Knowledge Transfer in Project Reviews
Knowledge Transfer in Project Reviews
Knowledge Transfer in Project Reviews
The first factor of interest to our study is the degree of involvement that the manager has with the
project under review, such as responsibility for initiating the project and/or length of active
management of the project. It is not uncommon that managers, who are involved in managing the
project under review, are also involved in the project review process (e.g. Kennedy and Mills,
1992; OBrien, 1998). Involving these managers can potentially bring in higher levels of project
specific knowledge compared to independent reviewers. However, we propose that there are
potential negative consequences as self-justification bias will cause the managers to place less
emphasis on, or even ignore negative project feedback (e.g. Caldwell and OReilly, 1982;
Aronson, 1995; Ryan, 1995); hence, reducing the effectiveness of project review as a knowledge
transfer tool.
Second, we are interested in the extent to which explicit performance evaluation attached to
performance reviews also affect managers willingness to transfer knowledge. We argue that
where project reviews are used to evaluate managerial performance, a moral hazard problem
exists, as managers (acting as agents) are given incentives not to reveal unfavorable private
information. Consistent with prior literature, which found that moral hazard is the major driving
force of escalation of commitment (e.g. Harrison and Harrell, 1993, 1994; Salter et al., 2004), we
propose that moral hazard will affect managers decision to share unfavorable information
regardless of the degree of prior involvement. We further argue that the degree of personal
responsibility and moral hazard have an interactive effect as the motivation to present a more
favorable outcome for evaluation purposes further adds to the cognitive bias of self-justification.
Using a laboratory experiment, we find support for the main effects of self-justification bias and
moral hazard on the managers willingness to share information via the project review process.
Both higher degree of personal involvement by the managers and the use of performance review
information in performance evaluation separately and significantly decrease managers
willingness to share information via the review process. This suggests that agency theory does
not completely explain managers unwillingness to share negative project feedback. Rather, the
managers self-justification bias also plays an important role. Our results have important
practical implications, suggesting that obstacles to knowledge transfer via project reviews can be
reduced by either severing the link between the project review process and the organizations
performance evaluation system, and/or the rotation of managers (e.g. via a job rotation program
where a new manager steps in and takes charge at some point during the life of the project).
However, the latter approach potentially leads to a reduction in local knowledge, which might
have a negative impact on project performance, especially if the rotation occurs too frequently.
Furthermore, it is also difficult to rotate other personnel closely associated with the project. An
alternative approach is to lower a managers feeling of personal responsibility by sharing the
project responsibility among a group of team members.
The results from the present study also contribute to the management accounting research
literature in a number of ways. First, prior literature investigating the effect of personal
responsibility on managerial decision bias has predominately focused on either agency theory
(e.g. Harrison and Harrell, 1993; Harrell and Harrison, 1994) or the effect of self-justification
bias (e.g. Staw, 1976; Cheng et al., 2003). Although Schulz and Cheng (2002) attempt to
integrate the two frameworks in the context of capital budgeting re-investment decisions, they
fail to find a significant moderating role for information asymmetry when managers who are
managers involved in the project review is also (i) the manager who initiated and managed the
project, and (ii) there is a direct link between the performance evaluation of the manager and
information elicited during the post-audit review.
2.1. Eliciting private negative information from the project manager to the organization: the
effect of personal involvement
The rationale for including managers actively involved with the project under review in the post
review process is based on these managers having private project information (e.g. information
that provides explanations for cost overruns and performance of various components of the
project). Once this information has been released by the manager, it can be pooled and shared
with other managers, and eventually be embedded within the organizations knowledge base as
part of the organizational learning process.
However, one impediment to sharing the information is likely to be the degree of the project
managers involvement with the project. Specifically, managers who have greater involvement
with the project (e.g. by initiating the project or having an active role in managing of the project)
are argued to have a greater perceived responsibility for the project. And higher personal
responsibility has been associated with a number of dysfunctional managerial behaviour, such as
the continuation of unprofitable projects in the escalation of commitment literature (e.g. Schulz
and Cheng, 2002).
