Anda di halaman 1dari 18

13

BEHAVIORAL RESEARCH IN ACCOUNTING


Volume 20, Number 1, 2008
pp. 1329
DCF Techniques and Nonnancial
Measures in Capital Budgeting:
A Contingency Approach Analysis
Shimin Chen
The Hong Kong Polytechnic University
ABSTRACT: This study empirically examines capital budgeting methods. Based on 115
responses from a cross-sectional survey and two approaches to contingency t, this
study produces three basic ndings. First, both discounted cash ow (DCF) techniques
and nonnancial measures are widely used in capital budgeting. However, DCF tech-
niques are more important than nonnancial measures, and nonnancial measures ap-
pear to serve as a partial substitute when DCF analysis is less efcient. Second, DCF
techniques and nonnancial measures are not unconditionally appropriate. Although
the impact of rm strategy on capital budgeting methods is not supported by the
present results, the study shows that product standardization affects both capital bud-
geting methods, as hypothesized. Firms with high product standardization tend to
place more emphasis on DCF analysis, while rms with low standardization are more
likely to focus on nonnancial measures. Third, an appropriate t under contingency
theory between product standardization and the two capital budgeting methods is sig-
nicantly associated with a rms satisfaction with the capital budgeting process.
Keywords: capital budgeting; DCF techniques; nonnancial measures; contingency
theory.
INTRODUCTION
T
here have been many surveys of capital budgeting techniques over the past three
decades (e.g., Graham and Harvey 2001; Kim et al. 1986; Klammer et al. 1991; Pike
1996; Ryan and Ryan 2002; Schall et al. 1978). Although there has been wide
acceptance of discounted cash ow (DCF) methods, very little is known about the under-
lying reasons for practice variations in the use of capital budgeting techniques. In particular,
the extant literature reects two shortcomings. First, although it has been long recognized
that DCF methods are not always appropriate (Haka 1987; Myers 1984), few studies have
examined the circumstances in which the use of DCF methods may or may not be benecial.
Second, existing surveys have mostly focused on quantitative nancial measures (Ittner and
Larcker 2001). While both normative literature (Kaplan 1986; Myers 1984; Shank and
Govindarajan 1992) and case/ eld studies (Carr and Tomkins 1996; Miller and OLeary
1997) suggest the importance of nonnancial considerations, studies based on cross-
sectional surveys have generally ignored nonnancial measures. This is surprising, given
calls for integrating nonnancial measures into management accounting systems during the
1990s (Kaplan and Norton 1992; Vaivio 1999).
My thanks go to Steve Kaplan, the editor, and two anonymous reviewers for their valuable comments and sug-
gestions that have helped improve this paper substantially.
14 Chen
Behavioral Research in Accounting, 2008
This study attempts to address these two shortcomings. The study rst measures and
compares the degree of DCF analysis and nonnancial consideration in capital budgeting.
1
Then, based on contingency theory (Fisher 1998), the study examines the extent to which
the use of DCF techniques and nonnancial measures is associated with product standard-
ization and rm strategy. Based on 115 responses from a cross-sectional survey, the study
reports the following ndings. First, while both methods play a signicant role in capital
budgeting, DCF techniques are more important than nonnancial measures. It appears that
nonnancial consideration is used as a partial substitute when DCF analysis is less effective.
Second, the study shows that product standardization is associated with both DCF tech-
niques and nonnancial measures, as hypothesized. Firms with high product standardization
tend to place more emphasis on DCF analysis, while rms with low standardization are
more likely to focus on nonnancial measures. Third, an appropriate t, under contingency
theory, between product standardization and the two capital budgeting methods is signi-
cantly associated with a rms satisfaction with the capital budgeting process.
The study contributes to the literature by constructing and providing preliminary evi-
dence for a contingency-based theory for capital budgeting. The study extends prior capital
budgeting research in three respects. First, the study examines the degree of DCF usage as
opposed to documenting the use versus non-use of DCF methods in the literature (e.g.,
Kim et al. 1986; Klammer et al. 1991; Ryan and Ryan 2002). Second, this study measures
nonnancial considerations in capital budgeting and compares the use of nonnancial mea-
sures with DCF analysis. Third, the study employs two approaches to contingency t and
identies specic conditions under which the use of DCF techniques or nonnancial mea-
sures are more appropriate.
