Description
The accounting literature has argued that ®rms overengage in outsourcing because they tend to ignore the transac-
tion costs involved in buying services from external suppliers. A ®eld experiment with managers of health care orga-
nizations shows that decision makers are actually quite sensitive to the asset speci®city associated with the ``buy''
option in an outsourcing decision. However, they also appear inappropriately sensitive to the sunk costs inherent in
most real-life outsourcing decisions, and may actually underengage in outsourcing. Prior commitment to internal pro-
curement systematically reduced the willingness to outsource, relative to a pure ``make or buy'' scenario.
On the role of sunk costs and asset speci®city in outsourcing
decisions: a research note
Filip Roodhooft *, Luk Warlop
Katholieke Universiteit Leuven, Department of Applied Economics, Naamsestraat 69, 3000 Leuven Belgium
Abstract
The accounting literature has argued that ®rms overengage in outsourcing because they tend to ignore the transac-
tion costs involved in buying services from external suppliers. A ®eld experiment with managers of health care orga-
nizations shows that decision makers are actually quite sensitive to the asset speci®city associated with the ``buy''
option in an outsourcing decision. However, they also appear inappropriately sensitive to the sunk costs inherent in
most real-life outsourcing decisions, and may actually underengage in outsourcing. Prior commitment to internal pro-
curement systematically reduced the willingness to outsource, relative to a pure ``make or buy'' scenario. # 1999
Elsevier Science Ltd. All rights reserved.
1. Introduction
Ecient ®rms allocate their own resources to
those activities within the value chain for which
they enjoy a comparative advantage over compe-
titors (Shank & Govindajaran, 1992). Other activ-
ities are increasingly being outsourced to external
suppliers. Outsourcing often involves important
production cost savings relative to internal pro-
duction because outside suppliers can aggregate
demand, which enables them to bene®t from
economies of scale, smoother production schedules
and centralization of expertise. Management
accountants play an important role in the decision
whether to ``make'' or ``buy'' an activity because it
requires an accurate analysis of the relevant costs
associated with both options.
Optimal choices between continued internal
production and a switch to market procurement of
an activity require more than mere consideration
of production cost di?erences. According to
transaction cost economics, the degree of asset
speci®city is an important consideration in the
outsourcing decision. Outsourcing is only desirable
when expected governance and coordination costs
resulting from asset speci®c investments in the
relationship with the future supplier are lower
than the production cost advantage that the supplier
may bring (Chalos, 1995). Because transaction costs
are often dicult to specify and estimate, many
authors in accounting warn against head-over-
heels commitment to outsourcing. Case studies
(Drtina, 1994; Lacity, Willcocks & Feeny 1996)
have suggested that decision makers within ®rms
Accounting, Organizations and Society 24 (1999) 363±369
0361-3682/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0361-3682(98)00069-5
* Corresponding author. E-mail: ®[email protected]
leuven.ac.be
overemphasize production cost advantages of
outsourcing and underestimate the role of trans-
action costs. However, little or no systematic
research has investigated the incorporation of
transaction costs in outsourcing decisions.
Even less documented is an intuitively impor-
tant factor that may discourage outsourcing when
it would be normatively appropriate. In practice,
outsourcing is not a make-or-buy decision, but
involves a switch from internal production to
external procurement. Although the past invest-
ments in internal production are sunk and should
be irrelevant, the act of giving up the correspond-
ing assets is very salient to the manager. If man-
agers are unable to ignore these sunk costs, they
may engage in outsourcing to a lesser extent than
would be normatively appropriate (Ghosh, 1995).
To our knowledge there have been no empirical
studies investigating the inherent sunk cost aspect
of outsourcing decisions.
``Good'' outsourcing decisions require that
decision makers are appropriately sensitive to
asset speci®c investments, and appropriately
insensitive to sunk costs. In this paper we experi-
mentally investigate the extent to which sophisti-
cated decision makers consider sunk costs and
asset speci®city while choosing between internal
production and outsourcing of a component of the
®rm's value chain. Section 2 introduces the trans-
action cost economics approach to vertical inte-
gration where anticipated asset speci®city
determines a ®rm's optimal degree of vertical
integration (Williamson, 1989). The third section
deals with the sunk cost bias and describes its
potential impact on the outsourcing decision. Sec-
tion 4 outlines our experimental design and
results. The last section o?ers a general discussion
and directions for future research.
