The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society...
-
Upload
retnaning-tyas -
Category
Documents
-
view
253 -
download
0
Transcript of The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society...
![Page 1: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/1.jpg)
www.elsevier.com/locate/aos
Accounting, Organizations and Society 31 (2006) 179–210
The relation between cost-system design, managers�evaluations of the relevance and usefulness of cost data,
and financial performance: an empiricalstudy of US hospitals
Mina J. Pizzini *
School of Management, University of Texas at Dallas, J043 Box 83688, Richardson, TX 75083-0688, United States
Abstract
This study examines the association between cost-system functionality, managers� beliefs about the relevance and
usefulness of cost data, and actual financial performance using a sample of 277 US hospitals. Results indicate that man-
agers� evaluations of the relevance and usefulness of cost data are positively correlated with the extent to which systems
can provide greater cost detail, better classify costs according to behavior, and report cost information more frequently.
However, only the ability to supply cost detail is favorably associated with measures of financial performance, including
operating margin, cash flow, and administrative expense. Interestingly, cost-system design was not associated with oper-
ating expense per admission, suggesting that accounting information had not yet been successfully used to manage clin-
ical costs.
� 2004 Elsevier Ltd. All rights reserved.
Introduction
This study investigates associations between
cost-system design, managers� beliefs about the rel-evance and usefulness of cost data, and actualfinancial performance using a sample of 277 US
hospitals. Accounting literature identifies at least
0361-3682/$ - see front matter � 2004 Elsevier Ltd. All rights reserv
doi:10.1016/j.aos.2004.11.001
* Tel.: +1 972 883 6337; fax: +1 972 883 6822.
E-mail address: [email protected]
four critical attributes of cost-system design: the
level of detail provided, the ability to disaggregate
costs according to behavior, the frequency with
which information is reported, and the extent to
which variances are calculated. The first attribute,the level of detail, refers to the system�s ability to
supply data about cost objects that vary in size
from entire divisions to individual products, com-
ponents, and services. Chenhall and Morris (1986),
Feltham (1977), Kaplan and Norton (1992), and
Karmarkar, Lederer, and Zimmerman (1990)
ed.
![Page 2: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/2.jpg)
1 Statistics can be found at www.cms.hhs.gov/statistics/nhe
and were compiled by Centers for Medicare and Medicaid
Services, Office of the Actuary.
180 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
incorporated level of detail in their characteriza-
tions of cost-system design. The second character-
istic of cost-system design, the ability to
disaggregate costs according to behavior, closely
relates to the first. To supply detail, the systemmust first separate and classify costs according to
behavior. Basic cost classifications explored in
the literature include fixed/variable costs, direct/
indirect costs, and controllable/non-controllable
costs (Feltham & Xie, 1994; Johnson, 1992; Kar-
markar et al., 1990; Khandwalla, 1972). The third
attribute, cost reporting frequency enables manag-
ers to expediently address problems and identifyopportunities for improvement (Hilton, 1979;
Karmarkar et al., 1990; Simons, 1987). The final
cost-system trait, variance analysis, highlights dif-
ferences between budgeted and actual outcomes
and seeks to explain such differences (Karmarkar
et al., 1990; Khandwalla, 1972; Simons, 1987).
More functional cost systems are those that can
provide greater detail, better classify costs accord-ing to behavior, report cost information more fre-
quently, and/or calculate more variances.
The conceptual model linking cost-system de-
sign to performance is typically stated in terms of
a causal chain in which more functional or refined
cost systems produce ‘‘better’’ (i.e., more relevant
and useful) data that enhance managerial decision
making, and thereby lead to improved economicperformance (e.g., Cooper & Kaplan, 1991; John-
son, 1992; Shank & Govindarajan, 1993). Despite
the intuitive appeal of this rather basic premise,
there is little systematic evidence linking cost-sys-
tem design to economic performance. Analytical
models of single-firm settings find that more de-
tailed and frequent cost data are more useful in
decision making (Feltham, 1977; Hilton, 1979);however, evidence in multi-firm settings indicates
that strategic behavior can bring about conditions
in which less informative product-cost data are
optimal (Banker & Potter, 1993; Callahan & Gab-
riel, 1999; Gal-Or, 1987; Gal-Or, 1998; Gal-Or,
1993). Research on institutional theory and infor-
mation overload further suggests that managers
may not effectively use the more detailed, disaggre-gated, and voluminous data produced by highly
functional systems (Covaleski, Dirsmith, & Mich-
elman, 1993; Schick, Gordon, & Haka, 1990).
Highly functional systems also cost more to imple-
ment and administer; consequently, the benefits of
such systems may not exceed their costs (Babad &
Balachandran, 1993; Banker & Potter, 1993; Kar-
markar et al., 1990). Finally, contingency theorysuggests that greater functionality is not a priori
better. Rather, performance improvement is a
function of the alignment between cost-system
functionality and the firm�s operating environment
(see Chenhall, 2003, for a review).
The US hospital industry is an ideal setting for
investigating the performance effects of cost-sys-
tem design. The industry provides the opportunityto sample from a large number of complex organi-
zations that operate under similar circumstances
and offer relatively standardized services. In the
United States, expenditures for hospital care ex-
ceeded $486 billion in 2002 and constituted almost
5% of the gross domestic product (GDP). 1
Healthcare literature overwhelmingly advocates
the use of highly refined cost systems to enablehospital managers to respond to growing pressure
to control costs in this rapidly changing industry
(e.g., Cooper & Suver, 1994; Chua & Degeling,
1991; Hill & Johns, 1994; Young & Pearlman,
1993). Such advocacy is consistent with manage-
ment accounting research (e.g., Chenhall & Mor-
ris, 1986; Gordon & Narayanan, 1984; Hill,
2001; Khandwalla, 1972), which suggests that asenvironmental uncertainty increases, decision
makers seek more information for planning and
control. Yet, in practice, US hospital cost systems
vary greatly in functionality, with many systems
presumably lagging the new managerial informa-
tion needs created by sweeping, industry-wide
changes (Hill & Johns, 1994; Orloff, Littell, Clune,
Klingman, & Preston, 1990; Serb, 1997).Empirical evidence on the performance implica-
tions of cost-system design is primarily limited to
small-sample studies focused upon managers� be-liefs about a single aspect of cost-system design
(activity-based costing); and here, results are mod-
est (e.g., Foster & Swenson, 1997; Ittner, Lanen, &
Larcker, 2002; Krumwiede, 1998). Using data
![Page 3: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/3.jpg)
2 Source: American Hospital Association website, www.
aha.org/aha/resource_center. Excludes federal government hos-
pitals, which constitute only a small proportion of the market
and primarily serve veterans and active members of the
military.3 Source: Centers for Medicare and Medicaid Services
website, www.cms.hhs.gov/medicare.
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 181
from 277 US hospitals, this study extends prior re-
search in several ways. First, rather than simply
identifying whether an organization has adopted
a single costing technique, comprehensive survey
data provide a relatively objective, multidimen-sional characterization of the functions a cost sys-
tem can perform. Second, the ability to link
descriptive data on cost-system design with actual
hospital-level financial data from external sources
permits some of the first direct tests of the relation
between specific cost-system attributes and actual
economic outcomes. Third, the relatively large
sample size and operational homogeneity of hospi-tals enable powerful tests of the research question.
In contrast, prior studies relied predominantly on
small-sample field studies or moderate-size sam-
ples of firms in diverse industries (e.g. Shields,
1995; Swenson, 1995). Finally, this study adds to
the literature on contingency theory by measuring
‘‘fit’’ between control system and organizational
context using both ‘‘selection’’ and ‘‘systems’’ ap-proaches (Selto, Renner, & Young, 1995, p. 679).
Results indicate that managers find cost data to
be more useful and relevant if supplied by systems
that provide greater detail, better classify costs
according to behavior, and provide cost informa-
tion on a more frequent basis. However, actual
financial performance, as characterized by operat-
ing margins, cash flow per bed, and administrativecost control, is significantly and positively associ-
ated only with those systems that provide greater
detail. There is some evidence that improved cost
classification capabilities can enhance cash flow
per bed; however, the performance effects of cost
classification cannot be separated from the effects
of providing detail. Cost-system design was not
associated with expense per admission, which re-flects clinical resource consumption, and hence, is
controlled primarily by physicians, not hospital
administrators. Performance tests that measured
cost-system functionality relative to hospitals with
similar strategies, structures, and environments
were consistent with tests that measured function-
ality relative to all hospitals in the sample. Taken
together, the results suggest that more functionalcost systems can help hospital administrators im-
prove non-clinical aspects of hospital operations,
such as cash management and administrative effi-
ciency; however, highly functional cost systems
have not been used to significantly reduce clinical
expenditures.
The remainder of this paper proceeds as follows.
The following section provides background on theUS hospital industry. Then the paper reviews the
literature on cost-system design and develops the
hypotheses within the context of the hospital indus-
try. The next section describes the research meth-
ods, followed by an analysis of the results of
statistical tests. The paper concludes with a sum-
mary of the results and limitations of the study.
Background on US hospital industry
In the US, the government is heavily involved in
the provision and regulation of healthcare; how-
ever, the US government still plays a much smaller
role in the health sector when compared with the
governments of most other countries. Privately-owned, nonprofit hospitals dominate the US hos-
pital market, accounting for approximately 60%
of all hospitals, while state and local govern-
ment-owned hospitals constitute only 22% of the
market. The remaining 18% of US hospitals are
organized as for-profit investor-owned firms. 2
Hospitals of different ownership type and for-prof-
it status often compete in the same markets. Al-most 55% of the total US healthcare bill is
funded from private sources including private
insurers (33%), individuals (15%), and other pri-
vate sources (6%). The US federal government
pays for roughly 33% of healthcare expenditures
primarily through its health insurance programs
for individuals age 65 or older (Medicare) and cer-
tain disadvantaged individuals (Medicaid). Stateand local governments cover another 13% of
healthcare expenditures. 3 Patient fees fund the
![Page 4: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/4.jpg)
6 Economic theories assume that individuals are rational
and that successful firms make appropriate decisions in
designing control systems. A number of alternative theories
suggest otherwise (see Luft & Shields, 2003, for a review); and,
several have used alternative theories to analyze management
control systems in healthcare (e.g., Abernethy & Chua, 1996;
Abernethy & Vagnoni, 2004; Chua & Degeling, 1991; Jones &
Dewing, 1997). However, systematic empirical evidence that US
182 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
bulk of expenditures in most nonprofit hospitals
and in all for-profit hospitals (Eldenburg, Herma-
lin, Weisbach, & Wosinska, 2004). Nonprofit pri-
vately-owned hospitals also receive significant
donations, while nonprofit government-ownedhospitals obtain substantial government subsidies.
For-profit hospitals receive few donations and, un-
like their nonprofit counterparts, their income is
taxed.
Despite differences in ownership and for-profit
status, US hospitals are strikingly similar in service
provision and organizational structure (Eldenburg
et al., 2004). All hospitals provide essentially thesame basic set of services using standard methods.
Across ownership types, hospitals are governed by
a board that is empowered to direct all hospital
functions (Phelps, 2003). The board, which is gen-
erally self-replicating (members elect their own
successors), represents the interests of various
stakeholders, including donors, community mem-
bers, and shareholders. The board provides overallstrategic direction, but normally hires nonphysi-
cian professional administrators to manage the
hospital. Senior hospital administers parallel their
counterparts in industry.
One rather unique aspect of US hospitals,
regardless of ownership type, is the distinction be-
tween hospital personnel and the medical staff of
physicians (Phelps, 2003). Hospitals employ bothadministrative professionals and nonphysician
clinical professionals (e.g., nurses and medical
technicians), but they do not generally employ
the physicians who are members of the medical
staff. Physicians gain entry to the medical staff
through a credentialing process controlled by the
existing medical staff. 4 The hospital has no expli-
cit contract with the medical staff, yet it is depen-dent upon staff physicians to attract and care for
hospital patients. Staff physicians direct patient
care within hospitals, and in doing so, control
the use of hospital employees and facilities. 5
4 Physicians apply to the hospital board for admission to the
medical and the board generally delegates responsibility for this
decision to existing medical staff (Phelps, 2003, p. 254).5 By law, most care provided to patients must be performed
by a licensed physician or under the direction of a licensed
physician (Phelps, 2003, p. 254).
For example, physicians admit patients, prescribe
all medications that hospital nurses administer,
and specify which tests hospital technicians
perform.
Given the prominence of the private sector inUS healthcare, researchers commonly use eco-
nomics to analyze and understand the US health-
care sector (for example, see Phelps, 2003,
Chapter 1). In fact, research in healthcare eco-
nomics provides strong empirical support for
the application of economic theory to the US
healthcare sector. For example, several studies
have found that physicians and hospital adminis-trators in the US respond to economic incentives
in a rational manner and there is growing evi-
dence that agency theory explains the choice of
compensation contracts in healthcare organiza-
tions (e.g., Brickley & Van Horne, 2002; DeBrock
& Arnould, 1992; Gaynor & Gertler, 1995; Gay-
nor & Pauly, 1990; Ittner, Larcker, & Pizzini,
2004; Lambert & Larcker, 1995; Lee, 1990).Empirical evidence also suggests that economic
theory underlies hospitals� strategic choices such
as location (Norton & Staiger, 1994; Wedig, Has-
san, & Sloan, 1989), investment (Krishnan, 2001),
cost allocations (Eldenburg & Kallapur, 1997),
operating strategy, and tax-status (Picone, Chou,
& Sloan, 2002). Following prior research on
healthcare institutions, this paper draws heavilyupon economic theory to develop the research
hypotheses. 6
hospitals conform to political, psychological and other non-
economic theories of behavior is quite limited. Conversely, a
significant healthcare economics literature provides strong
empirical evidence that physician behavior and managerial
decisions in US hospitals conform to economic theory (e.g.,
Brickley & Van Horne, 2002; Eldenburg et al., 2004; Gaynor &
Gertler, 1995; Gaynor & Pauly, 1990; Lambert & Larcker,
1995; Wedig et al., 1989).
![Page 5: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/5.jpg)
7 DRG categories, which reflect the nature and intensity of
the hospital services, can essentially be thought of as the
hospital�s ‘‘products.’’ Hence, DRG-level costing enables hos-
pital managers to assess the profitability of its ‘‘products.’’8 However, a very complex method for classifying costs is
not necessarily a prerequisite for supplying useful detailed
information (Bromwich & Hong, 1999; Datar & Gupta, 1994;
Noreen, 1991). Even if cost classification is not highly accurate,
detailed cost data can still be useful (Merchant & Shields, 1993).
Hill (2001) contends that hospitals with procedure-level cost
systems will have better information for decision making than
those without such systems, even if the procedure-level system
uses a relatively unsophisticated method to assign costs to
procedures (e.g., the ratio of costs to charges).
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 183
Hypothesis development
Proposed benefits of greater functionality in hospital
cost system design
While there is little empirical evidence support-
ing associations between cost-system design and
performance, the extant literature suggests that
cost systems can be used to enhance both financial
and operational performance. Each individual ele-
ment of cost-system design potentially improves
the relevance and usefulness of cost data. More-
over, in the hospital setting, the cost system as awhole can potentially mitigate agency problems.