In the context of information sharing, we postulate that higher project involvement makes project
managers more susceptible to the self-justification bias (also known as sponsorship bias),
which has been documented in both the accounting and psychology literature. Based on the more
general theory of cognitive dissonance (e.g. Festinger, 1957), prior studies have postulated that
managers who experience a higher level of personal responsibility for a project are more likely to
self-justify the existence and continuation of the project, as well as to seek out confirming rather
than disconfirming information about the project (Staw, 1976; Aronson, 1995).
Specifically, the theory of cognitive dissonance suggests that negative feedback (dissonant
feedback) about a prior decision generates cognitive dissonance. One way for managers to
reduce such dissonance is to ignore or re-interpret the dissonant feedback, and continue or even
increase their decision commitment (Festinger, 1957). A number of empirical studies in the
escalation literature support this view. For example, Beeler and Hunton (1997) and Conlon and
Parks (1987) both find that decision responsibility and the need for justification resulted in a
preference for retrospective (and, hence, supportive) information. Similarly, Ryan (1995) reports
that decision-makers sometimes re-interpreted critical events in a more positive light to support
their prior commitment. Biyalogorski et al. (2006) recently found support for a decision
responsibility effect on escalation behavior with the initial responsibility for the decision
resulting in significantly greater tendency to continue with a losing product introduction. The
framework used by Biyalogorski et al. (2006) is consistent with Festingers cognitive dissonance
theory.
Whereas prior literature has predominantly explored the consequence of self-justification in
terms of how these individuals seek out and use confirming or disconfirming information (e.g.
Conlon and Park, 1987; Keil, 1995; Ryan, 1995), in the present study we are interested in how it
manifests itself in willingness to share negative (and, hence, disconfirming) information. We
postulate
that
negative information creates cognitive dissonance with the initial decision to green light the
investment. Consistent with prior work into self-justification bias, we further expect that the
degree of dissonance between the negative information and the initial investment decision will
be greater for managers who were involved in making the initial decision and, therefore, perceive
a higher level of personal responsibility for the initial investment decision (Festinger, 1957;
Staw, 1976). We hypothesize that the outcome of this increased level of cognitive dissonance for
these managers will be a greater tendency to regard the negative information as less important to
both the initial decision and subsequent decisions and as such a lower likelihood of including this
information
in
performance
reports.
Stated
more
formally:
H1: Managers who perceive a higher level of personal responsibility for the project under
review are less likely to report negative project feedback, compared to those who perceive a
lower level of personal responsibility for the project.
2.2. Eliciting private negative information from the project manager to the organization: the
effect of performance evaluation
Information obtained during the post-review can also be used by the organization to jointly
evaluate the performance of the manager as well as the project. Although the primary aim of
project reviews is to analyze the performance (and factors contributing to the performance) of the
project, the review process can also be used to evaluate the performance of the project manager
with respect to his or her efficiency and effectiveness in implementing the project. Indeed, as a
project managers performance is at least partially reflected in the success of the projects he or
she manages, from the organizations viewpoint the project review process is a useful way to
gain insight into the project managers competence. For example, one of the aims of project
review is to gather root causes for problems and then to make recommendations on how such
problems can be avoided in the future (e.g. Von Zedtwitz, 2002). If the root causes identified
reveal mismanagement on the part of the project manager, it will affect top managements
perception of the competencies of the project manager.
Using project reviews to evaluate a project managers performance gives rise to the moral hazard
problem under agency theory. Agency theory states that a moral hazard problem exists when the
goals of the principal and the agent conflict (incentive to shirk), and when it is difficult or
expensive for the principal to verify the agents behavior (information asymmetry; Eisenhardt,
1989). When both incentive to shirk and information asymmetry occur, then agency theory
predicts that the agent will exploit any information advantage to maximize his or her own
welfare at the expense of the principal (e.g. by withholding private information; Sprinkle, 2003).
A number of prior studies has consistently shown that the problem of moral hazard exists in
capital investment/resource allocation decisions (e.g. Harrison and Harrell, 1993, 1994; Rutledge
and Karim, 1999; Booth and Schulz, 2004). In particular, Harrison and Harrell (1993) find that
managers who are responsible for a poorly performing project are more likely to continue with
an unprofitable project where both information asymmetry and financial/career incentives are
present. By continuing these failing projects, managers can delay the release of private negative
information (i.e. the project is failing) and, therefore, advance their personal interests (i.e.
financial incentives), which would otherwise be threatened by the release of this private
information.
