 8 0 6 . 1 0 3 9 . 1 3 2 9 . 1 7 7 8 . 1 8 2 2 . 2 5 6 0
XHM01 - . 2 6 0 1 . 6 9 1 4 . 5 6 6 3 . 1 6 6 5 - . 1 5 4 2 - . 2 1 6 5 - . 0 5 1 4 . 0 0 6 1 - . 0 3 9 3 . 0 1 4 2 . 0 8 0 8 1 . 0 0 0 0 .2111 . 2 1 4 0 . 2 4 4 1 . 2 5 9 7 . 5 9 2 7
XHM083 . 0 4 2 5 . 2 8 1 6 . 01 37 * * . 9 7 6 2 - . 0 4 4 8 - . 0 6 4 2 - . 0 9 5 2 - . 1 9 1 4 - . 1 0 5 1 - . 0 2 4 8 . 1 0 3 9 . 2 1 1 1 1 . 0 00 0 . 1 0 7 7 . 1 2 5 0 . 0 2 2 5 . 1 1 9 9
XMCL1 - . 1 3 5 6 . 4 4 7 6 - . 1 1 2 5 . 0 7 9 0 . 6 5 6 1 - . 1 0 6 6 - . 1 1 9 1 . 1 4 4 9 . 0 8 5 8 - . 0 5 7 1 . 13 29 . 2 1 4 0 . 1 0 7 7 1.00 00 . 1 9 3 8 . 0 7 1 9 . 4 7 2 1
XMCL83 - . 2 2 6 4 . 4 9 3 0 - . 1 1 4 7 . 09 34 - . 0 7 3 7 . 61 62 - . 0 1 3 3 - . 3 1 2 8 . 0 6 0 3 .0 694 . 17 76 . 2 4 4 1 . 1 2 5 0 . 1 9 3 8 1 . 00 00 . 2 1 1 4 . 2 2 3 8
XMCR1 - . 1 9 2 9 . 4 5 5 5 - . 0 6 8 1 - . 0 0 6 2 - . 1 6 2 1 - . 0 9 5 2 . 7 0 5 3 - . 0 4 5 2 . 0 3 4 3 .0 819 . 1 8 2 2 . 2 5 0 7 . 0 2 2 5 . 0 7 1 9 . 21 14 1. 00 00 3596
XMCR83 - . 1 5 8 9 * * . 8 5 2 5 - . 0 0 9 6 . 06 58 - . 0 0 7 4 - . 3 3 9 0 - . 0 2 8 3 . 4 3 6 8 - . 0 2 3 4 . 06 03 . 2 56 0 . 5 9 2 7 . 11 99 . 4 7 2 1 . 2 2 3 8 3598 1.0 00 0
** A Score of 1 or -1 = Correlation 'X' = Interaction Term (e.g., COMPTxPAYER)
179
CD
■D
O
Q .
C
g
Q .
■ D
CD
C/)
W
o"
3
0
3
CD
8
c i '
3 "
1
3
CD
"n
c
3 .
3"
CD
CD
■ D
O
Q .
C
a
O
3
■ D
O
CD
Q .
■ D
CD
C/)
C/)
TABLE B.3b
MODEL N2
CORRELATION COEFFICIENTS
N87 COMPT
HMO
2
HMO
87
MOL
2
MCU
87
M6R
2
MCR
87
MEM
2
MEM
87
OWN
XHM?