2. Transaction cost economics approach to out-
sourcing
Accounting norms and standard economic the-
ory prescribe a forward looking perspective for
outsourcing decisions. The decision whether or
not to outsource an activity should be based on a
comparison of the available ``make'' and ``buy''
options on the basis of expected future cash ¯ows.
Within this general prescriptive framework transac-
tion cost economics emphasizes that transactions
should occur in the market only when this is more
ecient than internal production (Anderson &
Weitz, 1986; Lieberman, 1991; Lyons, 1995).
Speci®c assets are specialized to the exchange
between buyer and seller rather than being usable
for other purposes without losing value. For
example, if outsourcing part of the production
process requires the outsourcing ®rm to invest in
dedicated transportation equipment, this invest-
ment is asset speci®c if it can not be used for other
purposes. Investments that can be put to other use
without costs are not asset speci®c.
While current production cost advantages of
external procurement are relatively easy to specify
on the basis of given information, future costs
associated with asset speci®city are not. They are
due to potential opportunistic behavior of the
supplier. At the time of the decision, it is not
known whether the partner in the transaction will
behave opportunistically. These costs can be made
more tangible by designing and enforcing con-
tracts between the parties, but even then the asso-
ciated dollar costs are hard to specify. From a
transaction cost economics point of view, we can
generate the hypothesis that the intention to
outsource is lower in the presence of asset speci-
®city.
3. Sunk cost bias in outsourcing decisions
In all cases where future cash ¯ows favor the
external supplier, outsourcing would be optimal.
From a normative point of view, the mere fact
that the ®rm is currently ``making'' the activity for
which outsourcing is contemplated, should be
irrelevant. Outsourcing±or vertical de-integra-
tionÐdecisions are normatively equivalent to ver-
tical integration decisions. Any historical
investments in a current ``make'' activity are to be
treated as sunk costs. These costs were incurred in
the past, are not changed by today's alternative
actions, and should therefore be ignored. Only
future and relevant cash ¯ows should be taken
into account.
364 F. Roodhooft, L. Warlop / Accounting, Organizations and Society 24 (1999) 363±369
Research in psychology, however, demonstrates
that individual decision makers are not immune to
sunk cost biases (Arkes & Blumer, 1985). Sensi-
tivity to sunk costs often leads to perseverance or
even escalation of normatively inappropriate
courses of action. These e?ects have been shown
in tasks that are related to various business func-
tions (Bazerman, Beekun & Schoorman, 1982;
Drummond, 1994; Garland & Newport, 1991;
Staw, 1976, 1981; Staw & Ross, 1978). In each
case the mere existence of prior investment (in
money or in time) interferes with the consideration
and adoption of alternative courses of action, with
which the manager would be normatively better
o?.
We propose that mostÐif not allÐoutsourcing
decisions are threatened by this bias. While out-
sourcing decisions should only take into account
anticipated costs and revenues of the make and buy
options, current ``make'' activities have usually
been the result of considerable prior investments. If
a parallel with sunk cost biases in other business
decisions exists, we should observe more reluctance
to choose for outsourcing than for choosing the
same ``buy'' option when no prior investment has
taken place, even if relevant accounting informa-
tion would prescribe the choice for outsourcing in
either case. This means that we can generate the
hypothesis that the presence of a sunk historical
investment in the current ``make'' activity reduces
the likelihood of opting for outsourcing.
Research into the sunk cost aspect of vertical
integration decisions is rare, and to our knowledge
there are no empirical studies investigating its role
in outsourcing decisions. In one study that is related
to our concerns, Whyte (1994) found an in¯uence
of sunk costs in vertical integration decisions. In
his study decision makers assumed the role of the
general manager of a manufacturing company,
and had to evaluate the possibility of acquiring a
distributor of the ®rm's products. The key manip-
ulation was whether prior transaction speci®c
investments had been made by the manufacturer
into the relationship with the distributor. Whyte
(1994) demonstrated that such sunk costs con-
siderably increase the likelihood that the manu-
facturer decides to integrate vertically even if
normative analysis would favor the preservation
of the current governance structure. In Whyte's
study, sunk costs and asset speci®city coincided.