In the context of project reviews, information asymmetry is likely to already exist between the
project manager and the review committee. Indeed, the existence of information asymmetry
gives rise to the need to use the project review as an information sharing tool. Where the review
is used solely for performance evaluation purposes, the project manager is provided with an
incentive not to include and, hence, reveal negative information, as doing so would negatively
affect their reported performance and, hence, the attached financial incentives. In contrast, where
the purpose of the review is solely for organizational learning, revealing negative information
will not affect their reported performance, as the organization is explicitly trying to learn not
only from prior successes but also from prior mistakes. In fact, in such situations the sharing of
and critical reflection on such negative information would be viewed as reflecting positively on
the project manager.
As a consequence, we postulate that in the presence of information asymmetry, the intended use
of the information obtained from the review has important consequences to the sharing of
negative information by managers. Where the primary intention by the organization is to use it as
a performance appraisal as opposed to an organizational learning tool, moral hazard facing the
managers will be higher and, hence, managers are less likely to share negative information in the
context of the former than the latter. Stated more formally:
H2: Managers who are involved in a project review as part of a performance evaluation process
are less likely to report negative project feedback, compared to managers who are involved in a
project review as part of an organizational learning process.
2.3. The interaction between personal involvement and the purpose of project review
It is unclear from the prior literature whether the negative effect of personal responsibility and
using the project review to evaluate performance are additive or whether the two factors interact.
Prior empirical studies in the accounting literature provide mixed evidence (Harrell and
Harrison, 1994; Schulz and Cheng, 2002). In particular, Harrell and Harrison (1994) find a moral
hazard effect in the context of managers being told that they were responsible for the project and,
hence, controlled for self-justification effects by pegging their study to a context of responsible
managers. Yet the subsequent study by Schulz and Cheng (2002) demonstrated that Harrell and
Harrison (1994) might have inadvertently weakened the level of responsibility successfully
created by studies, such as Staw (1976) (refer to Schulz and Cheng, 2002, for a discussion).
However, Schulz and Cheng (2002) fail to find a separate moral hazard effect, which can
potentially be attributed to a weaker moral hazard manipulation. 2 As such, it remains unclear
whether incentives have an independent addition effect to the level of responsibility.
The psychological literature also points to different predictions. On the one hand, the effect of
self-justification biases and moral hazard might be additive as the former is a cognitive bias
related to how managers process information, whereas the latter provides managers with a
motivational force not to share negative information in order to maximize their personal utility.
To the extent that these are separate effects, we would not expect an interaction between the level
of personal responsibility and the purpose of the project review.
In contrast, personal responsibility and project review purposes might interact. Prior literature on
motivated reasoning suggests that individuals cognitive processes are affected by their
motivation. In particular, peoples desire to arrive at a favorable outcome can lead to biases, such
as selective memory search and underestimation of the importance of unfavorable events (e.g.
Kunda, 1990; Browstein, 2003). Indeed, a high level of motivation to achieve a certain outcome
is said to provide a trigger for the operation of a set of cognitive biases that allow individuals to
arrive at a conclusion they prefer (Kunda, 1990). In the current context, the use of project
reviews to evaluate managerial performance is likely to increase managers motivation to draw
favorable conclusion about the projects outcome. As such, it would add to the devaluating of the
importance of negative information expected from the self-justification bias discussed previously
and, therefore, result in a greater reluctance by these managers to share negative project
information. On balance we postulate the latter and, hence, propose an interaction effect. Stated
more formally:
H3 : The relative difference in tendency to report negative project feedback between managers
perceiving a high level of personal responsibility for the project and managers perceiving a low
level of personal responsibility will be greater when the intended use of the project feedback is
performance evaluation of the managers than when the intended use is organizational learning.
3. Research method
3.1. Research design and subjects
To test the proposed hypotheses, we conducted an experiment with a 2 2 between subjects
design. The dependent variable was the degree to which two negative project feedback items
were included in the review.3 Experimental subjects were 129 middle managers who were
undertaking an MBA at a major Australian university. All subjects were enrolled in the same
management accounting subject, which was part of the core of the MBA program. All subjects
were volunteers and were given a random chance of winning a bookshop coupon. Their rewards
were not linked to their responses in the experiment. The average age of the subjects was 30
years, and the average work experience was 8 years. As mature subjects with real-world business
and management experience, the subjects were considered to have the appropriate background
for the decision-making task used in this experiment.