2
“XFIMÛ
83
XMCL
2
XMCT
83
XMCR
2
XMCR
83
N87 1 . 0 00 0 -.0518 -.0606 .1519 .0265 -.1548 .0623 - . 0 2 2 2 -.1055 .0634 .1941 -.0351 .0779 -.0739 -.0954 .0861 -.0650
COMPT -.0618 1 . 0 0 0 0 .0609 .0798 -.0534 -.1678 -.1229 .0066 -.0592 .0679 .3254 .7701 .4020 .7104 .4853 .4317 .8337
HM02 .0606 .0609 1 . 0 0 0 0 -.1485 -.0821 -.1513 -.0923 .0612 .0504 -.0568 .0943 .5370 -.1691 .0416 -.0742 -.0785 .0940
HM087 .1519 .0796 -.1485 1. 0 0 0 0 -.2240 -.1094 -.0533 -.4294 -.0205 -.0242 .0427 -.0978 .8628 -.1129 -.0680 .0957 -.1666
MCL2 .0265 -.0534 -.0821 -.2240 1 . 0 0 0 0 -.0453 -.2230 .0280 .0126 -.0755 -.1783 -.0427 -.2189 .5338 -.0641 -.2199 -.0271
MCL87 -.1548 -.1678 -.1513 -.1094 -.0453 1. 0 00 0 -.1176 -.2821 .0255 .0364 -.0485 -.2064 -.1664 -.1496 .6207 -.1457 -.2602
MCR2 .0623 -.1229 -.0923 -.0533 -.2230 -.1176 1. 00 00 .0485 -.0281 .0174 .0580 -.1685 .0041 -.2349 -.1286 .6605 -.1041
MCR87 - . 0 2 2 2 .0066 .0612 -.4294 .0280 -.2821 .0485 1.0 00 0 .1251 .0483 -.0064 .0622 -.3714 .0287 -.2051 . 0 0 0 0 .4904
MEM2 -.1055 -.0592 .0504 -.0205 .0126 .0255 -.0281 .1251 1. 0 0 0 0 -.3662 -.1981 .0217 -.0337 - . 0 0 1 0 -.0667 -.0634 .0401
MEM87 .0634 .0879 -.0568 -.0242 -.0755 .0364 .0174 .0483 -.3662 1 . 0 0 0 0 .4506 .0329 -.0184 -.0079 .1610 .1266 .0663
OWN .1941 .3254 .0943 .0427 -.1783 -.0485 .0580 -.0064 -.1981 .4506 1 . 0 00 0 .2424 .1353 .1419 . 2 2 1 1 .2568 .2583
XHM02 -.0351 .7701 .5370 -.0978 -.0427 -.2084 -.1685 .0622 .0217 .0329 .2424 1 . 0 0 0 0 .1489 .5456 .2876 .2471 .6779
XHM087 .0779 .4020 -.1691 -.8628 -.2189 -.1664 .0041 -.3714 -.0337 -.0184 .1353 .1489 1. 00 00 .1145 .0874 .2354 . 1 0 0 0
XMCL2 -.0739 .7104 .0416 -.1129 .5338 -.1496 -.2349 .0287 - . 0 0 1 0 -.0079 .1419 .5456 .1145 1.00 00 .3117 .1361 .6074
XMCL87 -.0954 .4853 -.0742 -.0680 -.0641 .6207 -.1286 -.2051 -.0867 .1610 . 2 2 1 1 .2876 .0674 .3117 1.00 00 .1298 .2737
XMCR2 .0661 .4317 -.0765 .0957 -.2199 -.1457 .6505 . 00 00 -.0634 .1266 .2568 .2471 .2354 .1361 .1298 1. 00 00 .3681
XMCR87 -.0650 -.8337 .0940 -.1666 -.0271 -.2602 -.1041 .4904 .0401 .0653 .2583 .8779 . 10 00 .6074 .2737 .3581 1 . 0 0 0 0
** A Score of 1 or -1 = Correlation 'X' = Interaction Term (e.g., COMPTxPAYER)
180
CD
■D
O
Q .
C
g
Q .
■ D
CD
C/)
W
o"
3
0
3
CD
8
■ D
( O '
3 "
1
3
CD
3 .
3"
CD
CD
■ D
O
Q .
C
a
O
3
■ D
O
CD
Q .
■ D
CD
( / )
( / )
TABLE B.3c
MODEL N3
CORRELATION COEFFICIENTS
N63 COMPT HMO
3
HMO
83
MCL
3
MCL
83
MCR
3
MCR
83
MEM
3
MEM
83
OWN XHMO
3
XHM08 XMCL
3
XMCL8 XMCR
3
XMCR8
N83 1. 00 00 - . 2 4 5 3 - . 1 1 0 7 . 0 5 4 3 - . 0 6 9 8 - . 0 9 5 1 . 1 5 6 7 . 0 7 5 0 - . 0 2 0 5 . 0 9 4 3 . 0 8 5 0 - . 2 3 6 2 . 0 4 2 5 - . 2 5 3 7 - . 2 2 6 4 - . 0 5 8 1 - . 1 5 8 9
COMPT . 2 4 5 3 1. 0 0 0 0 . 0 8 7 3 . 2 1 7 3 - . 0 0 5 7 - . 1 8 2 2 - . 1 6 1 3 . 0 0 1 2 - . 0 3 7 3 . 0 7 9 1 . 3 25 4 . 9 0 6 1 . 2 8 1 6 . 6 3 6 8 . 4 8 3 0 . 4 5 5 5 . 8 5 2 5
HM03 - . 1 1 0 7 . 0 8 7 3 1. 0 0 0 0 - . 0 9 4 0 - . 1 3 9 3 - . 1 0 6 5 . 0 3 0 9 . 0 5 3 1 . 0 3 2 1 . 0 7 2 7 . 0 6 8 4 . 3 8 1 5 - . 1 1 0 1 - . 0 1 2 6 - . 0 3 7 9 . 1 3 9 4 . 0 8 3 3