Our own research investigates exactly the opposite
problem. We study the impact of sunk cost on the
decision to de-integrate. In outsourcing, asset spe-
ci®city and sunk costs are dissociated: asset speci-
®city is forward looking and should be taken into
account; prior investments in internal production
are backward looking and should not.
Past research into sunk cost biases in manage-
rial decision making can be used to identify
potential reasons for the reluctance to engage in
outsourcing. First, it has been recognized that the
inclusion of sunk costs in a decision can result
from information asymmetry within an organiza-
tion where middle managers possess privately held
information and have an incentive to shirk (Harrell
& Harrisson, 1994; Harrisson & Harrell, 1993;
Kanodia, Bushman & Dickhout 1989). Out-
sourcing decisions may be postponed for similar
reasons: managers may have an incentive to with-
hold or distort information that would favor out-
sourcing, thereby threatening their own power
base within the organization.
Other research suggests a number of reasons for
postponing outsourcing that are based on the
individual psychology of the decision maker. One
common explanation is based on the fact that
outsourcing constitutes a discontinuation of pol-
icy. Managers who have been responsible for these
past ``make'' decisions may avoid outsourcing,
merely because it would create the appearance
that they are trying to correct for a prior mistake.
They would be reluctant to create such an
impression, either because they see it to as a threat
to their perceived competence by the other mem-
bers of the organization (Brockner, Rubin &
Lang, 1981; Fox & Staw, 1979), or even because it
would constitute a threat to their self esteem. Sup-
port for this latter hypothesis is found in studies
®nding a larger sunk cost bias when the decision
maker believes he was responsible for the past
investments (Bazerman, Guiliano & Appelman,
1984; Brockner, 1992; Chenhall & Morris, 1991;
Staw, 1976). Similar motivations may underlie the
reluctance to outsource when a decision maker has
been responsible for starting internal production
in a previous period. A di?erent motivational
F. Roodhooft, L. Warlop / Accounting, Organizations and Society 24 (1999) 363±369 365
explanation was provided by Arkes and Blumer
(1985). Some of their empirical results could not
be explained by the desire to appear consistent
with prior decisions. For example, participants in
one of their studies massively preferred to eat a
pizza for which they paid $5 rather than an iden-
tical pizza that had cost them only $3. The authors
formulated the hypothesis that many sunk cost
biases may be explained by the mere desire not to
be wasteful.
Our objective in this study is to test whether
``make or buy'' decisions are biased towards
internal production when the buy option is pre-
sented as a choice for outsourcing. We do this by
comparing preferences for make and buy options,
keeping relevant accounting information constant,
while experimentally manipulating whether asset
speci®city is present in the decision scenario, and
whether the current choice has been preceded by a
prior decision in favor of internal production.
4. Experiment
The experimental scenarios contained two
orthogonal experimental manipulations. A ®rst
manipulation involved the antecedents of out-
sourcing as suggested by transaction cost eco-
nomics. For half of the respondents, the scenario
speci®ed the requirement of asset speci®c invest-
ments associated with the outsourcing option.
Orthogonal to the asset speci®city manipulation
half of the participants were told that in-house
production would be a continuation of an existing
activity (sunk cost condition), while others were
said that is was a new activity (make or buy con-
dition).
The scenarios control for rational explanations
of the sunk cost bias. The decision maker in the
study was the only ``player''. Motivation to shirk
or social justi®cation were therefore excluded as
an explanation. Also, information about the deci-
sion alternatives was constructed such that out-
sourcing had unambiguous production cost
advantages. To the extent that only production costs
are taken into account, outsourcing would be the
rationally optimal choice under all circumstances
(cf. Northcraft & Wolfe, 1984). We carefully
avoided decision scenarios in which the sunk costs
would explicitly involve past investments in peo-
ple. Personnel decisions are extremely complex
and are determined by an interplay of eciency
considerations and norms about ethical responsi-
bility (Drummond, 1994).
4.1. Participants
One-hundred and ®fty-six managers of Belgian
hospitals and rest homes were presented with a
scenario in which they had to make a decision
regarding the outsourcing of patient catering to an
external catering company. In order to ensure a
high level of interest in the problem, we sent out a
questionnaire to managers who had already out-
sourced some activities, or who were actively con-
sidering the outsourcing of non-medical activities
such as catering and cleaning. This information
had been collected for other research purposes.