3.2. Experimental task
The experiment consisted of two stages: an initial investment decision followed by the
completion of a project review report. Subjects assumed the role of a national marketing
manager for a fast food company and were randomly assigned to one of the four experimental
groups. In the First Stage (initial investment decision 3 years prior to the current decision
period), subjects were either asked to make a decision to invest in one of two available projects
(high personal responsibility) or were told that their predecessor had made the initial investment
decision (low personal responsibility). The act of making the initial investment decision is
expected to increase the feeling of volition and, therefore, personal responsibility in the subjects
(Schulz and Cheng, 2002), as well as indicating to these subjects that they have been responsible
for the project for a long time (3 years). 4 This manipulation of personal responsibility was
consistent with previous studies in the self-justification literature (e.g. Staw, 1976; Ghosh, 1997;
Beeler, 1998; Schulz and Cheng, 2002).
The projects involved an intensive marketing campaign for a fast food product. All subjects were
given information relating to the two products, such as the products previous sales and earnings
figures, and a probability distribution of expected internal rate of return over the next 3 years. 5
This manipulation of personal responsibility was consistent with previous studies in the selfjustification literature, which found that decision choice could induce high personal
responsibility (e.g. Staw, 1976; Ghosh, 1997; Beeler, 1998; Schulz and Cheng, 2002). After all
subjects had made (or were given) the initial investment decision, they were given further
information indicating that 3 years after the initial investment decision, the chosen investment
project was not performing as well as expected. For example, subjects were told that the average
internal rate of return of the project was 14 per cent, which was below the company hurdle rate
of 15 per cent and the original expected return of 20.25 per cent. Furthermore, prospective
information was provided indicating that the expected internal rate of return for the next 2 years
(16 per cent) would be below those of an alternative investment project (19 per cent).6
The Second Stage of the experiment involved asking subjects to complete a project review report
by indicating the likelihood that they would include a list of information statements (representing
major factors that have contributed to the projects performance) in the review report. Subjects
were also told the purpose of the project review, either for reviewing the project itself in order to
facilitate organizational learning (the organizational learning condition), or to evaluate
managerial performance (the performance evaluation condition). Consistent with previous
studies that have examined the agency effect on decision biases (e.g. Harrell and Harrison,
1994), subjects in the performance evaluation group were given a substantial incentive to shirk
(to receive a favorable performance evaluation and a substantial year-end performance bonus).7
Subjects were then asked to indicate whether they would include each information statement in
the review report. Four information statements were provided: two negative feedback statements
and two positive feedback statements. Positive feedback refers to feedback information that had
a profit-increasing impact on the project and was consistent with the initial investment decision.
Negative feedback refers to feedback information which had a profit-decreasing impact on the
project and was inconsistent with the initial investment decision. Although our research interest
lies in managers bias against negative project feedback, positive feedback statements were also
incorporated in the instrument to provide a more balanced view of the project, and to avoid
leading subjects in their decisions. A summary of the two negative feedback statements
provided to subjects is shown in Table 1.8 After all the experimental materials were collected a
post-test questionnaire was administered, including demographic questions and manipulation
questions (discussed later).
TABLE 1
Furthermore, consistent with earlier research studies that have manipulated the personal
responsibility variable (e.g. Schulz and Cheng, 2002), two additional questions were included to
assess subjects perception of personal responsibility. Subjects were asked to indicate (on 7-point
scales) To what extent did you feel that [the chosen project] was a reflection on yourself and
How responsible did you feel for [the chosen project]s performance. These two questions were
then summed to form a perceived personal responsibility scale (theoretical range 214), which
had an acceptable Cronbach alpha of 0.749. Analyses of responses from across the treatment
groups (not tabulated) confirmed that subjects in the high personal responsibility groups
perceived their responsibility as significantly higher (mean = 9.673) than subjects in the low
personal responsibility group (mean = 7.725, t = 3.550, p = 0.001). Furthermore, subjects in the
high responsibility treatment groups perceived their level of personal responsibility to be
significantly higher (t = 7.375, p = 0.000) than the mean of the scale (average of 7 on the
aggregation of two 7-point scales). In contrast, subjects in the low personal responsibility
treatment group perceived their level of personal responsibility is not statistically different (t =
1.075, p = 0.086) from the mean of the scale. These findings provide support for a difference in
the perceived level of personal responsibility between the high and low personal responsibility
treatments.