The remainder of the paper is structured as follows. The next section provides a review
of the relevant literature and development of hypotheses. The research methods and em-
pirical results are then presented and discussed. The paper concludes with a summary of
ndings and a discussion of the implications and limitations of the study.
LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT
DCF Techniques and Nonnancial Measures
Survey evidence documents a trend of increasing use of DCF methods, with more recent
studies reporting DCF usage by most, if not all, responding rms (e.g., Graham and Harvey
2001; Kim et al. 1986; Klammer et al. 1991; Pike 1996; Ryan and Ryan 2002; Schall et
al. 1978). However, in an evaluation of the capital budgeting literature, Mukherjee (1987,
37) noted that much of the past effort has been spent on the what and not enough on
the why of capital budgeting practices. One exception is the nding that larger rms tend
to make more extensive use of DCF techniques than smaller rms (e.g., Pike 1996; Graham
and Harvey 2001; Schall et al. 1978).
Researchers have long recognized that DCF techniques are not unconditionally appro-
priate (e.g., Haka 1987; Kaplan 1986; Myers 1984). Specically, the usefulness of DCF
methods depends on a rms ability to estimate DCF parameters including: (1) a projects
future cash ow; (2) the risk-adjusted discount rate; (3) the projects impact on cash ows
generated from other assets; and (4) the projects impact on future investment opportunities
(Myers 1984). The ability of managers to estimate these parameters is likely to vary de-
pending on the environment in which rms operate (Haka 1987; Myers 1984). Firms in
1
A typical capital budgeting survey compares different DCF methods, such as net present value and internal rate
of return, with so-called na ve methods, such as payback and accounting rate of return. This study combines
DCF methods into one category and compares this with nonnancial considerations.
DCF Techniques and Nonnancial Measures in Capital Budgeting 15
Behavioral Research in Accounting, 2008
stable environments with a limited number of products, simple cross-sectional relationships,
and few growth opportunities are likely to nd DCF techniques more useful because of
their ability to estimate the parameters. Alternatively, rms in complex and risky environ-
ments may nd DCF methods less useful, as a result of their difculty in estimating the
parameters. However, this type of reasoning has seldom been incorporated into studies to
explain the variation in DCF usage.
The use of nonnancial measures is often prescribed by scholars in accounting and
nance as the other recommended approach to capital budgeting. For example, Myers
(1984), in discussing the inability of DCF analysis to capture the benets of future growth
and exibility, concludes that rational managers would simply bring these considerations
into project evaluation, in addition to DCF analysis, as strategic factors dressed in non-
nancial clothes. Both Klammer (1993) and Shank and Govindarajan (1992) suggest that
strategic cost management be integrated into capital budgeting using methods such as value-
chain analysis, cost-driver analysis, and competitive-advantage analysis. Additionally, case
study evidence, although limited, offers support for the use of nonnancial considerations
in capital budgeting. Carr and Tomkins (1996), drawing on 51 case studies in U.K., U.S.,
and German companies, nd that successful companies are more likely to use nonnancial
strategic information in investment decisions. Miller and OLeary (1997) describe how
Caterpillar redesigned its capital budgeting mechanisms to include substantial nonnancial
considerations because of the inability of DCF analysis to incorporate the benets and costs
of integrating diverse assets. However, as pointed out by Ittner and Larcker (2001), the case
study evidence is less than conclusive due to the use of ambiguous measures and the lack
of statistical tests. Also, no study has yet examined the use of nonnancial measures relative
to DCF analysis. This study attempts to measure and compare the degree to which DCF
techniques and nonnancial measures are used in capital budgeting.