Proposed benefits of individual cost system design
elements
Level of detail. Management accounting literature
generally contends that an important feature of a
cost system is its ability to provide sufficient detail
and flexibility to allow costs to be analyzed for dif-ferent purposes (e.g., Karmarkar et al., 1990;
Shank & Govindarajan, 1993). Chenhall and Mor-
ris (1986) proposed that systems that can isolate
the effects of specific events on different functions
are of greater use to managers in uncertain envi-
ronments. Feltham�s (1977) analytical model sup-
ported these assertions, finding that the expected
payoff from decisions based upon more detailedinformation is generally greater than that from
decisions based on more aggregated information.
In hospitals, managers face various decisions
regarding the services (procedures, tests, diagno-
ses, etc.) provided, the customers (patients) receiv-
ing these services, the individuals (physicians,
nurses, therapists, etc.) providing these services,
and the groups (private insurance companies, gov-ernment agencies, individuals, etc.) that pay for
the services. Unless the system can identify and
aggregate costs for these cost objects (i.e., provide
sufficient detail) with reasonable accuracy, manag-
ers will not be able to make well-informed deci-
sions on related issues (Comerford & Abernethy,
1999; Covaleski et al., 1993; Evans, Hwang, &
Nagarajan, 1995). The ability to specify the costsof individual diagnostic-related groups (DRG) is
particularly relevant because Medicare and many
private insurers reimburse hospitals a predeter-
mined fixed amount based upon a patient�s DRG
(e.g., Chua & Degeling, 1991; Preston, 1992). 7
Lawrence (1990) hypothesized that actual financial
and operational performance would be higher for
hospitals that used systems that supplied greaterdetail; however, her empirical tests did not support
this hypothesis.
Classify costs according to behavior. The ability to
disaggregate costs and classify them according to
their behavior directly supports the ability to pro-
vide useful detailed cost information. For example,
to aggregate costs by procedure, direct costs mustbe traced to the procedure, and indirect fixed and
variable costs must be allocated to the procedure.
This requires the system to classify costs as di-
rect/indirect and fixed/variable. If this cost classifi-
cation is arbitrary, than the detailed cost data will
not be useful; however, if the system can meaning-
fully classify costs according to behavior, detailed
cost data should be useful. In fact, severalresearchers contend that correctly identifying cost
behavior is the first step in supplying accurate cost
information at all levels of detail (Cooper & Kap-
lan, 1991, Chapter 5; McGown, 1998; Shank &
Govindarajan, 1993, Chapter 10; Swenson,
1995). 8 Classifying costs according to behavior
is particularly important in the hospital industry,
which is characterized by large amounts of facil-ity-related support and joint costs (e.g., buildings
and sophisticated equipment shared by many
users) that cannot be directly and easily traced to
individual services. In Coombs�s (1987) field study
of two Swedish hospitals, both administrators and
![Page 6: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/6.jpg)
184 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
physician clinic managers expressed strong interest
in systems that could better associate clinical activ-
ities with resource consumption.
Another important cost categorization is the
distinction between controllable and noncontrolla-ble costs because it aids principals in performance
evaluation (Feltham & Xie, 1994). In US hospitals,
the ability to identify functions that fall under a
manager�s control is particularly crucial because
hospital managers typically have less control over
basic business functions such as pricing, services
offered, and service delivery than their counter-
parts in other sectors. For example, if cost datashowed that emergency medicine was not profit-
able, the hospital could not simply turn people
away, raise prices, or shut down the emergency
room.
Frequency of cost reports. Cost-reporting fre-
quency enables managers to expediently address
problems and identify opportunities for improve-ment. Chenhall and Morris (1986) measured the
frequency of cost reports and contended that more
frequent reporting provides managers with feed-
back on decisions and information on recent
events that they can use to guide future courses
of action. Chenhall and Morris found that more
frequent reporting was believed to be particularly
useful to managers who operate in highly uncer-tain environments, such as hospitals. In response
to increased pressure to control costs, hospital
administrators in Coombs� (1987) field study
sought to increase the frequency with which they
received control information. Eldenburg (1994)
empirically examined whether more frequent
reporting of cost information to hospital physi-
cians could induce cost-conscious behavior. How-ever, contrary to her hypothesis, higher reporting
frequency was not significantly associated with
lower costs.
More frequent reporting may also indicate that
information is provided on a more timely basis.
For example, if cost information is reported
monthly rather than quarterly, managers can ad-
dress concerns that arise between quarters, ratherthan wait until the end of the quarter. Hilton
(1979) modeled the value of an information system
in a cost-volume-profit decision setting and found
that the more timely the information, the greater
the value of an information system. Karmarkar
et al. (1990) used timeliness to characterize cost-
system design, and they considered timeliness to
be indicative of the ‘‘elaborateness’’ of a costsystem.
Variance analysis. Proponents of variance analysis
contend that it aids in managerial decision making
by identifying corrective managerial actions (e.g.,
Johnson & Kaplan, 1987; Shank & Govindarajan,
1993). Budgetary control, which forms that basis
of variance analysis, may be appropriate for thehospital setting because uncertainty in both task
technology (Ouchi, 1979) and output (Cooper,
Hayes, & Wolf, 1981) renders administrative mon-
itoring difficult. Moreover, the prevalence of fixed-
price contracting (i.e., fixed reimbursements based
upon a DRG, patient-day, or procedure) should
make variance analysis very well suited for hospi-
tals because such contracts force hospitals to ac-cept the risk of unexpected costs and utilizations
(Abernethy & Stoelwinder, 1996; Lawrence,
1990; Preston, 1992). By specifying cost targets
and measuring performance in relation to these
targets, managers can use variance analysis to allo-
cate resources to clinical units and measure their
performance. In particular, instead of directly
monitoring physicians, hospital managers caninvestigate variances among amounts charged by
physicians to DRG-defined groups as compared
to hospital or industry standards (Covaleski
et al., 1993).
Potential benefits of a cost system as a whole:
reduction in agency conflicts
Conflicts between hospital board and hospital man-
agers. Most hospitals are nonprofit entities. The
nonprofit organizational form eliminates conflicts
between donors and residual claimants (Fama &
Jensen, 1983a); however, it does not eliminate
agency problems between donors and other inter-
nal agents, namely managers (Fama & Jensen,
1983b). Unlike for-profit entities, donors cannot
align managers� interests with their own by offeringmanagers an equity stake. Thus, donors rely on ac-
tive, independent boards to monitor nonprofit
managers and ensure that managers take actions
![Page 7: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/7.jpg)
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 185
consistent with the interests of donors and external
stakeholders. Consistent with the growing empha-
sis on cost containment in the healthcare sector,
board members are increasingly using cost, profit,
and operational data to monitor managerial per-formance (Brickley & Van Horne, 2002; Molinari,
Hendryx, & Goodstein, 1997). More functional
systems can better associate costs with a manager�sactions and decisions, and therefore, produce mea-
sures that contain more information about a man-
ager�s effort.
Conflicts between management and healthcare pro-
viders. Physicians play a critical role in determin-
ing hospital resource use (Pauly & Redisch, 1973)
and, consequently, hospital financial performance
(Feinglass, Martin, & Sen, 1991). 9 Yet, hospital
managers typically have no explicit control over
physicians because most US physicians are not
employed by the hospitals at which they practice.
Moreover, physician and hospital incentives arenot well aligned because hospital physicians are
usually compensated based upon the volume of
services provided, while hospitals increasingly face
fixed-price contracts (Eldenburg, 1994; Pauly &
Redisch, 1973). 10 Cost data can potentially be
used instead of an explicit contract to control phy-
sician behavior (Covaleski et al., 1993). Specifi-
cally, Eldenburg (1994) and Evans et al. (1995)found that presenting physicians with data on their
own costs, along with comparative cost and oper-
ational information, provides some incentive for
physicians to modify their behavior. However, to
use cost data to mitigate physician-hospital agency
9 Physician control over admissions and in-hospital treat-
ment led Pauly and Redisch (1973) to characterize hospitals as
physician cooperatives designed to maximize physician�s clinicalefficiency and income. Chilingerian and Sherman (1990)
estimate that physicians directly influence up to 80% of all
hospital expenditures.10 Medicare and a growing number of private payers
reimburse hospitals a fixed amount based upon the diagnos-
tic-related-group (DRG) into which an illness can be catego-
rized, regardless of how much it costs the hospital to care for
the illness. Conversely, Medicare and most other payers
reimburse most hospital physicians on a fee-for-service basis,
i.e., based upon the amount of service provided.
problems, hospitals must have a cost system that
provides appropriate data.
Potential drawbacks of greater cost-system
functionality
The aforementioned benefits associated with
greater functionality implicitly assume that more
functional cost systems produce more relevant cost
data, which managers use to make performance-
enhancing decisions. Yet, research has not estab-
lished whether managers find data produced by
highly functional cost systems more relevant anduseful. Nor has extant research convincingly dem-
onstrated that more useful cost data lead to deci-
sions that positively affect a firm�s objective
function. Furthermore, even if more functional
cost systems lead to better managerial decisions,
the benefits of those decisions may not offset costs
associated with a more functional cost system.
Each of these concerns is discussed below.
Limited evidence that greater functionality leads
to ‘‘better’’ cost data
While many contend that more functional cost
systems provide ‘‘better’’ cost data (i.e., more rele-
vant and useful), this is difficult to empirically sub-
stantiate because the ‘‘true’’ costs of products and
services are unknown (Christensen & Demski,1997; Dopuch, 1993). Evidence regarding the qual-
ity of data produced by more functional systems is
limited to analytical studies of overhead alloca-
tions, and here, results are mixed (Banker & Pot-
ter, 1993; Bromwich & Hong, 1999; Christensen
& Demski, 1997; Datar & Gupta, 1994; Noreen,
1991). 11 Institutional theory suggests that hospi-
tals may adopt refined cost systems simply to con-form to societal expectations of acceptable
practice, and thereby gain external credibility and
validation (Covaleski et al., 1993). In such cases,
11 Banker and Potter (1993) and Datar and Gupta (1994)
demonstrate that greater complexity in allocation schemes
introduces classification, measurement, and other errors that
can actually reduce data accuracy. Others show that greater
functionality leads to greater accuracy only if fairly restrictive
conditions are met (Bromwich & Hong, 1999; Christensen &
Demski, 1997; Noreen, 1991).
![Page 8: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/8.jpg)
12 In a 1992 survey on hospital cost systems, the majority of
respondents who did not use procedure-level cost systems
indicated that they did not have such systems due to resource
constraints, expense, or the concern that costs would outweigh
the benefits (Hill & Johns, 1994). Similarly, Counte and
Glandon (1988) found that 53% of respondents had not
upgraded their systems due to resource constraints, while 38%
feared the costs would not exceed the benefits.
186 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
hospital managers are unlikely to find the informa-
tion produced by more functional cost systems to
be useful.
Limitations on the usefulness of ‘‘better’’ cost data
Even if more functional systems produce ‘‘bet-
ter’’ cost data, such data may not necessarily lead
to improved firm performance. In fact, more infor-
mative cost data can actually hinder performance.
This has been demonstrated in multi-firm settings
characterized by strategic behavior (Gal-Or,
1987; Gal-Or, 1998; Gal-Or, 1993). Banker and
Potter (1993) specified demand conditions underwhich it was optimal for firms to use a relatively
simple system. Callahan and Gabriel (1999) found
that firms competing on the basis of product differ-
entiation (characterized as a Bertrand duopoly)
did not benefit from increased product-cost accu-
racy. The risk of information overload also sug-
gests that managers may be constrained in their
ability to effectively use data produced by morefunctional cost systems. Information that is too
detailed, disaggregated, and/or voluminous poten-
tially reduces decision-making effectiveness (see
Schick et al., 1990, for a review).
Finally, some researchers contend that variance
data specifically is not useful for managerial deci-
sion making. They argue that interdependence be-
tween variances may produce measurement errors,that variance analysis lacks information on the
causes of cost overruns, and that variances provide
incentives to pursue short-term cost reduction over
continuous improvement (e.g., Bastable & Bao,
1988; Cheatham & Cheatham, 1996; Cooper &
Kaplan, 1992; Mak & Roush, 1996). Others have
used political theories of organization to criticize
the application of budgeting, which underlies var-iance analysis, to the hospital sector. For example,
Abernethy and Vagnoni (2004), Covaleski and
Dirsmith (1986), and Kurunmaki (1999) suggest
that budgeting potentially serves to establish, dis-
tribute, and maintain power within hospitals,
and therefore, undermines the role that variance
analysis plays in performance-enhancing decisions.
The costs of greater functionality
The potential benefits of a more highly func-
tional cost system may not exceed the costs of that
system. Implementing a new system often entails
consulting, training, and software expenses. Com-
plex systems typically incur higher data collection,
processing, and management costs (Babad & Bal-
achandran, 1993; Banker & Potter, 1993). Surveysin the hospital industry suggest that the cost of
such systems is a major impediment to their use
(Counte & Glandon, 1988; Hill & Johns,
1994). 12 While the extant literature acknowledges
the costs of highly functional systems, it largely
ignores such costs in investigating the performance
effects of functionality. Analytical models gener-
ally do not address such costs (Banker & Potter,1993; Feltham, 1977; Hilton, 1979), and few
empirical studies use performance measures that
reflect the actual cost of administering the system
(Ittner et al., 2002; Lawrence, 1990).
In summary, there are several potential limita-
tions to cost-system functionality. However,
underlying economic theory suggests that the po-
tential benefits of more functional systems canovercome these limitations. In particular, the pro-
posed benefits of individual cost-system design ele-
ments, the potential to reduce agency problems,
and the need for more information created by
the rapidly changing hospital industry suggest
the following hypotheses:
H1a. Cost-system functionality is positively associ-
ated with managers� beliefs about the rele-
vance and usefulness of cost data in hospitals.
H1b. Cost-system functionality is positively associ-
ated with actual financial performance in
hospitals.
Contingency theory and cost-system design
The previous discussion posits that the absolute
level at which a cost-system functions is directly
![Page 9: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/9.jpg)
13 Hill (2001) found that during the 1980s, for-profit
hospitals were less likely to adopt more functional cost systems
than their nonprofit counterparts, presumably because for-
profit hospitals could better avoid the Medicare revenue
constraints of that period. However, she speculated that during
the 1990s, managed care would place increased pressure on for-
profits to control costs.14 Hospital districts, typically found in rural areas, are
municipal entities composed of elected officials. Hospital
district commissioners have the power to levy taxes to subsidize
operations.15 Managed-care organizations (MCOs) often impose strict
cost-control measures on hospitals and negotiate lower prices
than other insurers.
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 187
correlated to a firm�s performance. However, con-
tingency theory contends that a firm�s strategy,
organizational structure, and environment dictate
its choice of control system (see Chenhall, 2003,
for a review). Any associated benefits or draw-backs are a function of the degree of alignment be-
tween the design of a firm�s cost system and the
specific set of circumstances the firm faces. Bene-
fits or drawbacks are not solely attributable to
the absolute level of cost-system functionality.