Finally, analyses were conducted on the demographic data collected in the post-test
questionnaires. No significant differences were found in the distribution of work experience (in
years) and gender across the treatments. However, there were some significant differences in the
distribution of Age. Further analysis showed that Age was not correlated with our dependent
variable; therefore, it is suggested that the Age variable is unlikely to explain the between-subject
differences of the dependent variable. Finally the distribution of mode of study (full time/part
time) differed across treatments. Further analysis showed that variation in mode of study was not
related to the decision to include negative information.
4. Results
4.1. Descriptive statistics
To analyze subjects tendencies to include/exclude different types of negative project feedback,
we summed the response scores for the two feedback items into one variable: willingness to
share negative feedback.10 The descriptive statistics (means and standard deviations) for the
treatment groups are presented in Table 2.
4.2. Hypothesis testing
Hypothesis 1 predicts that managers who perceive a higher level of personal responsibility for
the project under review are less likely to report negative project feedback than those managers
who perceive a lower level of personal responsibility for the project. As predicted, our results
showed that subjects in the low responsibility treatment were significantly more likely to report
negative project information (mean = 10.392) than their high personal responsibility counterparts
(mean = 8.436). The two-way analysis of variance shows a significant main effect for personal
responsibility (F = 14.531, p = 0.000; Table 3); hence, we support Hypothesis 1.
Hypothesis 2 investigates whether managers who perceive high levels of personal responsibility
for the project under review are less likely to report negative project feedback if the project
reviews are used for their performance evaluation than organizational learning. Table 2 shows
that the level of information sharing is lower (mean = 8.681) for managers whose project review
was used for performance evaluation purposes than for managers whose project review purpose
was organizational learning (mean = 10.200). Our results show a significant main effect for
project review purpose (F = 10.853, p = 0.000), and provides support for Hypothesis 2.
Finally, Hypothesis 3 predicts that the relative difference in tendency to report negative project
feedback between managers perceiving a high level of personal responsibility for the project and
managers perceiving a low level of personal responsibility will be greater when the intended use
of the project feedback is performance evaluation of the managers than when the when the
intended use is organizational learning. We do not find empirical support for this hypothesis (F =
0.613, p = 0.435).
TABEL 2 &3
5. Discussion and conclusion
It has been increasingly well recognized that knowledge is an important organizational resource,
and as such, management accountants have responsibilities to manage this resource at least as
well as other more tangible organizational resources, such as financial and physical assets (e.g.
Dess et al., 2004). The important role of management accountants in helping organization to gain
knowledge by learning from its past operations is reflected in their involvement in the project
review process. For example, Kennedy and Mills (1992) find that a majority of companies
involve their accounting staff in their project reviews.
Yet we have argued in this study that there are a number of potential obstacles to learning from
project reviews based on the managers willingness or lack of willingness to share past private
information. Consistent with Chenhall and Morris (1993), this is likely to be the case where
environmental uncertainty provides managers with the opportunity to withhold private
information for longer periods of time. In our study, we explored two impediments to the
effectiveness of project reviews in the form of personal responsibility and the link between the
project review and the formal performance evaluation system of the organization. We
hypothesized two separate effects using self-justification theory for the former and agency theory
for the latter. Specifically, we stipulated both main effects for self-justification bias and the moral
hazard as well as an interaction effect for the two factors for managers unwillingness to report
negative project feedback during the project review process.
Our results lend support to both of our main effect hypotheses. Personal responsibility for the
project and the purpose of the project of the review had significant separate effects on the
managers willingness to incorporate negative project information in the project review.
Managers with higher personal responsibility for the initial project decision were significantly
less willing to disclose this type of information than managers with lower personal responsibility.
Likewise, where the performance evaluation was the purpose of the review, managers were
significantly less likely to disclose negative project information than when the purpose of the
review was organizational learning. We did not find empirical support for an interactive effect
between the level of personal responsibility and the purpose of the project review.