While the above literature review suggests the importance of both DCF analysis and
nonnancial considerations, there are reasons to believe that each approach plays a different
role in capital budgeting. According to nance theory, DCF analysis will lead to optimal
investment decisions as long as a rm is able to estimate DCF parameters accurately (Haka
1987; Myers 1984). Nonnancial considerations are recommended as an alternative when
the rm cannot adequately implement DCF analysis (Carr and Tomkins 1996; Kaplan 1986;
Klammer 1993; Myers 1984; and Shank and Govindarajan 1992). This suggests a more
important role for DCF analysis than for nonnancial considerations. In fact, recent surveys
have reported increasing acceptance of DCF analysis (e.g., Graham and Harvey 2001; Ryan
and Ryan 2002). However, most rms likely encounter some difculties in estimating DCF
parameters, which is consistent with anecdotal evidence pointing toward widespread use of
nonnancial considerations in capital budgeting (Burns and Walker 1997; Pike 1996). The
above discussion suggests a substitution effect (a negative relationship) between the two
capital budgeting methods. When managers have substantial condence in DCF analysis,
then it would have less of a need for nonnancial measures. In contrast, nonnancial mea-
sures are likely to become increasingly important in situations where managers lack con-
dence in DCF analysis (Carr and Tomkins 1996; Kaplan 1986; Myers 1984). These basic
expectations are summarized in the following hypotheses:
H1a: DCF methods are more important than nonnancial measures in capital
budgeting.
H1b: There is a negative relationship between DCF methods and nonnancial
measures.
16 Chen
Behavioral Research in Accounting, 2008
In addition to the two baseline hypotheses, this study further examines why there is a
variation in the use of DCF techniques and nonnancial measures among different com-
panies. Specically, based on contingency theory (Chenhall 2003; Fisher 1998), this study
proposes that product standardization and rm strategy affect the use of two capital bud-
geting methods.
Impact of Product Standardization and Firm Strategy
Contingency theory prescribes no universally applicable management accounting sys-
tems. The applicability of a specic system depends on a proper t between the system
and its environment. Researchers have applied this theory to various aspects of management
accounting systems (see reviews by Chenhall [2003] and Fisher [1998]). Particularly rele-
vant to this paper are the studies that examine determinants of nancial versus nonnancial
information usage (e.g., Chenhall and Langeld-Smith 1998; Chong and Chong 1997;
Govindarajan and Gupta 1985; Hoque and James 2000; Simons 1990). Although these
studies differ in specics, a central recurring theme is that nancial information plays a
more important role when a rm operates in a stable and predictable environment, while
nonnancial information is more important when the rm faces a dynamic and uncertain
environment. Haka (1987) employs a similar contingency theory framework to examine
DCF techniques. Based on 16 DCF rms and 16 matched non-DCF rms, the study pro-
vides some supporting evidence by correlating the effectiveness of DCF adoption, measured
as the difference in stock returns, with contingent variables including rm strategy, envi-
ronmental predictability and diversity, information systems, reward structure, and degree of
decentralization.
Product standardization measures the product dimension of company technology
(Brownell and Merchant 1990). Companies vary from low (one-of-a-kind) to high (com-
modities) standardization. High standardization implies an optimal input / output relationship
that is either known or can be learned through experience (Brownell and Merchant 1990).
In such an environment, investment projects are likely to have features that management
would be able to quantify accurately within the DCF framework. If DCF analysis can be
implemented with condence, consideration of nonnancial factors will be less important.
However, low standardization suggests product uniqueness, a complex manufacturing proc-
ess, and reliance on research and development. In this type of capital budgeting environ-
ment, management is more than likely to nd it difcult to estimate DCF parameters.
Moreover, one could argue that, under this type of environment, nonnancial measures are
relatively more effective at analyzing project benets related to new features, complex
processes and technology, and future opportunities (Kaplan 1986; Klammer 1993; Myers
1984; Shank and Govindarajan 1992). This leads to the following hypothesis:
H2a: The higher the level of a rms product standardization, the more the rm
will emphasize (de-emphasize) DCF methods (nonnancial measures).
Firm strategy is dened according to the Defender versus Prospector typology of Miles
and Snow (1978). This typology has been frequently used in management accounting re-
search (Chong and Chong 1997; Haka 1987; Simons 1990).