HM083 . 0 5 4 3 . 2 1 7 3 - . 0 9 4 0 1. 0 0 0 0 - . 1 1 6 9 - . 0 3 8 2 - . 0 7 5 2 - . 2 1 3 2 - . 1 3 7 3 - . 0 0 8 2 . 0 6 5 0 . 1 1 7 6 . 9 7 6 2 . 0 6 0 0 .0 9 3 4 . 0 2 7 3 . 0 6 6 8
MCL3 - . 0 6 9 6 - . 0 0 5 7 - . 1 3 9 3 - . 1 1 6 9 1. 0 0 0 0 . 1 4 0 8 - . 2 6 0 0 . 1 8 5 8 . 0 5 2 0 - . 1 0 0 7 - . 1 0 8 5 - . 0 4 6 8 - . 0 9 3 3 . 6 0 9 6 . 0 4 2 0 - . 1 4 0 6 . 0 8 5 1
MCL83 - . 0 9 5 1 - . 1 8 2 2 - . 1 0 6 5 - . 0 3 8 2 . 1 4 0 8 1. 0 0 0 0 - . 0 6 8 7 - . 3 9 0 3 . 0 6 5 1 - . 0 1 8 8 - . 1 1 2 4 - . 2 0 9 9 - . 0 6 4 2 - . 0 7 5 4 . 6 1 8 2 - . 1 2 9 9 - . 3 3 9 0
MCR3 . 1 5 6 7 - . 1 6 1 3 . 0 3 0 9 - . 0 7 5 2 - . 2 6 0 0 - . 0 6 8 7 1 . 0 0 0 0 - . 1 6 2 4 - . 0 6 5 1 . 0 7 1 2 . 0 6 5 1 - . 0 9 4 2 - . 1 0 8 8 - . 2 2 0 6 - . 1 2 3 7 . 6 2 4 1 - . 1 8 1 3
MCR83 . 0 7 5 0 . 0 0 1 2 . 0 5 3 1 - . 2 1 3 2 . 1 6 5 8 - . 3 9 0 3 - . 1 6 2 4 1. 00 00 . 0 6 2 9 . 0 6 3 0 - . 0 2 6 7 . 0 0 9 4 - . 1 9 1 4 . 1 3 9 6 - . 3 1 2 8 - . 0 7 8 8 . 4 3 6 8
MEM3 - . 0 2 0 5 - . 0 3 7 3 . 0 3 2 1 - . 1 3 7 3 . 0 5 2 0 . 0 6 5 1 - . 0 6 5 1 . 0 6 2 9 1 . 0 0 0 0 - . 4 4 2 8 - . 0 8 9 7 - . 0 2 7 0 - . 1 3 1 5 . 0 3 3 5 . 0 1 5 2 - . 0 4 2 3 - . 0 1 8 1
MEM83 . 0 9 4 3 . 0 7 9 1 . 0 7 2 7 - . 0 0 8 2 - . 1 0 0 7 - . 0 1 6 8 . 0 7 1 2 . 0 6 3 0 - . 4 4 2 8 1. 0 0 0 0 . 3 7 9 3 . 1 1 2 6 - . 0 2 4 8 - . 0 4 5 3 . 0 8 9 4 . 1 4 3 1 . 0 8 0 3
OWN . 0 6 5 9 . 3 2 5 4 .0 6 8 4 .0 6 5 0 - . 1 0 8 5 - . 1 1 2 4 . 0 6 5 1 . 0 2 6 7 - . 0 8 9 7 . 3 7 9 3 1.0 00 0 . 2 8 3 0 . 1 0 3 9 . 1 4 7 1 . 1 7 7 6 . 2 7 6 1 . 2 5 8 0
XHM03 - . 2 3 6 2 ".9061 .3 8 1 5 . 1 1 7 6 - . 0 4 6 8 - . 2 0 9 9 - . 0 9 4 2 . 0 09 4 - . 0 2 7 0 . 1 1 2 6 . 2 8 3 0 1. 0 0 0 0 . 1 7 5 4 . 5 3 6 7 . 3 9 8 8 . 4 7 1 7 . 7 7 7 7
XHM083 . 0 4 2 5 . 2 8 1 6 - . 1 1 0 1 ".9762 - . 0 9 3 3 - . 0 6 4 2 - . 1 0 8 8 - . 1 9 1 4 - . 1 3 1 5 - . 0 2 4 8 . 1 0 3 9 . 1 7 5 4 1. 0 0 0 0 . 0 9 0 4 . 1 2 5 0 . 0 6 6 2 . 1 1 9 9
XMCL3 - . 2 5 3 7 . 6 3 6 9 - . 0 1 2 6 . 0 5 0 0 . 6 0 9 6 - . 0 7 5 4 - . 2 2 0 5 . 1 3 9 8 . 0 3 3 5 - . 0 4 5 3 . 1 4 7 1 . 5 3 6 7 . 0 9 0 4 1. 0 0 0 0 . 3 5 7 4 . 1 5 6 3 . 6 3 0 1
XMCL83 - . 2 2 6 4 . 4 9 3 0 - . 0 3 7 9 . 0 9 3 4 . 0 4 2 0 . 6 1 8 2 - . 1 2 3 7 - . 3 1 2 8 . 0 1 5 2 . 0 8 9 4 . 1 7 7 6 . 3 9 8 8 . 1 2 5 0 . 3 5 7 4 1. 00 00 . 1 7 4 3 . 2 2 3 8
XMCR3 - . 0 5 8 1 . 4 5 5 5 .13 9 4 . 0 2 7 3 - . 1 4 0 6 - . 1 2 9 9 . 6 2 4 1 - . 0 7 8 8 - . 0 4 2 3 . 1 4 3 1 . 2 7 6 1 . 4 7 1 7 .0 5 6 2 . 1 5 6 3 . 1 7 4 3 1 . 0 0 0 0 . 3 3 8 5
XMCR83 - . 1 5 8 9 ".8525 .0 8 3 3 . 0 6 5 8 . 0 8 5 1 - . 3 3 9 0 - . 1 8 1 3 . 4 3 6 8 - . 0 1 8 1 . 0 8 0 3 . 2 5 6 0 . 7 7 7 7 .1 1 9 9 . 6 3 0 1 . 2 2 3 8 . 3 3 8 6 1. 0 0 0 0
** A Score of 1 or -1 = Correlation 'X' = Interaction Term (e.g., COMPTxPAYER)
161
APPENDIX c
DEA SCORES
182
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
g
w
s
o
U J
(D
è
UJ
s
1.00
.95
.90
.85
.80
.75
.70
FIGURE 1
Average DEA Score - Ancillary
,1^ '