Our results show that the intended use of the review information is an important consideration as
managers exhibited information sharing impediments when the project review also forms part of
the managers performance evaluation process. Overall, our study further shows that where
sharing of negative information is of importance to the organization, the most effective project
review design features is the combination of a manager with low personal responsibility for the
project, and where the project review process is independent from the performance evaluation
process. However, the latter is not necessarily feasible in practice as project information is
unlikely to be forgotten by a project managers supervisor during performance evaluation (the
curse of knowledge). An alternative is to ensure that the responsibility for a project is shared
between managers so no one manager believes that he or she is totally responsible for project
performance. In these circumstances, personal responsibility is likely to remain low (even though
collective responsibility should be high).
The contributions of our study to the research literature are twofold. First, our study provides an
extension to a limited number of studies that examine project reviews, an important stage of the
capital budgeting process. In particular, we provide evidence that the personnel involved as well
as the purpose of the review have significant effects on managers willingness to include
negative information in project reviews. Our study demonstrates that an understanding of the
effects of accounting control on managerial behavior can be better achieved by considering both
psychological-based theories (e.g. self-justification theory) and economic-based theories (e.g.
agency theory). As several researchers have pointed out (e.g. Eisenhardt, 1989; Waller, 1995;
Sprinkle, 2003), a multiple theory approach, in particular combining perspectives from
psychological theories and economic-based agency theories, can provide a more comprehensive
understanding of managerial behavior.
Our results also have important implications for practitioners responsible for the project review
processes. Previous studies on project review practices has found that organizations often assign
the responsibilities of preparing project review reports to an accountant associated with the
project being reviewed, or to a small team of people drawn from those involved in the project
(e.g. Mills and Kennedy, 1993b; Farragher et al. 1999). Although this practice might be the
logical result of limited organizational resources, our results highlight the risks of biased
information sharing among project review personnel who were also personally responsible for
the project.
The usual limitations associated with laboratory studies apply to the current research. For
example, although we have endeavored to capture the essential elements of capital budgeting
decisions and project review processes, the experimental task represents a much-simplified
decision-making context. As such, it omits some of the complex and often political factors
involved in this kind of decisions. Furthermore, for the low personal responsibility treatment we
have only looked at the decision point where a new manager came into the project just before the
project review. The effect of personal responsibility might be different where the manager has
taken over the project earlier, such that they are partially responsible for some of the events
contributing to the project performance. Indeed, it might be more appropriate to view the
concepts of responsibility and involvement as a continuous scale. As project work is often
non-linear in nature, a person might be involved with some parts of the project (e.g. idea
generation or brought in to complete the project) but not all stages of the project. The degree of
personal responsibility might, therefore, vary depending on the tasks undertaken by this
individual and his or her tenure. In our study, we only manipulated personal responsibility as a
dichotomous scale and did not consider various factors that might influence the degree of
personal responsibility. Finally, we did not control for social desirability bias, such that
participants in our performance evaluation treatment might be unwilling to withhold negative
feedback items despite the negative financial implications associated with reporting such items.
However, this limitation works against rather for our hypotheses and, therefore, does not reduce
the validity of our findings.
The current research represents a preliminary step in investigating the potential impediments to
the effectiveness of project reviews. Future studies can look at other feedback characteristics that
might affect managers decision to share and transfer the information; for example, whether the
feedback is based on objective data or subjective judgment. In addition, the way managers
attribute the causes of project failure might also affect their knowledge sharing behavior. For
example, managers might be less willing to share information that is directly attributed to
personal mismanagement of the project, but more willing to report feedback that can be blamed
on a second party. Another potentially fruitful research avenue relates to the other design features
of project reviews, such as team-based project reviewers versus individuals, the timing of the
project review and the extent of information dissemination after the project review process and
the subsequent effects on both individual and organizational learning. Finally, consistent with
earlier studies such as Harrell and Harrison (1994), our study only examines the situation where
information asymmetry exists and is water-tight. However, in a real-world setting, there might
be varying degree of information asymmetry. Future research can investigate whether our
findings and indeed escalation behavior in general still exist if there is only a moderate level of
information asymmetry.