Consistent with contingency theory, hospital
cost-system design has been found to vary system-
atically with internal, organizational factors andexternal, environmental factors (e.g., Counte &
Glandon, 1988; Hill, 2001; Hill & Johns, 1994;
Lawrence, 1990). For purposes of discussion, I
group determinants of hospital cost-system design
into three categories: strategic, structural, and
environmental.
Although there is little empirical research link-
ing organizational strategy to control-system de-sign in hospitals, research in other industries
generally contends that strategy influences con-
trol-system design (see Langfield-Smith, 1997, for
a review). Using Porter�s framework (1980,
1985), strategy can be measured along two dimen-
sions: product differentiation and low-cost produc-
tion. Hospitals pursuing a differentiation strategy
are expected to focus resources on clinical care tothe detriment of administrative systems, such as
the cost system. Conversely, hospitals emphasizing
cost control will have more functional cost systems
because managers will require more information
for monitoring costs.
Structural determinants include case mix, teach-
ing affiliation, size, and whether a hospital is a
member of a multi-hospital system. Prior studieshave found positive associations between these
determinants and cost-system functionality
(Counte & Glandon, 1988; Hill & Johns, 1994;
Lawrence, 1990). Case mix measures the severity
of cases treated. Teaching hospitals, which typi-
cally treat more severe cases, are more complex
organizationally due to their research and educa-
tional responsibilities. As complexity increases,so does the need for cost information (Karmarkar
et al., 1990). Larger hospitals and system hospitals
will benefit more from functional cost systems be-
cause they can potentially spread the fixed costs of
system development over more beds (Hill, 2001).
Other important structural variables include for-
profit status and hospital district membership.
For-profit hospitals, which have the explicit goalof increasing owners� wealth, are expected to
implement more elaborate cost systems to aid in
this endeavor. 13 Hospitals that are organized by
hospital districts can potentially subsidize opera-
tions with funds obtained from local taxes. Thus,
district hospitals may have less incentive to control
costs and, consequently, less need for highly func-
tional cost systems. 14
A firm�s external environment also has been
found to influence cost-system design in the hos-
pital and other industries (Hill, 2001; Khand-
walla, 1972). Hospitals that operate in markets
with strong competition and/or significant pene-
tration from managed-care organizations (MCOs)
face greater external pressure to control costs and
therefore require more extensive and detailed costinformation (Hill, 2001; Lawrence, 1990). 15 Sim-
ilarly, hospitals that treat a large proportion of
patients covered by Medicare or managed-care
plans will face greater external pressure to con-
trol costs, because these payers often impose
the risk of cost overruns upon the hospital
through the use of fixed-price reimbursement
contracts. Such hospitals will likely need morefunctional cost systems to control costs. Taken
together, literature on contingency theory and
hospital cost-system design suggests the following
hypotheses:
![Page 10: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/10.jpg)
Table 1
Descriptive statistics comparison of sample means to national
averages
National averages Sample means
Beds 185 272
Case mix 1.27 1.40
For profit 18.3% 7.2%
System members 39.5% 41.2%
Teaching hospitals 22.7% 36.8%
Medicare 51.6% 52.7%
188 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
H2a. Managers’ beliefs about the relevance and
usefulness of cost data are positively associ-
ated with the degree to which cost-system
functionality is aligned with the hospital’s
strategy, structure, and external environment.
H2b. Actual financial performance is positively
associated with the degree to which cost-sys-
tem functionality is aligned with the hospital’s
strategy, structure, and external environment.
Metropolitan 62.8% 67.7%17 The sample may be skewed towards larger hospitals and
teaching hospitals because such hospitals have greater active
representation in HFMA, the industry group that helped to
distribute the survey.18 Calculating the functionality measures from multiple
items should reduce the error associated with any single-item
Research method
Data
The hypotheses are tested using survey data col-
lected in conjunction with fellow researchers at the
University of Pennsylvania�s Wharton School of
Business and the Hospital Financial ManagementAssociation (HFMA). The initial questionnaire
was based upon the extant literature and inter-
views with financial managers from five hospitals.
The survey was then reviewed by personnel from
HFMA, and modified as required. We sent the
survey to financial managers at 1703 hospitals.
Personnel at HFMA distributed 1042 surveys to
members in September 1997, and the businessschool researchers sent another 661 surveys to hos-
pitals that were not included in the HFMA mail-
ing. 16 After two months, surveys were mailed a
second time to hospitals that did not respond to
the initial mailing.
We received 385 completed surveys from Octo-
ber 1997 through January 1998. Sixteen surveys
were rendered undeliverable and returned uno-pened because the contact could not be located
or the hospital had closed. Consequently, person-
nel at 1687 of the 1703 hospitals actually received
the surveys. Of these, 385 responded, yielding a
22.8% response rate. Eleven specialty hospitals
were eliminated because their activities were not
comparable to acute-care facilities. Of the remain-
ing 374 responses, 313 surveys provided informa-
16 We generally addressed the surveys to the Chief Financial
Officer, the Controller, or the Chief Executive Officer. However,
lower-ranking individuals in the accounting or finance func-
tions often responded to the surveys.
tion sufficient to characterize cost-system
functionality. External data on financial perfor-
mance and control variables were available for
277 of the 313 hospitals. Table 1 presents descrip-
tive statistics for the 277-hospital sample and all
US hospitals. The sample is predominantly com-
posed of metropolitan hospitals that are larger
than the national average and treat relatively morecomplex cases. Compared to all US hospitals, the
sample contains more teaching hospitals (36.8%
vs. 24.5%) and fewer for-profit hospitals (7.2%
vs. 15.4%). 17
Cost-system functionality measures
The survey asked 20 questions relating to cost-system attributes. The responses were standardized
to a common scale, and factor analysis was used to
reduce the dimensionality of the questions. 18 Four
factors emerged with eigenvalues in excess of one,
with the factor solution retaining 65% of the total
variation in the data (Table 2). Following ortho-
gonal rotation, the four factors are: DETAIL,
CLASSIFY, FREQUENT, and VARIANCE.
measure. Some of the hospitals failed to respond to all survey
questions. Rather than exclude these hospitals from the
analysis, missing values were imputed prior to the factor
analysis. Following Beal and Little (1975), the missing data
were imputed using conditional means obtained from a stepwise
ordinary least squares regression.
![Page 11: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/11.jpg)
Table 2
Factor structure correlations following orthogonal rotationa,b,c
DETAIL CLASSIFY FREQUENT VARIANCE Questionsd
Customize 0.55 . . . The cost accounting system can easily customize
reports to the specification of users
By payer 0.86 . . . To what extent does system provide data that
allow you to analyze costs by payer?
By contract 0.84 . . . To what extent does system provide data that
allow you to analyze costs by contract?
By physician 0.82 . . . To what extent does system provide data that
allow you to analyze costs by physician?
By procedure 0.74 . . . To what extent does system provide data that
allow you to analyze costs by procedure?
Per diem 0.75 . . . To what extent does system provide data that
allow you to analyze costs on a per diem basis?
By patient 0.85 . . . To what extent does system provide data that
allow you to analyze costs per patient?
Direct/indirect . 0.72 . . Does your cost system have a formalized method
of distinguishing direct and indirect costs?
Controllable . 0.77 . . Does your cost system have a formalized method
of distinguishing fixed and variable costs?
Fixed/variable . 0.55 . . Does your cost system have a formalized method
of distinguishing control/non-controllable costs?
Track F/V . 0.64 . . How sophisticated is method used to tracked fixed
and variable costs?
Use RCC . 0.63 . . To what extent is RCC used to estimate the costs
of individual services and procedures?
Frequent1 . . 0.83 . How often does the cost accounting system report
information to senior managers?
Frequent2 . . 0.85 . How often does the cost accounting system report
information to middle managers?
Frequent3 . . 0.79 . How often does the cost accounting system report
information to clinical managers?
Frequent4 . . 0.76 . How often does the cost accounting system report
information to nursing staff?
Frequent5 . . 0.77 . How often does the cost accounting system report
information to medical staff?
Efficiency variance . . . 0.68 Does your system calculate efficiency variances?
Mix variance . . . 0.58 Does your system calculate case mix variances?
Price variance . . . 0.76 Does your system calculate price variances?
Cronbach Alpha 0.93 0.83 0.88 0.53
a ‘‘.’’ Indicates correlation of less than 0.50.b Only factors with eigenvalues in excess of 1.0 were included. The lowest eigenvalue of included factors was 1.24 for the VARI-
ANCE construct.c DETAIL––the level of detail provided by the system. CLASSIFY––the system�s ability to classify costs according to behavior.
FREQUENT––the frequency with which cost information is disseminated throughout the organization. VARIANCE––the type and
number of variances calculated.d Questions were recoded, if necessary, such that higher values indicate greater functionality. For example, Use RCC was recoded
so that higher values indicate less use of RCC, and thus, greater functionality. All questions and response scales are included in the
Appendix.
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 189
DETAIL encompasses the extent to which the sys-
tem can accumulate costs at various levels of detail
(e.g., per patient, per diem, etc.). CLASSIFY mea-
sures the degree to which the system can classify
costs according to their behavior and the method-
ology used to classify costs. FREQUENT captures
the frequency with which cost-system reports are
distributed to various administrative and clinical
![Page 12: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/12.jpg)
Table 3
Comparison of mean values of functionality constructsa for hospitals that received JCAHO accreditation and those that did not
Not JCAHO, accredited (n = 23) JCAHO, accredited (n = 270) t-Statistica,b
DETAIL –0.830 0.080 31.23***
CLASSIFY –0.498 0.046 11.93***
FREQUENT –0.206 0.030 2.03*
VARIANCE –0.262 0.041 3.51**
*p < 0.10 (2-sided), **p < 0.05 (2-sided), ***p < 0.01 (2-sided).a Functionality constructs are calculated by taking the means of standardized responses to questions that loaded greater than 0.50
on a single factor from the factor analysis (Table 2). The actual responses were standardized to a mean of 0 and standard deviation of
1. DETAIL––the level of detail provided by the system. CLASSIFY––the system�s ability to classify costs according to behavior.
FREQUENT––the frequency with which cost information is disseminated throughout the organization. VARIANCE––the type and
number of variances calculated.b Two-sample t-statistic for test of hypothesis that values of functionality constructs for accredited hospitals are greater than those
of non-accredited hospitals. Results are similar if control variables for hospital structure and market conditions are included by using a
logistical regression.
190 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
personnel and proxies for the timeliness with
which the system can report on recent events and
provide feedback (Chenhall & Morris, 1986;
Eldenburg, 1994; Karmarkar et al., 1990). VARI-
ANCE measures the number of variances the sys-
tem calculates. 19
The questions were designed to be as objective
as possible, although we could not completelyignore subjective assessments. The goal was to
understand what kind of analysis and functions
the system could actually perform.
Each factor represents a different dimension of
cost-system functionality, which is measured using
the average standardized responses to questions
loading greater than 0.50 on a single construct
(see the Appendix for exact questions and theircoding). Cronbach�s coefficient alphas range from
0.53 to 0.93 (Table 2) and thus indicate that the re-
sponses used to calculate each functionality con-
struct are internally consistent. To further
establish that the four cost-system design con-
structs actually capture a system�s functional capa-bility, they are correlated with measures that proxy
for cost-system functionality. The first proxy for
19 The questions that load on the VARIANCE factor have
dichotomous (0, 1) responses. Technically, dichotomous
responses should not be included in a factor analysis. If
variance questions are removed, the factor solution gives three
factors with similar loadings to those reported when the
variance questions are included.
cost-system functionality is accreditation by the
Joint Commission on the Accreditation of Health-
care Organizations (JCAHO). The JCAHO is an
independent agency that evaluates the operational
efficiency of US hospitals. 20 Some of the specific
areas the JCAHO examines include organizational
improvement initiatives, leadership initiatives,
management of information, and management ofhuman resources. 21 Thus, it is expected that hos-
pitals with systems that meet accreditation stan-
dards also have more functional cost systems. As
shown in Table 3, the standardized values of all
four functionality measures are significantly greater
for hospitals that have received JCAHO accredi-
tation than those that have not. The difference is
especially significant (p < 0.001, one-sided) for theDETAIL and CLASSIFY constructs.
The remaining proxies for cost-system function-
ality are based upon individual survey questions
(see the Appendix for exact questions and coding).
One proxy is the extent to which a hospital�s costsystem differs from that used to comply with Medi-
care reporting requirements (denoted differs from
Medicare). Systems designed to comply with min-imum Medicare cost-reporting requirements pro-
20 Accreditation reports are available on the JCAHO�swebsite: www.JCAHO.org.perfrep.
21 While the JCAHO is responsible for attesting to the
quality of care, most of the measures address operational
efficiency. This has prompted many hospitals to forgo the
voluntary accreditation process (Burda, 2002).
![Page 13: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/13.jpg)
Table 4
Pearson correlations of functionality constructs with proxies for functionalitya,b
Differs from Medicare (n = 300) Accuracy (n = 304) Allocation bases (n = 304) Timely (n = 304)
DETAIL 0.42 0.42 0.54
CLASSIFY 0.54 0.35 0.61
FREQUENT 0.31
VARIANCE 0.28
a All correlations are statistically significant at the 0.001 level (two-sided).b Proxies for cost-system functionality were computed from the following questions: Differs from Medicare: ‘‘To what extent is you
organization�s cost accounting system different from the Medicare cost reporting system? (1 = same, 7 = completely different).
Accuracy: ‘‘The cost accounting system provides accurate data.’’ (1 = strongly agree, 7 = strongly disagree). Allocation bases: ‘‘In
allocating costs in revenue departments to a specific procedure, how many separate bases does your hospital use?’’ (N/A, 1, 2–3, 4–6,
>6). Timely: ‘‘Users receive information from the cost accounting function in a timely manner.’’ (1 = strongly agree, 7 = strongly
disagree). Functionality constructs are described in notes to Table 3.
23 Note that only FREQUENT is correlated with timely
because I do not expect the other functionality constructs
(DETAIL, for e.g.) to contribute to the timeliness with which
data are reported. Conversely, FREQUENT is not correlated
with the following functionality proxies, differs from Medicare,
accuracy, and allocation bases because it is not expected that
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 191
vide very little information that is relevant for
managerial decision making (Evans, Hwang, &
Nagarajan, 1997; Meeting & Harvey, 1998; Tsel-
epis, 1989; West & Balas, 1996). 22 Accordingly,
many hospitals have either augmented the Medi-
care cost-reporting system or developed a separate
system to manage costs. Thus, the more a hospi-tal�s cost system differs from the Medicare report-
ing system, the more functional it is expected to be.