M M #
Imj
■
1983 1987
YEAR
1991
UJ
O
K
<
UJ
Q
UJ
O
è
%
1.00
.95
.90
.85
.80
.75
.70
FIGURE 2
Average DEA Score - Nursing
1983 1987
YEAR
c u
########:
' LSLi
yoj
----- . -----
# # # * #
Li:— — sum—
1991
183
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.00
U J
ü
ü 5 .85
a
FIGURE 3
Average DEA Score - Ancillary
By Bed Size Category
W=74.6S;59
1^50,54
1983 1987
YEAR
1991
Large
hM O . 19,23
Bdra-Large
N = e . 1 5 ; 13
g
o
1.00
.95
.90
.85
.80
.75
.70
FIGURE 4
Average DEA Score - Nursing
By Bed Size Category
1983 1987 1991
YEAR
Hmficrp
■srrel
W=74.66,59
EHMBdum
hW6.50.54
H Large
hWO.19.23
HBctra-Large
hW. 15.13
184
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
FIGURE 5
Increase in DEA Score - Ancillary
Percent Out of Total Hospitals With Increase in DEA Score
1983-1987 1987-1991
YEAR
1983-1991
FIGURE 6
Increase in DEA Score - Nursing
Percent Out of Total Hospitals With Increase in DEA Score
HUjKMun
1983-1987 1987-1991
YEAR
1983-1991
185
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX D
PERCENT REDUCTION OF PTEs
IN INEFFICIENT HOSPITALS
186
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE D.la
Ancillary 1983-87
Percent Reduction of FTEs
in Inefficient Hospitals
By Percentile
Percentile Mangmnt Techs RNs Aides Clerks Other Total
25 th 12% 12% 11% 11% 11% 11% 22%
50th 27% 27% 21% 25% 26% 22% 29%
75 th 40% 40% 33% 44% 39% 36% 37%
TABLE D.lb
Ancillary 1987-91
Percent Reduction of FTEs
in Inefficient Hospitals
By Percentile
Percentile Mangmnt Techs RNs Aides Clerks Other Total
25th 22% 22% 18% 20% 22% 19% 26%
50th 31% 32% 27% 29% 31% 30% 33%
75th 42% 43% 37% 47% 44% 42% 41%
TABLE D.lc
Ancillary 1983-91
Percent Reduction of FTEs
in Inefficient Hospitals
By Percentile
Percentile Mangmnt Techs RNs Aides Clerks Other Total
25th 21% 24% 24% 24% 24% 21% 28%
50th 33% 35% 33% 35% 34% 33% 38%
75 th 45% 46% 44% 51% 44% 46% 46
187
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE D.2a
Nursing 1983-87
Percent Reduction of FTEs
in Inefficient Hospitals
By Percentile
Percentile Mangmnt RNs LVNs Aides Clerks Other Total
25th 25% 17% 23% 22% 30% 04% 27%
50th 38% 28% 34% 34% 47% 27% 35%
75th 61% 37% 43% 50% 70% 46% 44%
TABLE D.2b
Nursing 1987-91
Percent Reduction of FTEs
in Inefficient Hospitals
By Percentile
Percentile Mangmnt RNs LVNs Aides Clerks Other Total
25th 23% 18% 22% 19% 28% 12% 25%
50th 38% 31% 33% 33% 42% 37% 34%
75th 56% 38% 42% 43% 64% 50% 42%
TABLE D.2c
Nursing 1983-91
Percent Reduction of FTEs
in Inefficient Hospitals
By Percentile
Percentile Mangmnt RNs LVNs Aides Clerks Other Total
25th 24% 21% 22% 23% 25% 26% 27%
50th 38% 31% 34% 33% 40% 40% 34%
75th 49% 41% 45% 49% 58% 64% 46%
188
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
APPENDIX E
LOGISTIC REGRESSION MODEL OUTPUT
189
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
MODEL Al
(Ancillary 1983-87)
Number of cases included in the analysis: 149
Dependent Variable.. A1
-2 Log Likelihood 199.18818 * Constant is included in the model.