2
A Defender operates in rel-
atively stable product markets, has a narrow product range, engages in little research and
2
Porters cost leadership versus product differentiation is the other frequently used measurement for strategy (e.g.,
Chenhall and Langeld-Smith 1998). While separately developed, there are similarities between a Defender and
a Cost Leader and between a Prospector and a Differentiation rm (Simons 1990).
DCF Techniques and Nonnancial Measures in Capital Budgeting 17
Behavioral Research in Accounting, 2008
market development, and competes primarily through cost leadership, quality, and service.
Previous research has shown that Defender rms tend to use more objective nancial mea-
sures (e.g., Govindarajan and Gupta 1985; Simons 1990). By the same logic, these rms
are likely to nd DCF analysis suitable as a result of their ability to estimate DCF param-
eters, and, thus, there is less need to consider nonnancial measures. In comparison, a
Prospector competes primarily through new products and markets, which requires intensive
and continuous efforts to search for new products and market opportunities. When making
decisions in a Prospector rm, nancial considerations have been found not to be as im-
portant as nonnancial factors such as maintaining product leadership and seeking future
opportunities (Govindarajan and Gupta 1985; Simons 1990). Consequently, it is expected
that the Prospector will nd DCF analysis less benecial and will emphasize nonnancial
measures in capital budgeting. The following hypothesis summarizes the expected relation-
ship between rm strategy and the two capital budgeting methods:
H2b: The closer a rms strategy toward a Defender, the more the rm will
emphasize (de-emphasize) DCF methods (nonnancial measures).
While hypotheses H2a and H2b are consistent with the selection approach to contin-
gency theory (Fisher 1998), this study also adopts the interaction approaches to examine
the impact of t between a capital budgeting variable and a contingent variable on orga-
nizational outcome.
3
Many contingency-based studies adopt this interaction perspective
(e.g., Abernethy and Brownell 1999; Brownell and Merchant 1990; Govindarajan and Gupta
1985; Haka 1987; Hoque and James 2000). The literature has contained two types of
outcome measures (satisfaction and performance), and the underlying logic is similar and
intuitive: an organizational choice such as a capital budgeting method should be more
successful if it ts the environment in which the rm operates, and this t should then
generate a positive impact on satisfaction on part of managers involved in implementing
this method and/ or on rm performance as a result of using this method. This study em-
ploys a satisfaction measure because of the difculty in obtaining a performance measure
that is directly related to capital budgeting. The following are the two sets of interaction
approach hypotheses:
H3a: An appropriate t of DCF methods or nonnancial measures with product
standardization will be associated with a higher level of satisfaction in the
capital budgeting process.
H3b: An appropriate t of DCF methods or nonnancial measures with rm
strategy will be associated with a higher level of satisfaction in the capital
budgeting process.
RESEARCH METHODS AND EMPIRICAL RESULTS
Survey and Sample
This study conducted a survey of capital budgeting techniques. The survey approach
provided a basis to gather information about internal accounting practices on a large-sample
basis. To enhance the level of condence in the survey results, two mechanisms were put
3
The impact of t on outcome can also be examined in terms of the system approach that emphasizes t among
multiple contingent variables and multiple system design attributes (Fisher 1998). However, there has been a
lack of consensus on how to dene an optimal subsample resulting from contingency t, which seems to prevent
an effective implementation of the system approach in this current study.
18 Chen
Behavioral Research in Accounting, 2008
in place. First, established survey instruments were adopted, as far as possible, for variable
measurement. Second, the questionnaire incorporated comments from several academic col-
leagues with expertise in survey research and those received from a eld test in two large
manufacturing rms. This eld test provided an assurance that the questionnaire was rea-
sonably understandable for respondents.
The sample consisted of about 600 publicly traded manufacturing rms in the U.S.,
identied from the Disclosure Database.
4
As in most capital budgeting surveys (e.g.,
Graham and Harvey 2001; Klammer et al. 1991; Pike 1996; Ryan and Ryan 2002), the
questionnaire was sent to the CFOs of sample rms in a personalized letter.