As shown in Table 4, the DETAIL, CLASSIFY,
and VARIANCE constructs are all significantly
(p < 0.001, two-sided) and positively correlated
with the variable differs from Medicare. In particu-
lar, the DETAIL and CLASSIFY correlations are
0.42 and 0.54 respectively. The next proxy vari-able, accuracy, is calculated from survey partici-
pants� ratings of the accuracy of cost-system
data. Both DETAIL and CLASSIFY are posi-
tively (r = 0.42 and 0.35, respectively) and signifi-
cantly (p < 0.001, two-sided) correlated with
accuracy (Table 4). This suggests that systems that
can provide more detailed cost information are not
simply assigning costs arbitrarily to various costobjects. Furthermore, consistent with the asser-
tions of activity-based costing proponents, the
managers reported that systems that could better
classify costs according to behavior were more
accurate (Cooper & Kaplan, 1991, Chapter 5).
22 Medicare reimbursements received from the US Govern-
ment average close to 50% of a hospital�s total revenues. To
obtain these reimbursements, hospitals must file a Medicare
Cost Report.
The proxy variable, allocation bases, measures
the number of separate bases used to allocate costs
from revenue-producing departments to specific
procedures. A larger number of allocation bases
suggests greater homogeneity within individual
cost pools and a stronger association between
activities that give rise to the costs in the poolsand the activities used to allocate these costs. It
is expected that cost systems that can provide high
levels of detail and meaningfully classify costs will
use more allocation bases (Cooper & Kaplan,
1991, Chapter 5). Consistent with this premise,
allocation bases is positively and significantly
(p < 0.001, two-sided) correlated with DETAIL
(r = 0.54) and CLASSIFY (r = 0.61). Finally, thesurvey asks participants to rate the timeliness with
which the cost system provides data. This question
is reflected in the variable, timely, which is posi-
tively and significantly correlated with FRE-
QUENT (r = 0.31, p < 0.001). 23
more frequent reporting of cost data would contribute to the
extent to which a system differs from Medicare, system
accuracy, or the number of allocation bases used. Similarly,
the extent to which variances are calculated is not expected to
influence accuracy, nor is it expected to relate to the number of
allocation bases.
![Page 14: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/14.jpg)
192 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
Performance measures
Managers� beliefs about cost system data and
actual performance help determine whether more
functional cost systems provide more relevant costdata, which managers can potentially use to make
performance-enhancing decisions.
Managers’ beliefs about the relevance and
usefulness of cost data
The relevance of cost data is measured by com-
paring managers� beliefs about the need for cost
information to the information actually suppliedby the cost system. The survey asks hospital man-
agers to rate the importance of cost information in
performing ten managerial activities, including
performance evaluation, contract negotiation,
budgeting, and pricing (see the Appendix for ques-
tions and coding). The survey also requests that
managers specify the extent to which the existent
cost system provides information that can be usedto perform these 10 managerial activities. I take an
average of the differences between managers� rat-ings of: (1) the importance of cost information in
performing an activity and (2) the extent to which
the system actually provides information for per-
forming the activity. The smaller the difference,
the closer the system is to supplying all of a man-
ager�s information needs. To calculate a measureof relevance (denoted relevant), the differences
are subtracted from the maximum difference so
that the larger the value of relevant, the better
the system meets mangers� needs. The usefulness
of cost-system information is measured with a sin-
gle question that asks respondents to rate the ex-
tent to which users rely on cost-system data to
make decisions (use).
25 Financial performance is a key factor in HCIA�s deter-
mination of the ‘‘Top 100’’ hospitals. Five of the eight measures
Actual financial performance
Hospital financial performance is assessed
with objective, publicly available information. 24
24 While the quality of care is of utmost importance, it not
used as a performance measure because escalating healthcare
costs (and not concerns with quality) prompted the redesign of
cost systems.
Although a nonprofit hospital�s objective is not
the accumulation of residual profits, financial vari-
ables are still important in evaluating manage-
ment�s success in meeting hospital objectives. 25
In his classic characterization of the nonprofit hos-pital, Newhouse (1970) reasoned that a nonprofit
hospital�s mission is to maximize the quantity
and quality of service subject to a zero-budget con-
straint and that revenues in excess of expenses are
reinvested in facilities to improve quality and ac-
cess. More recently, Prince (1998) contended that
steady operating profits are necessary to make
ongoing investments in equipment and facilitiessimply to keep pace with the rapid advancements
in medical technology. Consistent with both New-
house and Prince, Nelson, Rust, Zahorik, and
Rose (1992) found that patient beliefs about qual-
ity were positively associated with financial perfor-
mance. The importance of financial performance
in hospitals is further evidenced by the recent trend
toward the use of financial measures in evaluatinghospital executives (Early & Cleverly, 1995; Moli-
nari et al., 1997). 26 Brickley and Van Horne
(2002) found that financial performance (return
on assets) was positively associated with CEO
pay and negatively associated with CEO turnover
in nonprofit hospitals. Moreover, the turnover/
performance relation appeared stronger in non-
profit hospitals than in for-profit hospitals.Perhaps even more important than the direct
pay-for-performance link is the opportunity for
managers to materially increase their wealth by
moving to larger, higher-paying hospitals (Leone
& Van Horne, 1999). Finally, in the US, many
nonprofit hospitals actively compete with for-prof-
it hospitals. In such instances, Duggan (2002)
found that nonprofit hospitals mimic the behaviorof their for-profit counterparts.
used to determine the ‘‘Top 100’’ are financial or operational
( ‘ ‘100 Top Hospitals : Benchmarks for Success , ’ ’
www.HCIA.com).26 According to a 1997 Hay Group Survey, 59% of nonprofit
hospitals and 92% of for-profit hospitals use a performance
incentive for senior executives that includes financial measures
(Fralicx & Liccione, 1997).
![Page 15: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/15.jpg)
Table 5
Sample means and Pearson correlations for financial performance measuresa,b (number of observations)
N Mean Cash flow Administrative expense Expense per admit
Profit margin 272 4.72% 0.627 (272)*** �0.078 (272) �0.213 (272)***
Cash flow 277 46,590 �0.060 (277) �0.050 (276)
Administrative expense 277 11.90% �0.053 (276)
Expense per admit 276 4,199
a *p < 0.10 (2-sided), **p < 0.05 (2-sided), ***p < 0.01 (2-sided).b Profit margin–operating profit margin (%). Operating profit as a percentage of operating revenue. Excludes funds from charitable
contributions and other non-operating revenue. Cash flow––cash flow per bed. Administrative expense––administrative expense ratio.
Administrative expense divided by total operating expense. Expense per admit––operating expense per case-mix adjusted admission.
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 193
Data on hospital profitability and cost control
come from the SDC Platinum Hospital Industry
Database. 27 Given the importance of investment
to a hospital�s objective function, I selected two
measures of profitability that reflect funds avail-
able on an ongoing basis to reinvest and servicedebt: operating profit margin (operating margin)
and cash flow per bed (cash flow). 28 Both mea-
sures exclude funds obtained from charitable con-
tributions and other non-operating additions to
income; and therefore, they provide a clear picture
of the hospital�s underlying operations. 29 As
shown in Table 5, operating margin and cash flow
per bed are highly correlated (r = 0.63). However,both performance measures are retained because
operating margin includes a major non-cash ex-
pense, depreciation, which serves as a barometer
for medical technology. One way a hospital man-
ager can improve operating margin is to forgo
quality-improving investments in medical technol-
ogy that increase depreciation expense.
Two measures of hospital costs are included be-cause cost control is typically cited as one of the
primary reasons hospitals adopt more functional
27 SDC Platinum is a proprietary product of the Securities
Data Company. SDC obtains information for its hospital
database from HCIA, Inc.28 All three profitability measures are significantly and
positively correlated with a hospital�s index of predictive
creditworthiness. The index of predictive creditworthiness is
HCIA�s comprehensive, quantitative assessment of a hospital�sfuture financial viability.
29 Similar results (in terms of statistical significance) are
obtained if operating margin and cash flow per bed are replaced
with operating revenue per admission, net income margin, and
cash margin. HCIA, Inc. uses cash flow margin in its evaluation
of the ‘‘Top 100 Hospitals’’.
cost systems. The administrative expense ratio
(administrative expense) is the ratio of administra-
tive expenses to total operating expenses, and it
measures management�s ability to control expenses
that are not directly related to patient care. Elden-
burg et al. (2004) used the administrative expenseratio to predict CEO turnover. Case-mix adjusted
operating cost per adjusted admission (expense per
admit) is a measure of the hospital�s average total
cost for delivering care on a per unit basis, ad-
justed for severity (HCIA, 1997, p. 150). 30
Control variables
As discussed during the development of the
hypotheses relating to contingency theory, various
elements of a hospital�s structure, strategy and
environment potentially influence cost-system de-
sign. Of these variables, those that also have been
found to influence performance are included as
control variables. Structural control variables are
size, case mix, teaching status, for-profit status,and system membership. Prior research has found
positive relations between size and costs (Elden-
burg, 1994; Lawrence, 1990). Size is measured by
the number of available beds set up and staffed
(beds). Case mix (denoted case mix) is expected
to be positively associated with profitability and
cost measures, as more severe cases typically gen-
erate higher costs and higher margins (Lawrence,
30 Admissions are adjusted by multiplying the number of
acute care discharges by a factor that inflates it to include
inpatient acute care, inpatient nonacute care, and outpatient
care. Case-mix adjustments account for differences in complex-
ity, according to the Medicare case mix.
![Page 16: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/16.jpg)
194 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
1990). Teaching hospitals (denoted teach, where
1 = teaching hospital, 0 otherwise) are expected
to have higher costs due to the added expenses
associated with teaching, research, and use of the
latest technology. For-profit hospitals (denotedfor profit, where 1 = for-profit, 0 otherwise) are
predicted to have higher profitability and lower
costs, consistent with their goal of increasing
owners� wealth. Hospitals that are members of
multi-hospital systems (denoted system, where
1 = system member, 0 otherwise) are also likely
to attain higher performance because they can
attract more capable managers, share knowledgeacross facilities, negotiate shared purchase agree-
ments with suppliers, obtain quantity discounts,
and negotiate more favorable labor contracts
(Lawrence, 1990). 31
Controls for environmental factors include
measures of competition (denoted competition),
MCO penetration (denoted MCO penetration),
the local wage level (denoted wage), and payermix. Competition is expected to be negatively
associated with all performance measures, as hos-
pitals in highly competitive environments typically
have lower operating margins and greater incen-
tive to control costs. MCO penetration is expected
to bring down local price levels. The wage index
accounts for geographic differences in the cost of
living and price levels. Payer mix is captured bythe percentage of revenues received from managed
care (denoted managed) and Medicare (denoted
Medicare). Both measures are expected to be neg-
atively associated with financial performance.
All control variables come from the SDC Plati-
num Database, except for competition, wage,MCO
penetration, and managed. Competition is calcu-
lated as one minus the sum of squared marketshares based upon beds in a county using data from
the American Hospital Association�s (AHA) an-
31 These variables also potentially control for managerial
skill (Fralicx & Liccione, 1997; Lawrence, 1990; Leone & Van
Horne, 1999). This poses a concern if more capable managers,
who enable hospitals to achieve better performance, are also
more likely to update hospital cost systems. Specifically, larger
hospitals, system hospitals, and for-profit hospitals are expected
to attract more sophisticated managers, and they are expected
to achieve better financial performance.
nual survey. The Health Care Financing Associa-
tion reports wage indices on its website.
InterStudy Publications calculates MCO penetra-
tion as the number of MCO enrollees in an area di-
vided by the area�s population. The variablemanaged is determined from a survey question
about payer mix (see the Appendix).
Correlations among measures of cost-system
functionality and control variables
Table 6 contains a correlation matrix of the cost-
systemmeasures and control variables. Because thefour cost-system design constructs relate to the
same system, they are significantly and positively
correlated. The variables DETAIL and CLAS-
SIFY are highly correlated (r = 0.68). However,
this correlation seems reasonable because the ex-
tent to which a system can classify costs (CLAS-
SIFY) affects the level of detail the system can
provide (DETAIL). 32 Several control variablesare also highly correlated with each other. Size is
significantly correlated with case mix (r = 0.69)
and the teaching indicator variable (r = 0.55) be-
cause large teaching institutions treat the most
complex cases. Competition is significantly corre-
lated with HMO penetration (r = 0.54) and the
wage index (r = 0.58). However, all control vari-
ables are retained to ensure a conservative and ro-bust test of the relation between performance and
functionality. 33
Results
The hypotheses are tested by examining rela-
tions between managers� beliefs about cost dataand actual financial performance and both the
absolute level of cost-system functionality and
the extent to which cost-system functionality is
32 Both measures were retained because factor analysis of
the relevant survey questions gave two different constructs.
Using factor scores from the orthogonal rotation, which gives
uncorrelated constructs, yields similar results in tests of the
research question.33 None of the independent variables has a variance inflation
factor in excess of 3.0 in any of the reported regressions.
![Page 17: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/17.jpg)
Table 6
Pearson correlations of functionality constructs and control variablesa,b (N = 277)
FREQUENT CLASSIFY VARIANCE Beds Case mix For profit System Competition Wages MCO penetration Managed Medicare
DETAIL 0.34*** 0.68*** 0.39*** 0.34*** 0.39*** 0.05 0.23*** 0.29*** 0.14** 0.23*** 0.19*** �0.08
FREQUENT 0.26*** 0.31*** 0.12** 0.14** 0.09 0.16*** 0.06 0.01 0.04 0.04 �0.11*
CLASSIFY 0.34*** 0.28*** 0.34*** 0.03 0.17*** 0.33*** 0.17*** 0.25*** 0.20*** �0.12*
VARIANCE 0.21*** 0.28*** 0.09 0.18*** 0.22*** 0.09 0.14** 0.13** �0.09
Beds 0.71*** �0.09 0.11* 0.42*** 0.22*** 0.31*** 0.24*** �0.36***
Case mix �0.10 0.22*** 0.51*** 0.26*** 0.38*** 0.35*** �0.41***
For profit 0.16*** 0.04 �0.03 �0.01 0.02 �0.05
System 0.17*** �0.02 0.05 0.06 �0.07
Competition 0.52*** 0.55*** 0.45*** �0.47***
Wages 0.57*** 0.42*** �0.35***
MCO
penetration
0.46*** �0.32***
Managed �0.42***
a *p < 0.10 (2-sided), **p < 0.05 (2-sided), ***p < 0.01 (2-sided).b DETAIL––the level of detail provided by the system. CLASSIFY––the system�s ability to classify costs according to behavior. FREQUENT––the frequency with
which cost information is disseminated throughout the organization. VARIANCE––the type and number of variances calculated. Size––total number of beds set up and
staffed divided by 1000. Mix––Medicare case-mix index. For profit––an indicator variable that is 1 if the hospital is a for-profit entity, and zero otherwise. System––an
indicator variable that is 1 if the hospital is part of a multi-hospital system and zero otherwise. Teach––an indicator variable that is 1 if the hospital is a teaching hospital
and zero otherwise. Competition––calculated as 1––sum of squared market shares, based upon beds in a county. Wages––Medicare wage index for MSA or state. MCO
penetration––MCO penetration. The number of MCO enrollees in an MSA divided by the area�s population. Managed––percentage of revenue received from managed
care. Medicare––percentage of revenue received from Medicare.