-2 Log Likelihood
Goodness of Fit
Model Chi-Square
Improvement
170.268
150.015
Chi-Square
28.920
28.920
df Significance
17 .0353
17 .0353
Predicted
.00 1.00
0 8 1
Observed ôôôôôôôôôôôôôôôôô
.00 0 o 76 6 15 a
ôôôôôôôôôôôôôôôôô
1.00 1 Ô 28 o 30 a
ôôôôôôôôôôôôôôôôô
Overall 71.14%
Variable
Percent Correct
83.52%
51.72%
B
Variables in the Equation
S.E. Wald df Sig Exp (B)
A83 -4.2232 1.4868 8.0687 1 .0045 -.1745 .0147
COMPT .1249 2.0433 .0037 1 .9513 .0000 1.1330
OWN .7148 .4803 2.2145 1 .1367 .0328 2.0438
HM083 -.4027 .3525 1.3050 1 .2533 .0000 .6685
MCL83 -.0810 .0459 3.1098 1 .0778 -.0746 .9222
MCR83 -.0234 .0326 .5161 1 .4725 .0000 .9768
MEM83 .1771 .4771 .1378 1 .7104 .0000 1.1938
HMOl .9597 .8150 1.3866 1 .2390 .0000 2.6109
MCLl 1.4950 .8586 3.0322 1 .0816 .0720 4.4594
MCRl .3078 .7517 .1677 1 .6822 .0000 1.3604
MEMl -.0089 .6330 .0002 1 .9888 .0000 .9911
INT 1 .4255 .3542 1.4433 1 .2296 .0000 1.5304
INT 2 .0860 .0524 2.6915 1 .1009 .0589 1.0898
INT 3 -.0121 .0399 .0926 1 .7609 .0000 .9879
INT 4 -1.2123 .9864 1.5105 1 .2191 .0000 .2975
INT 5 -.6871 .9907 .4811 1 .4879 .0000 .5030
INT_6 -.3196 .8940 .1278 1 .7207 .0000 .7264
Constant 3.5674 2.1461 2.7632 1 .0965
Interactions :
INT_1 COMPT by HM083 INT_4
INT_2 COMPT by MCL83 INT_5
INT_3 COMPT by MCR83 INT_6
COMPT by HMOl
COMPT by MCLl
COMPT by MCRl
190
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
MODEL A2
(Ancillary 1987-91)
Number of cases included in the analysis: 149
Dependent Variable.. A2
-2 Log Likelihood 204.61403 * Constant is included in the model.