We received one hundred and three usable
answers. The average experience of the respon-
dents in the sector was 14.33 years (S.D.=8.45).
On average they had been employed by their cur-
rent organization for 10.17 years (S.D.=6.79).
4.2. Decision task and procedure
Respondents were randomly assigned to one of
four experimental scenarios involving a hypothe-
tical choice between internal production or out-
sourcing of patient meal preparation. The ®rst
scenario was a control condition in which neither
asset speci®city or sunk cost were an issue. The net
present value of cash ¯ows related to future inter-
nal preparation was BEF 75 000 000. An external
company proposed to prepare and sell the meals
for BEF 65 000 000 and to employ the current
kitchen personnel of the hospital. The quality of
the meals would remain the same. Warming-up
the externally produced meals on-site would
require an investment of BEF9 000 000. This
investment was not asset speci®c, since it could
also be used for warming up internally prepared
meals or meals of other external catering compa-
nies.
The second scenario was identical to the ®rst,
except that the investment for the warming up
366 F. Roodhooft, L. Warlop / Accounting, Organizations and Society 24 (1999) 363±369
process was made asset speci®c. The scenario
mentioned that in the event of a switch to another
caterer the BEF9 000 000 investment would be
lost.
In scenarios 3 and 4 we introduced sunk costs.
The managers were told that 2 years before the
current decision their organization invested in new
kitchen equipment for a total of BEF35 000 000.
This investment would be lost with a switch to
external production. No other use of the installa-
tion was possible and it could not be sold. Table 1
gives an overview of the di?erent scenarios in our
®rst study.
4.3. Dependent variable
After reading the scenario, the managers had
two alternatives to choose from. They could either
decide to produce the patient meals internally
(make option), or outsource the activity to an
external catering company (buy option).
4.4. Results
The binary choice data were analyzed by logistic
regression. The proportions of respondents opting
for outsourcing in each of the four experimental
conditions are represented in Table 2.
As expected, we found that, relative to the con-
trol scenario, both the anticipation of asset speci®c
investments and the presence of sunk costs
reduced the likelihood of outsourcing. The analy-
sis revealed main e?ects for both asset speci®city
(Wald X
2
(1)=7.13; p
The accounting literature has argued that ®rms overengage in outsourcing because they tend to ignore the transac-
tion costs involved in buying services from external suppliers. A ®eld experiment with managers of health care orga-
nizations shows that decision makers are actually quite sensitive to the asset speci®city associated with the ``buy''
option in an outsourcing decision. However, they also appear inappropriately sensitive to the sunk costs inherent in
most real-life outsourcing decisions, and may actually underengage in outsourcing. Prior commitment to internal pro-
curement systematically reduced the willingness to outsource, relative to a pure ``make or buy'' scenario.
On the role of sunk costs and asset speci®city in outsourcing
decisions: a research note
Filip Roodhooft *, Luk Warlop
Katholieke Universiteit Leuven, Department of Applied Economics, Naamsestraat 69, 3000 Leuven Belgium
Abstract
The accounting literature has argued that ®rms overengage in outsourcing because they tend to ignore the transac-
tion costs involved in buying services from external suppliers. A ®eld experiment with managers of health care orga-
nizations shows that decision makers are actually quite sensitive to the asset speci®city associated with the ``buy''
option in an outsourcing decision. However, they also appear inappropriately sensitive to the sunk costs inherent in
most real-life outsourcing decisions, and may actually underengage in outsourcing. Prior commitment to internal pro-
curement systematically reduced the willingness to outsource, relative to a pure ``make or buy'' scenario. # 1999
Elsevier Science Ltd. All rights reserved.
1. Introduction
Ecient ®rms allocate their own resources to
those activities within the value chain for which
they enjoy a comparative advantage over compe-
titors (Shank & Govindajaran, 1992). Other activ-
ities are increasingly being outsourced to external
suppliers. Outsourcing often involves important
production cost savings relative to internal pro-
duction because outside suppliers can aggregate
demand, which enables them to bene®t from
economies of scale, smoother production schedules
and centralization of expertise. Management
accountants play an important role in the decision
whether to ``make'' or ``buy'' an activity because it
requires an accurate analysis of the relevant costs
associated with both options.