M.J.Pizzin
i/Acco
untin
g,Organiza
tionsandSociety
31(2006)179–210
195
![Page 18: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/18.jpg)
Table 7
Association between absolute level of cost-system functionality
and managers� evaluations of the relevance and usefulness of
cost dataa,b
Relevant Use
Intercept 2.074*** (7.73) �0.447*** (0.76)
DETAIL 0.141*** (3.05) 0.274*** (2.72)
CLASSIFY 0.081* (1.73) 0.039 (0.38)
FREQUENT 0.098*** (2.73) 0.112 (1.43)
VARIANCE 0.026 (0.63) 0.073 (0.82)
Control Variablesc
Beds 0.420* (1.96) 0.074 (0.16)
Case mix 0.299* (1.73) 0.143 (0.38)
System 0.167*** (2.97) 0.122 (0.99)
Medicare 0.004 (1.72) 0.002 (0.40)
7.25*** 4.03***
Adj. R2 0.228 0.125
N 277 277
a *p < 0.10 (2-sided), **p < 0.05 (2-sided), *** (2-sided), t-
statistics in ( ).b Relevant––Managers� perceptions regarding the extent to
which managers believe that cost data meets their information
needs. Use––Managers� perceptions regarding the extent to
which they rely on cost data to make decisions. See Appendix
(p. A-2) for questions used to calculate relevant and use.
DETAIL––the level of detail provided by the system. CLAS-
SIFY––the system�s ability to classify costs according to
behavior. FREQUENT––the frequency with which cost infor-
mation is disseminated throughout the organization. VARI-
ANCE––the type and number of variances calculated.c Only statistically significant control variables are included
in table. Control variables used in the regression are as follows:
Size––total number of beds set up and staffed divided by 1000.
Mix––Medicare case-mix index. For profit––an indicator vari-
able that is 1 if the hospital is a for-profit entity, and zero
otherwise. System––an indicator variable that is 1 if the hospital
is part of a multi-hospital system and zero otherwise. Teach––
an indicator variable that is 1 if the hospital is a teaching
hospital and zero otherwise. Competition––calculated as 1––
sum of squared market shares, based upon beds in a county.
Wages––Medicare wage index for MSA or state. MCO pene-
tration––MCO penetration. The number of MCO enrollees in
an MSA divided by the area�s population. Managed––percent-
age of revenue received from managed care. Medicare––per-
centage of revenue received from Medicare.
196 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
aligned with a hospital�s strategy, structure, and
environment.
The level of cost-system functionality
In tests of the level of functionality, variables
capturing managers� beliefs and actual perfor-
mance are regressed on the four cost-system attri-
butes (DETAIL, CLASSIFY, FREQUENT, and
VARIANCE) and the control variables. The re-
sults for managers� beliefs about the relevance
and usefulness of cost data are contained in Table
7. Both models are significant (p > 0.001) and theadjusted r-squareds are 0.228 and 0.125 for rele-
vant and use respectively. 34 The coefficient on DE-
TAIL is positive and highly significant (p < 0.01,
two-sided) for both measures. The coefficients for
CLASSIFY and FREQUENT are positively and
significantly associated with data relevance
(p < 0.10 and p < 0.01, two-sided, respectively),
but neither is associated with usefulness. However,recall that DETAIL and CLASSIFY are highly
correlated (r = 0.68). If DETAIL is eliminated
from either model, the coefficients on CLASSIFY
and FREQUENT increase in significance in the
relevant regression (p < 0.01, two-sided), and they
become significant in the use regression (p < 0.05,
two-sided). Because all the variables are standard-
ized, one can interpret the coefficients on theDETAIL construct, for example, as follows: a
1-standard deviation increase in DETAIL is asso-
ciated with a 0.14-standard deviation increase in
relevance and a 0.27-standard deviation increase
in usefulness of cost data. The VARIANCE con-
struct is not associated with managers� beliefs
about data relevance or usefulness. This result is
consistent with claims that traditional varianceanalysis is not a useful management tool (e.g.,
Bastable & Bao, 1988; Cheatham & Cheatham,
34 White�s (1980) test indicates no evidence of heteroskedas-
ticity. Shapiro and Wilk (1965) tests indicate that the residuals
are not distributed normally; however, histograms and normal
probability plots indicate the residuals are close to normal.
Elimination of observations with the greatest influence, as
measured by the change in a regression coefficient when an
observation is omitted from the analysis (DFBETA), had no
material impact on the results.
1996; Cooper & Kaplan, 1992; Mak & Roush,
1996), and that budgeting is not well suited to hos-
pital environments (Abernethy & Vagnoni, 2004;
Covaleski & Dirsmith, 1986; Kurunmaki, 1999).
The relevance and usefulness of cost-system
data are critical to the assertion that cost-system
design can be used to enhance financial perfor-
![Page 19: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/19.jpg)
Table 8
Association between the absolute level of cost-system functionality and financial performance a,b
Operating profit Cash flow Administrative expensec Expense per admitc
Intercept 8.874* (1.92) 16,598 (0.66) 9.518*** (4.265) 2,070*** (3.36)
DETAIL 1.264** (1.97) 11,899*** (3.36) �0.555* (1.77) 83 (0.75)
CLASSIFY �0.257 (0.40) �135 (0.04) 0.287 (0.92) 11 (0.10)
FREQUENT �0.515 (1.06) �814 (0.30) 0.086 (0.36) �15 (�0.18)
VARIANCE �0.809 (1.46) �19,264** (2.53) 0.820*** (3.02) 48 (0.51)
Control variablesd
Beds 0.984 (0.34) �21,864 (1.36) �3.771*** (2.65) 924** (2.11)
Case mixe 4.913** (2.06) 52,869*** (4.07) �1.176 (1.02)
For profit 3.673** (2.55) �2,003 (0.25) 1.590** (2.28) 81 (0.33)
System �1.389* (1.80) �9,107** (2.14) 1.147*** (3.05) 18 (0.14)
Teach �0.827 (0.82) 2,836 (0.52) �0.408 (1.02) 516*** (3.07)
Competition �2.504 (1.56) 7,487 (0.86) 1.876** (2.42) 169 (0.62)
Wages �7.047** (2.31) �29,216* (1.74) 3.506** (2.35) 2,333*** (4.45)
MCO penetration �4.693 (1.59) �37,314** (2.27) �0.609 (0.42) �827 (1.62)
F 2.862*** 5.382*** 5.553*** 7.371***
Adj. R2 0.088 0.182 0.188 0.232
N 272 277 277 276
a *p 0.10 (2-sided), **p < 0.05 (2-sided), ***p < 0.01 (2-sided), t-statistics in ( ) below coefficient.b DETAIL––the level of detail provided by the system. CLASSIFY––the system�s ability to classify costs according to behavior.
FREQUENT––the frequency with which cost information is disseminated throughout the organization. VARIANCE––the type and
number of variances calculated.c Smaller values are considered favorable because they indicate better cost control or greater efficiency.d Only statistically significant control variables are included in table. Control variables used in the regression are as follows: Size––
total number of beds set up and staffed divided by 1000. Mix––Medicare case-mix index. For profit––an indicator variable that is 1 if
the hospital is a for-profit entity, and zero otherwise. System––an indicator variable that is 1 if the hospital is part of a multi-hospital
system and zero otherwise. Teach––an indicator variable that is 1 if the hospital is a teaching hospital and zero otherwise. Compe-
tition––calculated as 1––sum of squared market shares, based upon beds in a county. Wages––Medicare wage index for MSA or state.
MCO penetration––MCO penetration. The number of MCO enrollees in an MSA divided by the area�s population. Managed––
percentage of revenue received from managed care. Medicare––percentage of revenue received from Medicare.e Expense per admit is adjusted for case-mix; therefore, the case-mix variable is not included as a control.
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 197
mance. Unless more highly functional cost systems
provide more relevant and useful data that lead to
‘‘better’’ decisions, they should be unrelated to dif-
ferences in financial performance. 35 Accordingly,
the tests of managers� beliefs about the cost systemprovide a basis for the expectation that cost sys-
tems that can supply greater detail, better classify
35 As an alternative to separate regressions, I constructed a
path model of the relation between cost-system functionality,
managers� beliefs about relevance and usefulness, and actual
financial performance. The results of the path analysis were
consistent with the causal model linking greater functionality
with increased data relevance and use. The total effects of the
DETAIL and CLASSIFY constructs on financial performance
were positive and significant. However, the results of the path
model were not tabulated because the approach mixes the
relatively subjective relevance and use measures with the more
objective cost-system attribute and financial measures.
costs, and distribute information more frequently
and broadly can lead to enhanced financial
performance. 36
Table 8 contains coefficients from the regression
of actual financial performance measures on the
36 In theory, if all hospitals have optimally selected their cost
systems, there should be no relation between cost-system design
and performance. While hospitals may choose optimally in the
long run, observers suggest that, in the short run, many
hospitals have not yet adjusted their cost systems to accom-
modate the increased information needs created by sweeping
environmental changes (Hill & Johns, 1994; Orloff et al., 1990;
Young & Pearlman, 1993). For example, a recent survey found
that while percent 80% of managers surveyed placed a high
value on obtaining specific cost information, only 33%
expressed confidence in the quality of their cost data (Serb,
1997).
![Page 20: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/20.jpg)
198 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
level of cost-system functionality. 37 All four mod-
els are significant (p > 0.001) and the adjusted r-
squareds range from 0.088 for operating profit to
0.232 for expense per admit. 38 The capability to
provide more detail (DETAIL) is positively andsignificantly associated with operating margin
(p < 0.05, two-sided) and cash flow per bed
(p < 0.01, two-sided). 39 DETAIL is negatively
associated with the administrative expense ratio
(p < 0.10, two-sided), which indicates that admin-
istrative expenses are lower when a cost system
can provide greater detail. The significant associa-
tions between DETAIL and measures of bothmanagers� beliefs and financial performance sup-
port the premise that more highly functional sys-
tems provide more relevant and useful cost data,
which lead to better financial performance.
Neither CLASSIFY nor FREQUENT is signif-
icantly associated with performance in Table 8.
However, if the highly correlated DETAIL con-
struct is removed from this regression, CLASSIFYis positively and significantly associated with cash
flow per bed (p < 0.05, two-sided). These results
are consistent with prior findings regarding man-
agers� beliefs in that the ability to better classify
data becomes highly significant only when DE-
TAIL is removed from the regression. Although
the factor analysis suggested that the ability to
provide greater detail and the ability to classifycosts according to behavior are separate con-
structs, it is difficult to isolate their relations with
managers� beliefs about performance and actual
cash flow. The variable, FREQUENT, is not asso-
37 I also investigated two-way and three-way interactions
among the cost-system design constructs by including multipli-
cative terms. However, the high degree of multicolinearity
between the multiplicative terms and the individual constructs
precluded a meaningful interpretation of the coefficients.38 White (1980) test indicates no evidence of heteroskedas-
ticity. Shapiro and Wilk (1965) tests suggest that only the
residuals of the cash flow regression are distributed normally.
However, histograms and normal probability plots indicate the
residuals are close to normal. Elimination of observations with
the greatest influence, as measured by the change in a regression
coefficient when an observation is omitted from the analysis
(DFBETA), had no material impact on the results.39 DETAIL is also positively and significantly associated
with revenue per bed, investment per bed, and the number of
admissions per bed (p < 0.10, two-tailed), an efficiency measure.
ciated with actual financial performance even
though managers find information that is reported
more frequently to be more relevant and useful.
Interestingly, hospitals that engage extensively in
variance analysis exhibit significantly lower cashflow per bed (p < 0.05, two-sided) and relatively
higher administrative costs (p < 0.01, two-sided).
As discussed earlier, a potential explanation is that
greater use of variance analysis may lead to inap-
propriate decisions. 40
Finally, operating expense per admission, which
is driven by clinical resource consumption, is not
significantly associated with any of the functional-ity measures. This result is consistent with prior re-
search suggesting that initial efforts to improve
hospital performance typically focus on admin-
istrative processes and not clinical processes
(Shortell, Levin, O�Brien, & Hughes, 1995).
Furthermore, as discussed previously, hospital
managers have little direct control over clinical
expenditures because physicians control clinicaldecisions. Only if managers present physicians
with appropriate cost data can managers poten-
tially influence physician behavior (Eldenburg,
1994; Evans et al., 1995). Even then, the use of
accounting information in directing physician
behavior raises significant concerns (e.g., Aber-
nethy & Stoelwinder (1996); Abernethy & Vag-
noni, 2004; Chua & Degeling, 1991; Evans,Hwang, & Nagarajan, 2003; Jones & Dewing,
1997; Preston, 1992).
In summary, there is some evidence that sys-
tems that are better at supplying detail and classi-
fying costs provide more relevant and useful cost
data, which lead to better financial performance.
However, a major concern with cross-sectional
tests is that they do not address causality. Specifi-cally, better performing hospitals may have the
means to invest in cost systems development; thus,
strong financial performance could give rise to
more functional cost systems. Still, if it is the case
40 To confirm the validity of the VARIANCE construct, I
substituted each of the individual questions used to calculate
the VARIANCE construct in place of VARIANCE in the
performance regressions The only notable difference was that
those hospitals using price variances had significantly higher
expenses per admission.
![Page 21: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/21.jpg)
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 199
that hospitals invest in cost systems simply because
they have excess funds, it is not clear that manag-
ers would find information supplied by more func-
tional systems to be more relevant or more useful.
Only when managers believed that a specific sys-tem capability (i.e., supplied greater detail, better
classified costs) provided more relevant and useful
information was financial performance better.
Causality is further examined in the extensions.
Alternatively, performance and cost-system func-
tionality may be simultaneously determined.
Simultaneity is considered in the next section. 41
Alignment with operating environment
The levels tests of the previous section assumed
that the appropriate benchmark for determining if
a system is ‘‘highly’’ functional is the average level
of cost-system functionality for all hospitals in the
sample. However, contingency theory contends the
appropriate benchmark is the expected level offunctionality, given the hospital�s organizational
context. Accordingly, I investigate associations be-
tween financial performance and the residuals
from an empirical model estimating the bench-
mark level of cost-system functionality as a func-
tion of hospital structure, strategy, and
environment (Hypotheses 2a and 2b). The estima-
tion of the benchmark model itself is an operation-alization of the selection approach to measuring fit
between control system design and organizational
context (Selto et al., 1995, p. 669). The selection
approach assumes that context drives organiza-
tional design, and therefore, defines fit in terms
of associations between context and control sys-
tem design. The residuals from the benchmark
model are a variant of the systems approach tomeasuring fit (Selto et al., 1995, p. 670). The resid-
uals measure variations in overall systematic fit as
41 Hausman�s specification error test can potentially identify
whether the independent variables are exogenous. However,
this test requires an instrumental variables estimate of the
regression coefficient on the design construct, and there are no
obvious choices for instruments. Moreover, a simultaneous
equations approach renders a system of five equations with five
endogenous variables. and thus would require four instrumen-
tal variables.
determined by the hospital�s organizational
context. 42
Following Chenhall and Langfield-Smith
(1998), strategy is characterized along two dimen-
sions, product differentiation and cost control.Product differentiation, denoted lead, is measured
by taking the average of three survey questions that
ask respondents to rate the importance of estab-
lishing clinical leadership, technological leadership,
and leadership in business processes. The extent to
which hospitals follow a low-cost strategy, denoted
cost, is quantified with two survey questions that
ask respondents to rate the importance of control-ling and eliminating unprofitable services. 43
To model cost-system functionality, I regress
each functionality construct (DETAIL, CLAS-
SIFY, FREQUENT, VARIANCE) on measures
of a hospital�s strategy, structure, and environ-
ment. Hospital structure is characterized by hospi-
tal size (size), case mix (case mix), teaching status
(teaching), system ownership (system), and hospi-tal district membership (district). The variable, dis-
trict, is set equal to 1 if a hospital is a member of a
hospital district association, and it is 0 otherwise.