-2 Log Likelihood
Goodness of Fit
Model Chi-Square
Improvement
180.091
146.561
Chi-Square
24.523
24.523
df Significance
17 .1059
17 .1059
Predicted
.00 1.00
0 8 1
Observed ôôôôôôôôôôôôôôôôô
.00 0 Ô 37 Ô 29 Ô
ôôôôôôôôôôôôôôôôô
1.00 1 ô 18 ô 65 ô
ôôôôôôôôôôôôôôôôô
Overall 68.46%
Variable
Percent Correct
56.06%
78.31%
Variables in the Equation
S.E. Wald df Sig Exp(B)
A87 -2.0204 1.3167 2.3547 1 .1249 -.0416 .1326
COMPT 1.4474 2.5024 .3345 1 .5630 .0000 4.2519
OWN -.5038 .4560 1.2206 1 .2692 .0000 .6042
HM087 -.0082 .0395 .0430 1 .8358 .0000 .9918
MCL87 .0274 .0288 .9046 1 .3415 .0000 1.0277
MCR87 -.0432 .0369 1.3660 1 .2425 .0000 .9578
MEM87 1.0573 .4630 5.2150 1 .0224 .1253 2.8785
HM02 .8351 .7252 1.3262 1 .2495 .0000 2.3050
MCL2 .2641 .7612 .1203 1 .7287 .0000 1.3022
MCR2 -.3768 .6739 .3126 1 .5761 .0000 . 6860
|MEM2 1.5445 .7530 4.2074 1 .0402 .1039 4.6855 1
INT 1 .0220 .0461 .2277 1 .6333 .0000 1.0222
INT 2 -.0110 .0360 .0928 1 .7607 .0000 .9891
INT 3 -.0059 .0438 .0181 1 .8931 .0000 .9941
INT 4 -1.4325 .9767 2.1511 1 .1425 -.0272 .2387
INT 5 -.2597 .9544 .0741 1 .7855 .0000 .7712
INT_6 .1281 .8362 .0234 1 .8783 .0000 1.1366
Constant 1.7320 2.3008 .5667 1 .4516
Interactions :
INT_1 COMPT by HM087 INT_4
INT_2 COMPT by MCL87 INT_5
INT 3 COMPT by MCR87 INT 6
COMPT by HM02
COMPT by MCL2
COMPT by MCR2
191
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
MODEL A3
(Ancillary 1983-91)
Number of cases included in the analysis: 149
Dependent Variable.. A3
-2 Log Likelihood 203.58825 * Constant is included in the model.
-2 Log Likelihood
Goodness of Fit
Model Chi-Square
Improvement
171.478
169.462
Chi-Square
32.110
32.110
df Significance
17 .0146
17 .0146
Predicted
.00 1.00
0 Ô 1
Observed ôôôôôôôôôôôôôôôôô
.00 0 Ô 65 S 20 a
ôôôôôôôôôôôôôôôôô
1.00 1 a 27 a 37 o
ôôôôôôôôôôôôôôôôô
Overall
Percent Correct
76.47%
57.81%
68.46%
— —— — -------- —--- Variables in the Equation----—-
Variable B S.E. Wald df Sig Exp (B)
A83 -4.1605 1.4934 7.7615 1 .0053 -.1682 .0156
COMPT .5711 2.2965 .0619 1 .8036 .0000 1.7703
OWN -.8961 .4679 3.6677 1 .0555 -.0905 .4081
HM083 -.0236 .1760 .0180 1 .8932 .0000 .9767
MCL83 -.0342 .0376 .8296 1 .3624 .0000 .9663
MCR83 -.0511 .0323 2.4971 1 .1141 -.0494 .9502
MEM83 .2205 .4768 .2138 1 .6438 .0000 1.2467
HM03 1.2667 .9765 1.6826 1 .1946 .0000 3.5491
MCL3 .5514 .7206 .5855 1 .4441 .0000 1.7357
MCR3 -.0113 .7035 .0003 1 .9871 .0000 .9887
MEM3 1.5722 .5790 7.3718 1 .0066 .1624 4.8171 1
INT 1 .0141 .1801 .0062 1 .9374 .0000 1.0142
INT 2 .0720 .0481 2.2384 1 .1346 .0342 1.0747
INT 3 .0180 .0394 .2087 1 .6478 .0000 1.0182
INT 4 -1.6178 1.3691 1.3965 1 .2373 .0000 .1983
INT 5 -.7658 .9118 .7054 1 .4010 .0000 .4650
INT_6 -.3159 .8849 .1275 1 .7211 .0000 .7291
Constant 4.1184 2.2656 3.3044 1 .0691
Interactions :
INT 1 COMPT by HM083 INT 4 COMPT by HM03
INT 2 COMPT by MCL83 INT 5 COMPT by MCL3
INT 3 COMPT by MCR83 INT 6 COMPT by MCR3
192
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
MODEL Ml
(Nursing 1983-87)
Number of cases included in the analysis: 149
Dependent Variable.. N1
-2 Log Likelihood 205.74504 * Constant is included in the model.