Optimal choices between continued internal
production and a switch to market procurement of
an activity require more than mere consideration
of production cost di?erences. According to
transaction cost economics, the degree of asset
speci®city is an important consideration in the
outsourcing decision. Outsourcing is only desirable
when expected governance and coordination costs
resulting from asset speci®c investments in the
relationship with the future supplier are lower
than the production cost advantage that the supplier
may bring (Chalos, 1995). Because transaction costs
are often dicult to specify and estimate, many
authors in accounting warn against head-over-
heels commitment to outsourcing. Case studies
(Drtina, 1994; Lacity, Willcocks & Feeny 1996)
have suggested that decision makers within ®rms
Accounting, Organizations and Society 24 (1999) 363±369
0361-3682/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0361-3682(98)00069-5
* Corresponding author. E-mail: ®[email protected]
leuven.ac.be
overemphasize production cost advantages of
outsourcing and underestimate the role of trans-
action costs. However, little or no systematic
research has investigated the incorporation of
transaction costs in outsourcing decisions.
Even less documented is an intuitively impor-
tant factor that may discourage outsourcing when
it would be normatively appropriate. In practice,
outsourcing is not a make-or-buy decision, but
involves a switch from internal production to
external procurement. Although the past invest-
ments in internal production are sunk and should
be irrelevant, the act of giving up the correspond-
ing assets is very salient to the manager. If man-
agers are unable to ignore these sunk costs, they
may engage in outsourcing to a lesser extent than
would be normatively appropriate (Ghosh, 1995).
To our knowledge there have been no empirical
studies investigating the inherent sunk cost aspect
of outsourcing decisions.
``Good'' outsourcing decisions require that
decision makers are appropriately sensitive to
asset speci®c investments, and appropriately
insensitive to sunk costs. In this paper we experi-
mentally investigate the extent to which sophisti-
cated decision makers consider sunk costs and
asset speci®city while choosing between internal
production and outsourcing of a component of the
®rm's value chain. Section 2 introduces the trans-
action cost economics approach to vertical inte-
gration where anticipated asset speci®city
determines a ®rm's optimal degree of vertical
integration (Williamson, 1989). The third section
deals with the sunk cost bias and describes its
potential impact on the outsourcing decision. Sec-
tion 4 outlines our experimental design and
results. The last section o?ers a general discussion
and directions for future research.
2. Transaction cost economics approach to out-
sourcing
Accounting norms and standard economic the-
ory prescribe a forward looking perspective for
outsourcing decisions. The decision whether or
not to outsource an activity should be based on a
comparison of the available ``make'' and ``buy''
options on the basis of expected future cash ¯ows.
Within this general prescriptive framework transac-
tion cost economics emphasizes that transactions
should occur in the market only when this is more
ecient than internal production (Anderson &
Weitz, 1986; Lieberman, 1991; Lyons, 1995).
Speci®c assets are specialized to the exchange
between buyer and seller rather than being usable
for other purposes without losing value. For
example, if outsourcing part of the production
process requires the outsourcing ®rm to invest in
dedicated transportation equipment, this invest-
ment is asset speci®c if it can not be used for other
purposes. Investments that can be put to other use
without costs are not asset speci®c.
While current production cost advantages of
external procurement are relatively easy to specify
on the basis of given information, future costs
associated with asset speci®city are not. They are
due to potential opportunistic behavior of the
supplier. At the time of the decision, it is not
known whether the partner in the transaction will
behave opportunistically. These costs can be made
more tangible by designing and enforcing con-
tracts between the parties, but even then the asso-
ciated dollar costs are hard to specify. From a
transaction cost economics point of view, we can
generate the hypothesis that the intention to
outsource is lower in the presence of asset speci-
®city.
3. Sunk cost bias in outsourcing decisions
In all cases where future cash ¯ows favor the
external supplier, outsourcing would be optimal.
From a normative point of view, the mere fact
that the ®rm is currently ``making'' the activity for
which outsourcing is contemplated, should be
irrelevant. Outsourcing±or vertical de-integra-
tionÐdecisions are normatively equivalent to ver-
tical integration decisions. Any historical
investments in a current ``make'' activity are to be
treated as sunk costs. These costs were incurred in
the past, are not changed by today's alternative
actions, and should therefore be ignored. Only
future and relevant cash ¯ows should be taken
into account.