The remaining variables were defined in the section
on control variables. Environmental factors are
also captured by variables defined in the section
on control variables: the level of competition (com-
petition), the degree of MCO penetration (MCO
penetration), and the percentage of managed (man-
aged) and Medicare (Medicare) patients in a hospi-
tal�s patient mix. As discussed in the section on
contingency theory, all structural and market vari-
ables are expected to be positively associated with
functionality, except district, which is predicted to
be negatively associated with functionality.
Results of the benchmark regressions are con-tained in Table 9. The model has the greatest
explanatory power for DETAIL (adjusted
42 Others have used the top performing organizations as the
benchmark in determining systematic fit (Selto et al., 1995).
This approach was not used here because, as discussed in
footnote 29, it is unlikely that hospital cost system design is in a
state of equilibrium.43 Each set of questions loads on a different factor. Cron-
bach�s alphas are 0.56 and 0.52 for lead and cost, respectively.
Each variable is the average of the standardized responses to
each set of questions.
![Page 22: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/22.jpg)
Table 9
Regression of functionality constructs on hypothesized determinants of functional capabilitya,b,c
DETAIL CLASSIFY FREQUENT VARIANCE
INTERCEPT �1.857*** (4.15) �1.597 (3.75) �0.228 (0.48) �1.485*** (3.59)
Cost + 0.115* (1.86) (1.40) 0.243*** (3.69) 0.178*** (3.12)
Lead � 0.080 (1.47) 0.040 (0.77) 0.050 (0.85) �0.032 (0.63)
Teach � 0.180 (1.45) 0.079 (0.67) 0.028 (0.22) 0.156 (1.36)
Size +(1.33) 0.487 0.023 (0.07) 0.104 (0.27) �0.079 (0.23)
Case mix + 0.591** (1.99) 0.651** (2.31) 0.336 (1.06) 0.689** (2.52)
For profit + 0.188 (1.05) 0.165 (0.97) 0.298 (1.56) 0.329*** (1.99)
System + 0.204** (2.13) 0.142 (1.56) 0.224*** (2.20) 0.162* (1.84)
Competition + 0.194 (1.01) 0.284 (1.56) �0.241 (1.18) 0.111 (0.63)
MCO penetration + 0.226 (0.65) 0.396 (1.20) 0.109 (0.29) 0.106 (0.33)
Managed + 0.005 (1.08) 0.002 (0.57) �0.001 (0.32) 0.002 (0.46)
Medicare + 0.010** (2.49) 0.006 (1.64) �0.005 (1.12) 0.005 (1.35)
District � �0.540*** (2.65) �0.601*** (3.11) 0.163 (0.75) 0.064 (0.34)
F 7.95*** 6.35*** 2.53*** 3.94***
Adj. R2 0.232 0.188 0.063 0.113
N 277 277 277 277
a *p < 0.10 (2-sided), **p < 0.05 (2-sided), ***p < 0.01 (2-sided), t-statistics in ( ).b DETAIL––the level of detail provided by the system. CLASSIFY––the system�s ability to classify costs according to behavior.
FREQUENT––the frequency with which cost information is disseminated throughout the organization. VARIANCE––the type and
number of variances calculated.c Cost––the extent to which the hospital follows a cost-focused strategy. Lead––the extent to which the hospital follows a leadership
strategy. Size––total number of beds set up and staffed divided by 1000. Mix––Medicare case-mix index. For profit––an indicator
variable that is 1 if the hospital is a for-profit entity, and zero otherwise. System––an indicator variable that is 1 if the hospital is part of
a multi-hospital system and zero otherwise. Teach––an indicator variable that is 1 if the hospital is a teaching hospital and zero
otherwise. Competition––calculated as 1––sum of squared market shares, based upon beds in a county. Wages––Medicare wage index
for MSA or state. MCO penetration––MCO penetration. The number of MCO enrollees in an MSA divided by the area�s population.Managed––percentage of revenue received from managed care.Medicare––percentage of revenue received fromMedicare. District––an
indicator variable that equals 1 if the hospital is a member of a hospital district association.
200 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
R2 = 0.23) and the least explanatory power for
FREQUENT (adjusted R2 = 0.06). All statistically
significant associations have the predicted sign.
Low-cost strategic orientation, case mix, system
ownership, and district membership have the
strongest associations with cost-system functional-
ity. Residuals from the benchmark model measure
the extent to which the functionality of a hospital�scost system differs from that expected, given the
hospital�s operating environment. Positive residu-
als measure the degree to which a system is more
functional than expected; they are denoted with
the variables more DETAIL, more CLASSIFY,
more FREQUENT, and more VARIANCE. These
variables equal the residual if the residual is posi-
tive, and zero otherwise. Similarly, negative devia-tions from expected indicate that the system has
relatively little functionality, and they are denoted
with the variables less DETAIL, less CLASSIFY,
less FREQUENT, and less VARIANCE. These
variables equal the absolute value of the residual
if it is negative, and zero otherwise. For example,
a large value of less FREQUENT indicates that
the system reports cost information less frequently
than expected, given its operating environment.
The relation between performance and cost-sys-
tem alignment is investigated by regressing perfor-
mance measures on the positive and negativeresiduals and the control variables described in
prior tests. In addition to addressing hypotheses
2a and 2b, this approach has two advantages.
First, it does not constrain the coefficients on the
functionality measures to be the same regardless
of whether a system has relatively ‘‘too much’’ or
‘‘too little’’ functional capability. Second, the ap-
proach provides a partial control for endogeneityby modeling cost-system functionality as a func-
tion of a number of exogenous variables (e.g.,
competition, MCO penetration, etc.), as well as
three additional instrumental variables (cost, lead,
![Page 23: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/23.jpg)
Table 10
Regression of managers� beliefs about relevance and usefulness
on positive and negative residuals from benchmark
regressionsa,b
Relevant Use
Intercept �2.220*** (�2.02) �1.508** (2.47)
More DETAIL 0.297** (2.01) 0.561*** (2.75)
Less DETAIL �0.744*** (�6.08) �0.043 (0.25)
More CLASSIFY 0.034 (0.22) 0.077 (0.37)
Less CLASSIFY �0.159 (1.36) �0.034 (0.21)
More FREQUENT 0.046 (0.36) 0.120 (0.67)
Less FREQUENT �0.115 (1.00) �0.090 (0.57)
More VARIANCE 0.013 (0.91) �0.096 (0.49)
Less VARIANCE 0.047 (0.35) �0.240 (1.26)
F 12.223*** 4.393***
Adj. R2 0.409 0.181
N 277 277
a **p < 0.05 (2-sided), ***p < 0.01 (2-sided), t-statistics in ( ).b Control variables are not included in table. Control vari-
ables used in the regression are as follows: Size––total number
of beds set up and staffed divided by 1000. Mix––Medicare
case-mix index. For profit––an indicator variable that is 1 if the
hospital is a for-profit entity, and zero otherwise. System––an
indicator variable that is 1 if the hospital is part of a multi-
hospital system and zero otherwise. Teach––an indicator vari-
able that is 1 if the hospital is a teaching hospital and zero
otherwise. Competition––calculated as 1––sum of squared
market shares, based upon beds in a county. Wages––Medicare
wage index for MSA or state. MCO penetration––MCO pene-
tration. The number of MCO enrollees in an MSA divided by
the area�s population. Managed––percentage of revenue
received from managed care. Medicare––percentage of revenue
received from Medicare.
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 201
and district) that are not included in the perfor-
mance regressions. 44
Table 10 contains regressions of managers� beliefsabout the relevance and usefulness of cost informa-
tion on the measures of cost-system alignment.Man-
agers find cost information to be significantly more
relevant (p < 0.05, two-sided) and useful (p < 0.01,
two-sided) when it is supplied by systems that pro-vide relatively more detail than expected (more
DETAIL). Correspondingly, managers find infor-
44 Obviously, the strategy variables are also choice variables
for the hospital. I assume that these choices precede the choice
of cost-system design, and are therefore exogenous to the choice
of cost-system attributes. Additional untabulated tests indicate
that these three variables are generally correlated with func-
tional capabilities and uncorrelated with the performance
measures, supporting their use as instruments in the function-
ality regressions and omission in the performance regressions.
mation to be significantly less relevant (p < 0.01,
two-sided) when it comes from systems that provide
relatively less detail than expected (less DETAIL).
Residuals from the CLASSIFY benchmark model
are significant only if the highly correlated residualsfrom the DETAIL benchmark model are re-
moved. 45 In this case, more CLASSIFY is posi-
tively associated with relevant (p < 0.05, two-sided)
and use (p < 0.10, two-sided), and less CLASSIFY
is negatively associated with relevant (p < 0.01,
two-sided). Residuals from the FREQUENT and
VARIANCE benchmark models are not signifi-
cantly associated with managers� beliefs about therelevance and usefulness of cost data at all.
Table 11 contains regressions of financial per-
formance on the measures of cost-system align-
ment. With respect to the DETAIL construct,
hospitals that have relatively greater functional
capability than hospitals in similar operating envi-
ronments have significantly higher operating mar-
gins (p < 0.05, two-sided) and cash flow per bed(p < 0.05, two-sided). Hospitals with cost systems
that provide less detail than expected have lower
cash flow per bed (p < 0.10, two-sided) and incur
proportionally higher administrative expenses
(p < 0.10, two-sided). As in prior tests, residuals
from the CLASSIFY benchmark model are signif-
icant only if the DETAIL residuals are removed.
In this case, less CLASSIFY is negatively associ-ated with cash flow per bed (p < 0.05, two-sided),
suggesting that hospitals that do a relatively poor
job of classifying costs have significantly lower
cash flows. The relative frequency with which costs
are reported is not associated with performance.
Finally, hospitals that conduct relatively more var-
iance analysis have significantly lower operating
margins (p < 0.05, two-sided). Overall, the resultsfrom residuals regressions are consistent with
those of the levels tests. This suggests that cost sys-
tems that have greater detail and classification
capabilities, relative to all hospitals and relative
to their peers, achieve better performance. More-
over, there is nothing to indicate that the levels
tests have been confounded by endogeneity.
45 More DETAIL and more CLASSIFY have a correlation
of 0.53 (p < 0.01, two-sided); less DETAIL and less DETAIL
have correlation of 0.51 (p < 0.01, two-sided).
![Page 24: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/24.jpg)
Table 11
Regression of profitability and cost measures on positive and negative residuals from benchmark regressionsa,b
Operating profit Cash flow Administrative expensec Expense per admitc
INTERCEP 8.196* (1.72) 4,877 (0.19) 8.923*** (3.89) 1,645** (2.51)
More DETAIL 2.862** (2.27) 14,788** (2.11) �0.187 (0.30) 271 (1.24)
Less DETAIL 0.124 (0.12) �10,138* (1.75) 0.873* (1.70) 4 (0.02)
More CLASSIFY �1.190 (0.92) �7,714 (1.07) 0.697 (1.10) �61 (0.27)
Less CLASSIFY �0.912 (0.91) �6,231 (1.13) 0.054 (0.11) 22 (0.13)
More FREQUENT �1.152 (1.04) �516 (0.08) �0.435 (0.80) 71 (0.37)
Less FREQUENT �0.198 (0.20) 77 (0.01) �0.557 (1.16) 74 (0.44)
More VARIANCE �2.539** (2.01) �10,690 (1.57) 0.826 (1.37) 88 (0.41)
Less VARIANCE �0.739 (0.63) 3,508 (0.54) �0.859 (1.49) 68 (0.33)
F 2.510*** 4.203*** 4.420*** 5.671***
Adj. R2 0.091 0.173 0.182 0.225
N 277 277 277 277
a *p < 0.10 (2-sided), **p < 0.05 (2-sided), ***p < 0.01 (2-sided), t-statistics in ( ).b Control variables are not included in table. Control variables used in the regression are as follows: Size––total number of beds set
up and staffed divided by 1000. Mix––Medicare case-mix index. For profit––an indicator variable that is 1 if the hospital is a for-profit
entity, and zero otherwise. System––an indicator variable that is 1 if the hospital is part of a multi-hospital system and zero otherwise.
Teach––an indicator variable that is 1 if the hospital is a teaching hospital and zero otherwise. Competition––calculated as 1––sum of
squared market shares, based upon beds in a county. Wages––Medicare wage index for MSA or state. MCO penetration––MCO
penetration. The number of MCO enrollees in an MSA divided by the area�s population. Managed––percentage of revenue received
from managed care. Medicare––percentage of revenue received from Medicare.c Smaller values are considered favorable because they indicate better cost control or greater efficiency. Control measures were
included in regression, but they are not shown in the table.
202 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
Extensions––robustness of results
Because causality is a major concern in cross-
sectional tests such as these, I examine the research
question using alternative specifications that can
potentially detect evidence of causality and omit-
ted variables.
System changes
To examine causality, I estimate the likelihood
that the cost system is currently changing or has
undergone a significant change within the past
two years as a function of past performance and
the control variables. 46 This specification exam-
46 The timing of cost-system changes is determined by the
following questions: When was the last major change to your
organization�s cost accounting system completed? (Major
changes include the installation of new software, integration
of the cost accounting system with other control systems,
adoption of a new overhead allocation scheme, etc.) 36%
responded that the system was currently changing, 28%
completed last major change within the past two years, 18%
within last three-to-five years, 18% more than five years since
last major change.
ines whether above (below) average past perfor-
mance is associated with the decision to change
the cost system. A positive association between
past performance and major cost-system change
is consistent with the premise that better-perform-
ing hospitals invest in cost-system upgrades and,
consequently, have more functional systems. A
negative association may indicate that poor per-formers attempt to improve performance by
upgrading cost systems. However, I find that the
decision to change the cost system is not signifi-
cantly associated with past performance at all.
Therefore, the results do not support the premise
that strong financial performance drives invest-
ments in cost-system functionality.
Performance changes
To test for omitted variables bias, I regress
changes in performance over a one-year period
on cost-system functionality, past performance,
and the control variables. Hospitals that have
undergone major cost-system changes within the
past two years are removed from the sample in this
![Page 25: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/25.jpg)
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 203
specification so that the functionality measures de-
scribe systems that have been in place for at least
three years. This test again controls for past per-
formance. It also helps control for omitted vari-
ables that may be related to both current levelsof cost-system functionality and financial perfor-
mance, such as managerial skill. Finally, simulta-
neity is less of a concern in this specification
because the functionality constructs describe a cost
system that has been in place for at least three
years prior to measuring the improvement in per-
formance. Thus, the system was selected prior to
many of the actions taken to affect the currentchange in performance.