-2 Log Likelihood
Goodness of Fit
Model Chi-Square
Improvement
176.722
149.682
Chi-Square
29.024
29.024
df Significance
17 .0343
17 .0343
Predicted
.00 1.00
0 Ô 1
Observed ôôôôôôôôôôôôôôôôô
.00 0 Ô 40 6 29 6
ôôôôôôôôôôôôôôôôô
1.00 1 ô 12 ô 68 ô
ôôôôôôôôôôôôôôôôô
Overall
Percent Correct
57.97%
85.00%
72.48%
—— — —-------——————— Variables in the Equation — —-
Variable B S.E. Wald df Sig Exp(B)
N83 -3.2284 1.2230 6.9682 1 .0083 -.1554 .0396 1
COMPT -.7449 2.0679 .1298 1 .7187 .0000 .4748
|OWN 1.3900 .4825 8.3003 1 .0040 .1750 4.0149 1
HM083 -.0812 .1728 .2209 1 .6384 .0000 .9220
MCL83 -.0103 .0412 .0620 1 .8034 .0000 .9898
MCR83 -.0140 .0342 .1670 1 .6828 .0000 .9861
MEM83 -.2949 .4686 .3960 1 .5292 .0000 .7446
HMOl .3271 .7932 .1701 1 .6800 .0000 1.3870
MCLl -1.0722 .7743 1.9173 1 .1662 .0000 .3423
MCRl 1.0571 .7365 2.0605 1 .1512 .0171 2.8782
MEMl -.2993 .6326 .2238 1 .6362 .0000 .7414
INT 1 .0813 .1767 .2115 1 .6456 .0000 1.0847
INT 2 -.0067 .0478 .0197 1 .8883 .0000 .9933
INT 3 .0167 .0407 .1688 1 .6812 .0000 1.0169
INT 4 -.1705 .9645 .0313 1 .8597 .0000 .8432
INT 5 1.7021 .9149 3.4610 1 .0628 .0843 5.4852
INT_6 -1.1587 .8753 1.7526 1 .1856 .0000 .3139
Constant 2.9568 2.0715 2.0372 1 .1535
Interactions :
INT 1 COMPT by HM083 INT 4
INT""2 COMPT by MCL83 INT""5
INT""3 COMPT by MCR83 INT""6
COMPT by HMOl
COMPT by MCLl
COMPT by MCRl
193
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
MODEL N2
(Nursing 1987-91)
Number of cases included in the analysis: 149
Dependent Variable.. N2
-2 Log Likelihood 203.58825 * Constant is included in the model.
-2 Log Likelihood
Goodness of Fit
Model Chi-Square
Improvement
171.765
147.069
Chi-Square
31.823
31.823
df Significance
16 .0105
16 .0105
Predicted
.00 1.00
0 Ô 1
Observed ôôôôôôôôôôôôôôôôô
.00 0 Ô 36 5 28 Ô
ôôôôôôôôôôôôôôôôô
1.00 1 ô 19 ô 66 ô
ôôôôôôôôôôôôôôôôô
Overall
Percent Correct
56.25%
77.65%
68.46%
—— —————————————————— Variables in the Equation — — ————— — ————————
Variable B S.E. Wald df Sig R Exp (B)
|N87 -3.1752 1.1715 7.3463 1 .0067 -.1621 .0418
COMPT -.2827 2.0875 .0183 1 .8923 .0000 .7537
OWN -.0733 .4860 .0227 1 .8802 .0000 .9293
IHM087 .0443 .0199 4.9524 1 .0261 .1204 1.0453 1
MCL87 -.0013 .0297 .0019 1 .9656 .0000 .9987
MCR87 .0084 .0352 .0570 1 .8113 .0000 1.0084
MEM87 .4209 .4702 .8012 1 .3707 .0000 1.5233
HM02 .1660 .7225 .0528 1 .8183 .0000 1.1806
MCL2 1.1777 .7649 2.3709 1 .1236 .0427 3.2470
MCR2 .9532 .6562 2.1100 1 .1463 .0232 2.5941
MEM2 -.8014 .7041 1.2955 1 .2550 .0000 .4487
INT 1 .0047 .0372 .0161 1 .8992 .0000 1.0047
INT 2 .0215 .0405 .2814 1 .5958 .0000 1.0217
INT 3 .4290 .9795 .1918 1 .6614 .0000 1.5357
INT 4 -.3515 .9435 .1388 1 .7095 .0000 .7036
|INT 5 -2.4543 .8429 8.4778 1 .0036 -.1784 .0859
Constant .8826 2.0540 .1846 1 .6674
Interactions :
INT_1 COMPT by MCL87
INT_2 COMPT by MCR87
INT_3 COMPT by HM02
* Dropped due to colinearity
INT 4 COMPT by MCL2
INT 5 COMPT by MCR2
INT_6 COMPT by HM087 *
194
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
MODEL N3
(Nursing 1983-91)
Number of cases included in the analysis : 149
Dependent Variable.. N3
-2 Log Likelihood 206.0139 * Constant is included in the model.