364 F. Roodhooft, L. Warlop / Accounting, Organizations and Society 24 (1999) 363±369
Research in psychology, however, demonstrates
that individual decision makers are not immune to
sunk cost biases (Arkes & Blumer, 1985). Sensi-
tivity to sunk costs often leads to perseverance or
even escalation of normatively inappropriate
courses of action. These e?ects have been shown
in tasks that are related to various business func-
tions (Bazerman, Beekun & Schoorman, 1982;
Drummond, 1994; Garland & Newport, 1991;
Staw, 1976, 1981; Staw & Ross, 1978). In each
case the mere existence of prior investment (in
money or in time) interferes with the consideration
and adoption of alternative courses of action, with
which the manager would be normatively better
o?.
We propose that mostÐif not allÐoutsourcing
decisions are threatened by this bias. While out-
sourcing decisions should only take into account
anticipated costs and revenues of the make and buy
options, current ``make'' activities have usually
been the result of considerable prior investments. If
a parallel with sunk cost biases in other business
decisions exists, we should observe more reluctance
to choose for outsourcing than for choosing the
same ``buy'' option when no prior investment has
taken place, even if relevant accounting informa-
tion would prescribe the choice for outsourcing in
either case. This means that we can generate the
hypothesis that the presence of a sunk historical
investment in the current ``make'' activity reduces
the likelihood of opting for outsourcing.
Research into the sunk cost aspect of vertical
integration decisions is rare, and to our knowledge
there are no empirical studies investigating its role
in outsourcing decisions. In one study that is related
to our concerns, Whyte (1994) found an in¯uence
of sunk costs in vertical integration decisions. In
his study decision makers assumed the role of the
general manager of a manufacturing company,
and had to evaluate the possibility of acquiring a
distributor of the ®rm's products. The key manip-
ulation was whether prior transaction speci®c
investments had been made by the manufacturer
into the relationship with the distributor. Whyte
(1994) demonstrated that such sunk costs con-
siderably increase the likelihood that the manu-
facturer decides to integrate vertically even if
normative analysis would favor the preservation
of the current governance structure. In Whyte's
study, sunk costs and asset speci®city coincided.
Our own research investigates exactly the opposite
problem. We study the impact of sunk cost on the
decision to de-integrate. In outsourcing, asset spe-
ci®city and sunk costs are dissociated: asset speci-
®city is forward looking and should be taken into
account; prior investments in internal production
are backward looking and should not.
Past research into sunk cost biases in manage-
rial decision making can be used to identify
potential reasons for the reluctance to engage in
outsourcing. First, it has been recognized that the
inclusion of sunk costs in a decision can result
from information asymmetry within an organiza-
tion where middle managers possess privately held
information and have an incentive to shirk (Harrell
& Harrisson, 1994; Harrisson & Harrell, 1993;
Kanodia, Bushman & Dickhout 1989). Out-
sourcing decisions may be postponed for similar
reasons: managers may have an incentive to with-
hold or distort information that would favor out-
sourcing, thereby threatening their own power
base within the organization.
Other research suggests a number of reasons for
postponing outsourcing that are based on the
individual psychology of the decision maker. One
common explanation is based on the fact that
outsourcing constitutes a discontinuation of pol-
icy. Managers who have been responsible for these
past ``make'' decisions may avoid outsourcing,
merely because it would create the appearance
that they are trying to correct for a prior mistake.
They would be reluctant to create such an
impression, either because they see it to as a threat
to their perceived competence by the other mem-
bers of the organization (Brockner, Rubin &
Lang, 1981; Fox & Staw, 1979), or even because it
would constitute a threat to their self esteem. Sup-
port for this latter hypothesis is found in studies
®nding a larger sunk cost bias when the decision
maker believes he was responsible for the past
investments (Bazerman, Guiliano & Appelman,
1984; Brockner, 1992; Chenhall & Morris, 1991;
Staw, 1976). Similar motivations may underlie the
reluctance to outsource when a decision maker has
been responsible for starting internal production
in a previous period. A di?erent motivational
F. Roodhooft, L. Warlop / Accounting, Organizations and Society 24 (1999) 363±369 365
explanation was provided by Arkes and Blumer
(1985). Some of their empirical results could not
be explained by the desire to appear consistent
with prior decisions. For example, participants in
one of their studies massively preferred to eat a
pizza for which they paid $5 rather than an iden-
tical pizza that had cost them only $3. The authors
formulated the hypothesis that many sunk cost
biases may be explained by the mere desire not to
be wasteful.