Regressions using the change specification show
that the ability to provide detailed information is
significantly and positively associated with in-
creases in cash flow per bed, and negatively associ-
ated with reductions in the administrative expense
ratio. In contrast, hospitals that engaged more
extensively in variance analysis also experiencedgreater increases in administrative expenses. These
results are consistent with the findings from the
levels specification, and therefore, provide no evi-
dence that omitted variables or simultaneity con-
founded the levels tests.
Conclusion
This study examines relations among cost-sys-
tem functionality, managers� beliefs about the rel-
evance and usefulness of cost data, and actual
financial performance in US hospitals. The results
indicate that managers believe that systems that
supply greater cost detail, on average and relative
to hospitals in similar organizational contexts,provide more relevant and useful data. Hospitals
with such systems are significantly more profitable,
generate greater cash flows, and have proportion-
ately lower administrative expenses. These findings
support the premise that more functional cost sys-
tems supply managers with more relevant data,
which they use to make performance-enhancing
decisions. Interestingly, there was no relation be-tween the detail capability and expense per admis-
sion, which is largely driven by physicians� clinicaldecisions and not managers� decisions. This is con-
sistent with previous research that suggests cost
containment efforts have been directed at adminis-
trative processes and not clinical process (Shortell
et al., 1995). Ultimately, the ability to significantly
reduce healthcare expenditures lies in containmentof the direct costs of patient care, and not in sim-
ply improving administrative efficiency.
There is also some evidence that better clas-
sification of costs, in general and relative to a
homogeneous peer group, is associated with higher
managerial evaluations of data relevance and use,
as well as actual financial performance. However,
it was not possible, with this data, to isolate the ef-fects of systems that better classify costs from
those that provide greater detail. Increased report-
ing frequency was favorably associated with
managers� beliefs about data relevance and time-
lines, but it was not associated with financial
performance.
I found no relation between the extent to
which systems calculate variances and managers�evaluations of the relevance and usefulness of cost
information. Interestingly, hospitals with systems
that calculate more variances had significantly
lower profits and relatively higher administrative
expenses. These results could indicate reverse cau-
sality, as poorly performing hospitals may engage
in variance analysis to identify and correct prob-
lems. However, poor performance in the pastwas not associated with decisions to change the
cost system. Alternatively, these findings support
claims that variance analysis can be problematic,
especially in the hospital setting. Consistent with
this explanation, tests that control for prior per-
formance and the timing of cost-system changes
find increases in administrative expenses among
those hospitals that performed more varianceanalysis.
The similarity of the findings regarding the level
of functionality on average and relative to contex-
tually similar peers is consistent with claims from
the practitioner literature that hospitals have not
yet adjusted their cost systems to accommodate
the increased information needs created by sweep-
ing environmental changes (e.g., Hill & Johns,1994; Orloff et al., 1990; Serb, 1997; Young &
Pearlman, 1993). Specifically, more detail capabil-
ity relative to all hospitals and relative to similar
![Page 26: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/26.jpg)
N Mean Median St. Dev.
Payer level
(e.g. Medicare)
324 5.1 6 2.0
Contract level 321 4.5 5 2.2
Physician level 323 4.9 6 2.0Procedure level 324 4.9 6 2.1
Per Diem level 320 4.3 5 2.2
Patient level 324 5.0 6 2.1
204 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
hospitals is positively associated with perfor-
mance, while less, in absolute and relative terms,
is negatively associated with performance. How-
ever, while healthcare management literature lar-
gely supports cost-system refinement alongseveral dimensions, this study indicates that only
the detail capability, and possibly the classification
capability, warrant investment. Moreover, even
improved detail capability has not helped manag-
ers to reduce clinical costs.
The findings are subject to a number of limita-
tions. Cross-sectional studies such as this can
establish associations, but not causality. Futurestudies using panel data would enable researchers
to investigate performance changes arising from
decisions made using cost system data. Similarly,
endogeneity is also an obvious concern with this
test design. Although alternative specifications re-
veal no evidence of reverse causality or endogene-
ity with regard to the detail and classify
constructs, these potentially confounding factorscannot be eliminated. Another factor that may
affect these results is the noisiness of the mea-
sures. A mail survey prevents an assessment of
the survey respondent�s actual knowledge of the
cost-accounting system, although the surveys
were mailed to chief financial officers. A mail sur-
vey also prevents the respondent from effectively
clarifying his or her understanding of the ques-tions. Moreover, the single-industry sample limits
the ability to generalize these results to other
industries. Finally, the hypotheses were primarily
based upon economic theories of organization,
and some argue that such theories do not ade-
quately capture the complex interplay between
various internal and external organizational ac-
tors in hospitals (e.g., Abernethy & Vagnoni,2004; Chua & Degeling, 1991; Covaleski & Dir-
smith, 1986; Jones & Dewing, 1997). Despite
these limitations, this study has important impli-
cations for research in both accounting and
healthcare financial management. The results pro-
vide the first empirical evidence of a relation be-
tween cost-system functionality and actual
financial performance and thereby help justifyclaims regarding the need for greater investment
in certain dimensions of hospital cost systems.
However, the results also suggest that cost-system
functionality has not yet been successfully intro-
duced into the management of clinical expendi-
tures. Until this takes place, the usefulness of
cost-system design in containing hospital costs is
limited.
Acknowledgements
I thank Christopher Ittner, David Larcker,
Rick Lambert, John Core, Keith Weigelt, and
workshop participants at the University of Penn-
sylvania, Emory University, Rice University, theUniversity of Washington, Stanford University,
Harvard�s Kennedy School of Government, the
University of Texas at Dallas, the University of
Virginia�s Darden School, and the 2000 AAA
Managerial Accounting Conference for their
comments. This paper has benefited greatly from
the comments of two anonymous referees. I also
thank Taylor Randall and Steve Walston, whowere instrumental in collecting the data used in
this study. The financial support of the Deloitte
and Touche Foundation is gratefully acknowl-
edged.
Appendix. Number of responses, mean response,
and coding method for survey questions
Cost-system design constructs
Detail––level of detail provided
To what extent does the cost accounting system
provide data that allow you to analyze costs at the
following levels? 1 = not at all, 7 = completely
![Page 27: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/27.jpg)
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 205
The cost accounting function can easily custom-
ize reports to the specification of users. 1 = strongly
agree, 7 = strongly disagree, N = 320, mean = 4.2.
N Not
reported
Daily Weekly Monthly Quarterly Semi-annually Annually
Response codinga 1 7 6 5 4 3 2
Sr. Managers
(CEO, CFO)
309 40 10 12 159 40 18 30
Middle Managers
(Dept. Heads)
307 59 5 8 168 27 14 26
Clinical Managers
(Phys. Mgrs.)
307 107 5 7 115 43 11 18
Nursing Staff 307 133 0 6 107 27 10 21Medical Staff 307 136 2 1 64 64 17 20
a Method of coding responses. Higher numbers were assigned to more frequent reporting.
Classify––Classification of costs according to
behavior
Does your cost system have a formalized meth-
od of distinguishing between the following costs?
1 = not at all, 7 = completely
N Mean Median St. Dev.
Direct and indirect 321 5.1 6 1.9Fixed and variable 318 4.7 5 2.1
Controllable and
non-controllable
317 3.7 4 2.0
If your hospital tracks fixed and variable costs,
how are they separated? (Check one, N = 317)
22% N/A, hospital does not track fixed
and variable costs (1)a
34% Account analysis (3)
40% Subjective/managerial experience (2)4% Statistical regression techniques (4)
a Number indicates how responses were coded; 4 = most
advanced, 1 = least advanced.
Towhat extent does the cost accounting function
use the ratio of cost to charge (RCC) to estimate the
cost of individual services and procedures? 1 = use
only RCC, 7 = do not use RCC at all: N = 318,
mean = 4.1, median = 4, St. Dev. = 2.1.
Frequent––frequency and breadth of cost-report
distribution
How often does the cost accounting system re-
port information to each of the following groups?
Variance––number and type of variances are
calculated
Which of the following variances does your orga-
nization calculate? (Check all that apply.) 48%: effi-
ciency, 53%: case mix, 58%: price, (N = 317) (coded
as 0, 1).
Perceptual performance measures
Relevant––This measure was based upon the
two sets of questions below (1 and 2). Responses
to the second set of questions (2. Actual infor-
mation supplied) were subtracted from the corre-
sponding responses to the first set of questions
(1. User Needs). The average of these differences
captures the extent to which a system meets theuser�s needs. The smaller the difference, the bet-
ter the system meets the user�s needs. To calcu-
late the variable relevant, the average difference
for each hospital was subtracted from the
maximum difference so that larger values indi-
cate that a system is better at meeting a user�sneed.
1. User Needs: Rate the importance of costaccounting information in performing each of
the following activities. You may not actually have
the information necessary to perform the activities
listed below; however, in responding, please indi-
cate whether you would consider cost information
![Page 28: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/28.jpg)
N Mean Median St.Dev.
206 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
of value in performing these activities. 1 = not at
all important, 7 = very important
N Mean Median St.
Dev.
Evaluate individual
performance
322 4.4 5 1.6
Evaluate group
(e.g., departmental)
performance
324 5.6 6 1.3
Negotiate contracts
with payers
323 6.2 7 1.1
Budget 324 5.8 6 1.2Control costs 324 6.2 6 1.0
Price hospital services 323 5.7 6 1.3
Benchmark your
hospital�s costsagainst industry
323 5.6 6 1.2
Evaluate acquisitions
of new technology/
equipment
324 5.2 5 1.3
Evaluate strategic
alternatives
324 5.3 6 1.4
Negotiate contracts
with suppliers
(e.g. Pharm. Co.)
324 5.1 5 1.4
Budget 323 4.9 5 1.6Control costs 323 4.9 5 1.6Price hospital
services324 4.9 5 1.7
Benchmark yourhospital�scosts againstindustry
321 4.3 5 1.8
Evaluateacquisitionsof newtechnology/equipment
322 4.2 4 1.8
Evaluate strategicalternatives
321 4.1 4 1.8
Negotiate contractswith suppliers(e.g. Pharm. Co.)
321 4.0 4 1.8
2. Actual information supplied by system: Can
the cost accounting system provide your organiza-
tion with information for the following activities.
1 = not at all, 7 = completely.
N Mean Median St.Dev.
Evaluate
individual
performance
322 3.2 3 2.0
Evaluate group
(e.g., depart-mental)
performance
324 4.8 5 1.7
Negotiate
contracts
with payers
324 4.9 6 1.8
Use––Use in decision-making: Rate agreement
with the following statements: 1 = strongly agree,
7 = strongly disagree. Users rely on cost account-
ing data to make decisions. N = 324, mean = 4.5,
median = 5, St. Dev. = 1.5.
Construct validity measures––(Table 4)
Differs from Medicare: To what extent is your
organization�s cost accounting system different
from the Medicare cost-reporting system? 1 = ex-
actly the same, 7 = completely different. N =319,
mean = 4.35, median = 4, St. Dev. = 2.1.
Allocation bases: In allocating costs in revenue
departments to a specific procedure, how many
separate bases does your hospital use? (Circleone; circle N/A if your organization does not allo-
cate costs at the procedure level.)
No. of responses N/A 1 2–3 4–6 >6
309 114 27 60 47 61
Accurate: Rate agreement with the following
statements: 1 = strongly agree, 7 = strongly dis-
agree. The cost accounting system provides accu-
rate data. N = 321, mean = 3.5, median = 4, St.
Dev. = 1.4.
![Page 29: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/29.jpg)
N Mean Median St. Dev.
Business
process
innovation
(e.g., new
billing system)
321 4.1 4 1.6
Patient-caretechnology
(e.g., advanced
operating room)
323 4.3 4 1.7
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 207
Timely: Users receive information from the
cost accounting function in a timely manner.
1 = strongly agree, 7 = strongly disagree. N = 316,
mean = 3.9, median = 4, St. Dev. = 1.6.
Control variable––(Table 6)
Managed: Percentage of revenues derived from
managed care––(control variable): Indicate your
approximate payer mix by assigning the percent-
age of total inpatient revenue that is received from
the following: N = 316.
46% Medicare; 19% Managed care
13% Medicaid; 7% Self-pay/bad debt/charity
12% Indemnity; 4% Other
Strategy constructs––used to model cost-system
functionality (Table 9)
Cost––Extent to which the hospital follows a
cost-focused strategy: Rate the importance of the
following objectives in achieving your hospital�smission: 1 = not at all important, 7 = very important.
N Mean Median St.
Dev.
Control or reduce
the costs of existing
services
321 6.4 7 0.9
Identify and eliminate
unprofitable services
320 4.9 5 1.4
Lead––Extent to which hospital follows a prod-
uct differentiation strategy: Rate the importance
of the following objectives in achieving your hospi-
tal�s mission: 1 = not at all important, 7 = very
important.
N Mean Median St. Dev.
Establishclinical
leadership
322 5.8 6 1.3
Is your organization a leader or follower with
respect to the time of adoption of the following:1 = leader, 7 = follower.
References
Abernethy, M. A., & Chua, W. F. (1996). A field study of
control system ‘‘redesign’’: The impact of institutional
process on strategic choice. Contemporary Accounting
Research, 13, 596–606.
Abernethy, M. A., & Stoelwinder, J. Y. (1996). The role of
professional control in the management of complex
organizations. Accounting, Organizations and Society, 20,
1–17.
Abernethy, M. A., & Vagnoni, E. (2004). Power, organization
design and managerial behaviour. Accounting, Organiza-
tions and Society, 29, 207–225.
Babad, Y. M., & Balachandran, B. V. (1993). Cost driver
optimization in activity-based costing. The Accounting
Review, 68, 563–575.
Bastable, C. W., & Bao, D. H. (1988). The fiction of sales-mix
and sales-quantity variances. Accounting Horizons, 2, 10–17.
Banker, R. D., & Potter, G. (1993). Economic implications of
single cost driver systems. Journal of Management Account-
ing Research, 5, 15–31.
Beal, E., & Little, R. (1975). Missing values in multivariate
analysis. Journal of the Royal Statistical Society, Series B,
129–145.
Brickley, J. A., & Van Horne, R. L. (2002). Managerial
incentives in nonprofit organizations: Evidence from hospi-
tals. Journal of Law and Economics, 45, 227–249.
Bromwich, M., & Hong, D. (1999). Activity-based costing
systems and incremental costs. Management Accounting
Research, 10, 39–60.
Burda, D. (2002). Deja vu all over again? Joint Commission�shistory casts doubt on the sincerity of its newest makeover.
Modern Healthcare, 32, 22.
Callahan, C. M., & Gabriel, E. A. (1999). The differential
impact of accurate product cost information in imperfectly
competitive markets: A theoretical and empirical investiga-
tion. Contemporary Accounting Research, 15, 419–455.
Cheatham, C. B., & Cheatham, L. R. (1996). Redesigning cost
systems: Is standard costing obsolete?. Accounting Horizons,
10, 23–31.