-2 Log Likelihood
Goodness of Fit
Model Chi-Square
In^irovement
165.481
165.511
Chi-Square
40.533
40.533
df Significance
17 .0011
17 .0011
Predicted
.00 1.00
0 o 1
Observed ôôôôôôôôôôôôôôôôô
.00 0 Ô 50 Ô 20 Ô
ôôôôôôôôôôôôôôôôô
1.00 1 ô 19 ô 60 ô
ôôôôôôôôôôôôôôôôô
Overall
Percent Correct
71.43%
75.95%
73.83%
Variable
Variables in the Equation
S.E. Wald df Sig Exp (B)
|N83 -5.0696 1.3323 14.4788 .0001 -.2461 .0063
COMPT .7062 2.4235 .0849 .7708 .0000 2.0262
OWN .0532 .4797 .0123 .9116 .0000 1.0547
HM083 .0951 .1888 .2536 .6145 .0000 1.0998
MCL83 -.0602 .0408 2.1799 .1398 -.0296 .9416
MCR83 -.0639 .0343 3.4627 .0628 -.0843 .9381
MEM83 -.1315 .4969 .0700 .7913 .0000 .8768
HM03 2.5335 1.2624 4.0274 .0448 .0992 12.5975
MCL3 2.2957 .8789 6.8229 .0090 .1530 9.9311
MCR3 1.2270 .7707 2.5346 .1114 .0509 3.4108
MEM3 -.3036 .5686 .2852 .5933 .0000 .7381
INT 1 -.0890 .1929 .2128 .6446 .0000 .9149
INT 2 .1144 .0512 4.9923 .0255 .1205 1.1212
INT 3 .0685 .0409 2.8029 .0941 .0624 1.0709
INT 4 -2.0038 1.5891 1.5900 .2073 .0000 .1348
INT 5 -2.0941 1.0346 4.0966 .0430 -.1009 .1232
INT 6 -1.7607 .9251 3.6226 .0570 -.0887 .1719
Constant 2.7144 2.2455 1.4612 .2267
Interactions :
INT 1 COMPT by HM083
IINT~2
INT
COMPT by MCL83
COMPT by MCR83
INT 4
I INT"
COMPT by HM03
COMPT by MCL3
INT_6 COMPT by MCR3
195
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Balancing community benefit and economic survival in nonprofit hospitals: A comparative perspective on charitability and tax-exemption policies in competitive health care markets in Pennsylvania ...
PDF
The impact of state abortion policy on selective abortion ratios in the United States
PDF
Alternate models of women's health care policy in the United States
PDF
The effects of case mix, hospital competition, and managed care penetration on hospital costs and revenues
PDF
Patient life in a nursing home: an experiential study
PDF
The use of transaction cost economics, fuzzy set and systems theories in the development of a three-dimensional pricing model of asset specificity and contracts, and applied to health care
PDF
Managing change, market and value: Reengineering cardiac care in a tertiary care academic medical center
PDF
A study of the effects of familiarity on public opinions about public schools: What they don't know can hurt!
PDF
The organizational effectiveness of public management information systems in Korea: A principal-agent perspective
PDF
Evidence of age-based rationing of health care to the elderly: The effect of age and death on the use of health care services by elderly enrollees in an HMO
PDF
Evaluating differences in patient satisfaction with inpatient hospital care categorized by diagnostic related groups
PDF
Receipt of and perceived difficulty in accessing needed services among HIV-seropositive men and women
PDF
The extent and effectiveness of nursing home regulation in the 50 states
PDF
The effects of position classification on need satisfaction and work attitudes
PDF
Hospital versus home delivery in Lesotho: A comparison of mountain and non-mountain locations
PDF
HIV/AIDS education in the Arabian Gulf state of Bahrain
PDF
Technological advances in long-term care: Barriers to implementation.
PDF
Leadership in restructuring schools: The development of an expanded framework
PDF
Nurse executive leadership and the relationship to organizational commitment among nurses
PDF
Two essays on linear regression models: Solving shortest length least-squares problems and the determination of birth outcome
Asset Metadata
Creator
Smith, Suzanne Lucille
(author)
Core Title
The effect of rate regulation and price competition on labor efficiency in ancillary and nursing service areas of hospitals in California, 1983-1991
Degree
Doctor of Philosophy
Degree Program
Public Administration
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
education, health,health sciences, health care management,OAI-PMH Harvest,Political Science, public administration
Language
English
Contributor
Digitized by ProQuest
(provenance)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c17-530470
Unique identifier
UC11350016
Identifier
9705177.pdf (filename),usctheses-c17-530470 (legacy record id)
Legacy Identifier
9705177.pdf
Dmrecord
530470
Document Type
Dissertation
Rights
Smith, Suzanne Lucille
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA
Tags
education, health
health sciences, health care management