Our objective in this study is to test whether
``make or buy'' decisions are biased towards
internal production when the buy option is pre-
sented as a choice for outsourcing. We do this by
comparing preferences for make and buy options,
keeping relevant accounting information constant,
while experimentally manipulating whether asset
speci®city is present in the decision scenario, and
whether the current choice has been preceded by a
prior decision in favor of internal production.
4. Experiment
The experimental scenarios contained two
orthogonal experimental manipulations. A ®rst
manipulation involved the antecedents of out-
sourcing as suggested by transaction cost eco-
nomics. For half of the respondents, the scenario
speci®ed the requirement of asset speci®c invest-
ments associated with the outsourcing option.
Orthogonal to the asset speci®city manipulation
half of the participants were told that in-house
production would be a continuation of an existing
activity (sunk cost condition), while others were
said that is was a new activity (make or buy con-
dition).
The scenarios control for rational explanations
of the sunk cost bias. The decision maker in the
study was the only ``player''. Motivation to shirk
or social justi®cation were therefore excluded as
an explanation. Also, information about the deci-
sion alternatives was constructed such that out-
sourcing had unambiguous production cost
advantages. To the extent that only production costs
are taken into account, outsourcing would be the
rationally optimal choice under all circumstances
(cf. Northcraft & Wolfe, 1984). We carefully
avoided decision scenarios in which the sunk costs
would explicitly involve past investments in peo-
ple. Personnel decisions are extremely complex
and are determined by an interplay of eciency
considerations and norms about ethical responsi-
bility (Drummond, 1994).
4.1. Participants
One-hundred and ®fty-six managers of Belgian
hospitals and rest homes were presented with a
scenario in which they had to make a decision
regarding the outsourcing of patient catering to an
external catering company. In order to ensure a
high level of interest in the problem, we sent out a
questionnaire to managers who had already out-
sourced some activities, or who were actively con-
sidering the outsourcing of non-medical activities
such as catering and cleaning. This information
had been collected for other research purposes.
We received one hundred and three usable
answers. The average experience of the respon-
dents in the sector was 14.33 years (S.D.=8.45).
On average they had been employed by their cur-
rent organization for 10.17 years (S.D.=6.79).
4.2. Decision task and procedure
Respondents were randomly assigned to one of
four experimental scenarios involving a hypothe-
tical choice between internal production or out-
sourcing of patient meal preparation. The ®rst
scenario was a control condition in which neither
asset speci®city or sunk cost were an issue. The net
present value of cash ¯ows related to future inter-
nal preparation was BEF 75 000 000. An external
company proposed to prepare and sell the meals
for BEF 65 000 000 and to employ the current
kitchen personnel of the hospital. The quality of
the meals would remain the same. Warming-up
the externally produced meals on-site would
require an investment of BEF9 000 000. This
investment was not asset speci®c, since it could
also be used for warming up internally prepared
meals or meals of other external catering compa-
nies.
The second scenario was identical to the ®rst,
except that the investment for the warming up
366 F. Roodhooft, L. Warlop / Accounting, Organizations and Society 24 (1999) 363±369
process was made asset speci®c. The scenario
mentioned that in the event of a switch to another
caterer the BEF9 000 000 investment would be
lost.
In scenarios 3 and 4 we introduced sunk costs.
The managers were told that 2 years before the
current decision their organization invested in new
kitchen equipment for a total of BEF35 000 000.
This investment would be lost with a switch to
external production. No other use of the installa-
tion was possible and it could not be sold. Table 1
gives an overview of the di?erent scenarios in our
®rst study.
4.3. Dependent variable
After reading the scenario, the managers had
two alternatives to choose from. They could either
decide to produce the patient meals internally
(make option), or outsource the activity to an
external catering company (buy option).
4.4. Results
The binary choice data were analyzed by logistic
regression. The proportions of respondents opting
for outsourcing in each of the four experimental
conditions are represented in Table 2.
As expected, we found that, relative to the con-
trol scenario, both the anticipation of asset speci®c
investments and the presence of sunk costs
reduced the likelihood of outsourcing. The analy-
sis revealed main e?ects for both asset speci®city
(Wald X
2
(1)=7.13; p