Chenhall,R.K.(2003).Managementcontrolsystemsdesignwithin
its organizational context: Findings from contingency-based
![Page 30: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/30.jpg)
208 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
research and directions for the future. Accounting, Organi-
zations and Society, 28, 127–168.
Chenhall, R. H., & Langfield-Smith, K. (1998). The relation-
ship between strategic priorities, management techniques
and management accounting: An empirical investigation
using a systems approach. Accounting, Organizations and
Society, 23, 243–264.
Chenhall, R. H., & Morris, D. (1986). The impact of structure,
environment, and interdependence on the perceived useful-
ness of management accounting systems. The Accounting
Review, 61, 16–35.
Chilingerian, J., & Sherman, H. (1990). Managing physician
efficiency and effectiveness in providing hospital services.
Health Services Management Research, 3, 3–15.
Christensen, J., & Demski, J. S. (1997). Product costing in the
presence of endogenous subcost functions. Review of
Accounting Studies, 2, 65–87.
Chua, W. F., & Degeling, P. (1991). Information technology
and accounting in the accomplishment of public policy a
cautionary tale. Accounting, Management, and Information
Technology, 1, 109–137.
Comerford, S. E., & Abernethy, M. A. (1999). Budgeting and
the management of role conflict in hospitals. Behavioral
Research in Accounting, 11, 1–17.
Coombs, R. W. (1987). Accounting for the control of doctors:
Management information systems in hospitals. Accounting,
Organizations, and Society, 12, 389–404.
Cooper, D., Hayes, D., & Wolf, F. (1981). Accounting in
organized anarchies. Accounting, Organizations, and Soci-
ety, 6, 175–191.
Cooper, R., & Kaplan, R. (1991). The design of cost manage-
ment systems: Text, cases, and readings. Englewood Cliffs,
NJ: Prentice Hall.
Cooper, R., & Kaplan, R. (1992). Activity-based systems:
Measuring the costs of resource usage. Accounting Horizons,
6, 1–13.
Cooper, J., & Suver, J. (1994). Variance analysis refines overhead
cost control. Healthcare Financial Management, 48, 40–52.
Counte, M., & Glandon, G. (1988). Managerial innovation in
the hospital: An analysis of the diffusion of hospital cost-
accounting systems. Hospital and Health Services Adminis-
tration, 42, 371–384.
Covaleski, M. A., Dirsmith, M. W., & Michelman, J. E. (1993).
An institutional theory perspective on the DRG Frame-
work: Case-mix accounting systems health-care organiza-
tions. Accounting, Organizations and Society, 18, 65–80.
Covaleski, M. A., & Dirsmith, M. W. (1986). The budgetary
process of power and politics. Accounting, Organizations
and Society, 11, 193–214.
Datar, S., & Gupta, M. (1994). Aggregation, specification and
measurement errors in product costing. The Accounting
Review, 69, 567–591.
DeBrock, L., & Arnould, R. J. (1992). Utilization controls in
HMOs. Quarterly Review of Economics and Finance, 32,
31–53.
Dopuch, N. (1993). A perspective on cost drivers. The
Accounting Review, 68, 615–620.
Duggan, M. (2002). Hospital market structure and the behavior
of not-for-profit hospitals, specification and measurement
errors in product costing. Rand Journal of Economics, 33,
433–446.
Early, L., & Cleverly, W. (1995). CFO compensation increas-
ingly linked to performance. Healthcare Financial Manage-
ment, 49, 44–49.
Eldenburg, L. (1994). The use of information in total cost
management. The Accounting Review, 69, 96–121.
Eldenburg, L., Hermalin, B. E., Weisbach, M. S., & Wosinska,
M. (2004). Governance, performance objectives, and orga-
nizational form: Evidence from hospitals. Journal of Cor-
porate Finance, 10, 527–548.
Eldenburg, L., & Kallapur, S. (1997). Changes in hospital
service mix and cost allocations in response to changes in
Medicare reimbursement schemes. Journal of Accounting
and Economics, 23, 31–50.
Evans, J. H., III., Hwang, Y., & Nagarajan, N. J. (1995).
Physician�s response to length-of-stay profiling. Medical
Care, 33, 1106–1119.
Evans, J. H., III., Hwang, Y., & Nagarajan, N. J. (1997). Cost
reduction and process reengineering in hospitals. Journal of
Cost Management, 11, 10–27.
Evans, J. H., III., Hwang, Y., & Nagarajan, N. J. (2003).
Management conrol and hospital cost reduction: Additional
evidence. Journal of Accounting and Public Policy, 20, 73–88.
Fama, E. F., & Jensen, M. C. (1983a). Separation of ownership
and control. Journal of Law and Economics, 26, 301–325.
Fama, E. F., & Jensen, M. C. (1983b). Agency problems and
residual claimants. Journal of Law and Economics, 26,
327–349.
Feinglass, J., Martin, G., & Sen, A. (1991). The financial effect
of physician practice style on hospital resource use. Health
Services Research, 26, 183–205.
Feltham, G. A. (1977). Cost aggregation: An information
economic analysis. Journal of Accounting Research, 15,
42–70.
Feltham, G. A., & Xie, J. (1994). Performance measure
congruity and diversity in multi-task principal/agent rela-
tions. The Accounting Review, 69, 429–453.
Foster, G. F., & Swenson, D. (1997). Measuring the success of
activity-based cost management and its determinants. Jour-
nal of Managerial Accounting Research, 9, 109–141.
Fralicx, R. D., & Liccione, W. J. (1997). What Are You
Worth?. Hospital and Health Networks, 69(17), 8.
Gal-Or, E. (1987). First mover disadvantages with private
information. The Review of Economic Studies, 54, 279–292.
Gal-Or, E. (1998). The advantages of imprecise information.
Rand Journal of Economics, 19, 266–275.
Gal-Or, E. (1993). Strategic cost allocation. Journal of Indus-
trial Economics, 41, 387–402.
Gaynor, M., & Gertler, P. (1995). Moral hazard and risk
spreading in partnerships. RAND Journal of Economics, 26,
591–613.
Gaynor, M., & Pauly, M. V. (1990). Competition and produc-
tive efficiency in partnerships: Evidence fromMedical Group
Practice. Journal of Political Economy, 98, 544–573.
![Page 31: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/31.jpg)
M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210 209
Gordon, L. A., & Narayanan, V. K. (1984). Management
accounting systems, perceived environmental uncertainty
and organization structure: An empirical investigation.
Accounting, Organizations, and Society, 9, 37–47.
HCIA, Inc. (1997). Financial and clinical benchmarking: The
strategic use of data. Baltimore.
Hill, N. T. (2001). Adoption of costing systems in U.S.
hospitals: An event history analysis 1980-1990. Journal of
Accounting and Public Policy, 19, 41–71.
Hill, N. T., & Johns, E. (1994). Adoption of costing systems by
U.S. hospitals. Hospital and Health Services Administration,
39, 521–537.
Hilton, R. (1979). The determinants of cost information value:
An illustrative analysis. Journal of Accounting Research, 17,
411–435.
Ittner, C. D., Lanen, W. N., & Larcker, D. F. (2002). The
association between activity-based costing and manufactur-
ing performance. Journal of Accounting Research, 40,
711–725.
Ittner, C. D., Larcker, D. F., & Pizzini, M. J. (2004).
Compensation contracts in professional service firms.
Working Paper. Wharton School, University of
Pennsylvania.
Johnson, H. (1992). It�s time to stop overselling activity-based
concepts. Management Accounting, 74, 26–35.
Johnson, H., & Kaplan, R. (1987). The rise and fall of
management accounting. Boston MA: Harvard Business
School Press Books.
Jones, C. S., & Dewing, I. P. (1997). The attitudes of NHS
clinicians and medical managers towards changes in
accounting controls. Financial Accountability and Manage-
ment, 13, 261–280.
Kaplan, R. S., & Norton, D. P. (1992). The balanced
scorecard––measures that drive performance. Harvard Busi-
ness Review, 70, 71–79.
Karmarkar, U. S., Lederer, P. J., & Zimmerman, J. L. (1990).
Choosing manufacturing production control and cost
accounting systems. In R. Kaplan (Ed.), Measures for
manufacturing excellence. Boston, MA: Harvard Business
School Press.
Khandwalla, P. (1972). The effect of different types of compe-
tition on the use of management controls. Journal of
Accounting Research, 10, 275–285.
Krumwiede, K. R. (1998). The implementataion stages of
activity-based costing and the impact of contextual organi-
zational factors. Journal of Management Accounting
Research, 10, 239–277.
Krishnan, R. (2001). Market restructuring and pricing in the
hospital industry. Journal of Health Economics, 20, 213–237.
Kurunmaki, L. (1999). Professional vs financial capital in the
field of health care – struggles for the redistribution of
power and control. Accounting, Organizations and Society,
24, 95–124.
Lambert, R. A., & Larcker, D. F. (1995). The prospective
payment system, hospital efficiency, and compensation
contracts for senior-level hospital administrators. Journal
of Accounting and Public Policy, 14, 1–31.
Lee, R. H. (1990). Monitoring physicians: a bargaining model
of medical group practice. Journal of Health Economics, 9,
463–481.
Lawrence, C. H. (1990). The effect of ownership structure and
accounting system type on hospital costs. Research in
Governmental and Nonprofit Accounting, 6, 35–60.
Langfield-Smith, K. (1997). Management control systems and
strategy: a critical review. Accounting, Organizations and
Society, 22, 207–232.
Leone, A. J., & Van Horne, R. L. (1999). Earnings management
in not-for-profit institutions: Evidence from hospitals.
Working Paper. Simon School, University of Rochester.
Luft, J., & Shields, M. J. (2003). Mapping management
accounting: Graphics and guidelines for theory-consistent
empirical research. Accounting, Organizations and Society,
28, 169–250.
Mak, Y., & Roush, M. J. (1996). Managing Activity Costs with
Flexible Budgeting and Variance Analysis. Accounting
Horizons, 10, 141–146.
McGown, A. S. (1998). Perceived benefits of ABCM imple-
mentation. Accounting Horizons, 12, 31–50.
Meeting, D. T., & Harvey, R. O. (1998). Strategic cost
accounting helps create a competitive edge. Healthcare
Financial Management, 52, 42–51.
Merchant, K. A., & Shields, M. D. (1993). When and why to
measure costs less accurately to improve decision-making.
Accounting Horizons, 7, 76–81.
Molinari, C., Hendryx, M., & Goodstein, J. (1997). The effect
of CEO-board relations on hospital performance. Health
Care Management Review, 22, 7–15.
Nelson, E. G., Rust, R. T., Zahorik, A., & Rose, R. L. (1992).
Do patient perceptions of quality relate to hospital finan-
cial performance?. Journal of Health Care Marketing, 12,
6–13.
Newhouse, J. P. (1970). Toward a theory of nonprofit institu-
tions: An economic model of a hospital. The American
Economic Review, 60, 64–74.
Noreen, E. (1991). Conditions under which activity-based cost
systems provide relevant costs. Journal of Management
Accounting Research, 3, 159–168.
Norton, E. C., & Staiger, D. O. (1994). How hospital ownership
affects access to care for the uninsured. Journal of Health
Economics, 7, 259–284.
Orloff, T., Littell, C., Clune, C., Klingman, D., & Preston, B.
(1990). Hospital cost accounting: Who�s doing what and
why. Health Care Management Review, 15, 73–78.
Ouchi, W. G. (1979). A conceptual framework for the design of
organizational control mechanisms. Management Science,
79, 833–848.
Pauly, M., & Redisch, M. (1973). The not-for-profit hospital as
a physician�s cooperative. The American Economic Review,
63, 87–99.
Phelps, C. E. (2003). Health economics (3rd ed.). New York:
Addison Wesley.
Picone, G., Chou, S. Y., & Sloan, F. (2002). Are for-profit
hospital conversions harmful to patients and to medicare?
Rand Journal of Economics, 33, 507–523.
![Page 32: The-relation-between-cost-system-design,-managers'-evaluations-of-the-relevance-and-usefulness-of-cost-data,-and-financial-performance-an-empirical-study-of-US-hospitals_2006_Accounting,-Organizations-and-Society](https://reader030.fdocuments.us/reader030/viewer/2022032513/55cfeb3b5503467d968bdb0d/html5/thumbnails/32.jpg)
210 M.J. Pizzini / Accounting, Organizations and Society 31 (2006) 179–210
Porter, M. E. (1980). Competitive strategy. New York: The Free
Press.
Porter, M. E. (1985). Competitive advantage. New York: The
Free Press.
Preston, A. M. (1992). The birth of clinical accounting: A study
of the emergence and transformations of discourses on costs
and practices of accounting in U.S. hospitals. Accounting,
Organizations, and Society, 17, 63–100.
Prince, T. R. (1998). Strategic management for health care
entities: Creative frameworks for financial and operational
analysis. Chicago: American Hospital Publishing Inc.
Schick, A. G., Gordon, L. A., & Haka, S. (1990). Information
overload: A temporal approach. Accounting, Organizations,
and Society, 5, 199–220.
Serb, C. (1997). Data rich, information poor. Hospital and
Health Networks 5 May, 43–44.
Shank, J., & Govindarajan, V. (1993). Strategic cost manage-
ment: The new tool for competitive advantage. New York:
The Free Press.
Shapiro, S. S., & Wilk, M. B. (1965). An analysis of variance
test for normality (complete samples). Biometrica, 52,
591–611.
Shields, M. D. (1995). An empirical analysis of firms� imple-
mentation experience with activity-based costing. Journal of
Management Accounting Research, 7, 1–28.
Shortell, S. M., Levin, D. Z., O�Brien, J. L., & Hughes, E. F. X.
(1995). Assessing the evidence on CQI: Is the glass half
empty or half full?. Journal of Hospital and Health Services
Administration, 40, 4–24.
Simons, R. (1987). Accounting control system and business
strategy: And empirical analysis. Accounting, Organizations,
and Society, 12, 357–374.
Swenson, D. (1995). The benefits of activity-based cost man-
agement to the manufacturing industry. Journal of Man-
agement Accounting Research, 7, 167–180.
Tselepis, J. (1989). Refined cost accounting produces better
information. Healthcare Financial Management, 43, 27–34.
Young, D., & Pearlman, L. (1993). Managing the stages of hos-
pital cost accounting. Healthcare Financial Management, 47,
58–80.
Selto, F. H., Renner, C. J., & Young, S. M. (1995). Assessing
the organizational fit of a just-in-time manufacturing
system: Testing selection, interaction and systems models
of contingency theory. Accounting Organizations and Soci-
ety, 20, 665–684.
Wedig, G., Hassan, M., & Sloan, F. (1989). Hospital invest-
ment decisions and the cost of capital. Journal of Business,
62, 517–536.
West, T. D., & Balas, E. A. (1996). Contrasting RCC, RVU,
and ABC for managed care decisions. Healthcare Financial
Management, 50, 54–61.
White, H. (1980). A heteroskedasticity-consistent covariance
matrix estimator and a direct test for heteroskedasticity.
Econometrica, 48, 817–838.