5
About a dozen
companies returned the questionnaire, citing their corporate policy of no survey participa-
tion. The survey resulted in 115 usable responses,
6
a response rate of approximately 20
percent based on the number of potential respondents. Although a 20 percent response rate
is not very high, it is commensurate with many capital budgeting surveys, including more
recent ones (e.g., Graham and Harvey 2001; Ryan and Ryan 2002).
Several tests were conducted to assess potential nonresponse biases. First, the respond-
ing and nonresponding rms were compared in terms of industry distribution, size, and
protability. The responding rms all came from manufacturing industries based on two-
digit SIC codes, except for those producing textile products (SIC code 22). No single
industry dominated the responding rms, and the largest industry group included 16 rms
producing industrial and commercial machinery and computer equipment products (SIC
code 35). A Chi-square test (p-value 0.8676) of homogeneity does not suggest any
signicant differences in industry distribution. The total assets of the responding rms
averaged about $5,450 million, which is signicantly higher than those of the nonrespond-
ing rms, according to the t-test (p-value 0.0178). Thus, relative to the mailing sample,
the sample used in this study was biased toward larger rms. The average return on assets
(ROA) of the responding rms was 6.62 percent, which is not signicantly different from
that of the nonresponding rms, based on the t-test (p-value 0.7959). Second, the 23
rms that responded three weeks after the survey were identied as late respondents
7
and
were compared with the 92 early respondents for all variables measured in this study, in
addition to size and protability. No signicant difference was found between the early and
late responding groups. Finally, the responding rms were divided into equal groups by
their median assets. The resulting groups were compared and two signicant differences
were observed. Consistent with previous studies (Graham and Harvey 2001; Pike 1996;
Schall et al. 1978), the larger responding rms placed more emphasis on DCF techniques
than did the smaller rms. Second, product standardization was higher for the group con-
taining the larger rms. Consequently, total assets were included in this study as a control
variable, to minimize the potential bias of size.
4
The sample was determined based on three criteria: (1) two-digit SIC Codes (2039) were used to select publicly
traded manufacturing rms; (2) all rms had been listed for at least ten years; and (3) mailing information,
including CFO name, must be available.
5
A relevant issue is whether a capital budgeting survey should be conducted at the corporate level to measure
overall practice, or at lower levels to capture diversied practices within the rm. Almost all published capital
budgeting surveys are conducted at the corporate level. One probable reason is that capital budgeting practice
may, indeed, be fairly uniform within a rm as a result of top-management involvement.
6
The survey was sent to the CFO of each rm, but about half of the respondents were CFOs, treasurers, or
controllers, and the other half were assistant treasurers or controllers, or managers of capital budgeting depart-
ments. Most of them had graduate-level education and had been in their positions for more than three years.
7
The respondents were asked to respond within two weeks of receiving the survey.
DCF Techniques and Nonnancial Measures in Capital Budgeting 19
Behavioral Research in Accounting, 2008
TABLE 1
Descriptive Statistics
(n 115)
Variable Mean Median S.D. Minimum Maximum
DCF 4.017 4.100 1.067 1.000 5.000
Nonnancial 3.616 4.000 1.156 1.000 5.000
Standardization 2.461 2.000 0.882 1.000 4.000
Strategy 2.921 2.920 0.461 1.670 4.420
Satisfaction 3.487 4.000 0.842 1.000 5.000
Specic measures for each variable are contained in the Appendix.
DCF average score of discounted cash ow techniques used in three categories of capital budgeting
projects;
Nonnancial average score of nonnancial measures used in three categories of capital budgeting projects;
Standardization product standardization, with a high score indicating high standardization;
Strategy company strategy, with a high (low) score indicating a Prospector (Defender) strategy; and
Satisfaction satisfaction with capital budgeting process.
Variable Measurement
Capital Budgeting Techniques
As shown in the Appendix, two questions were asked to measure the importance of
DCF techniques and nonnancial measures. The questions were repeated three times for
three types of investment projects: (1) equipment replacement; (2) expansion of existing
products; and (3) expansion into new products. Klammer et al. (1991) used similar classi-
cations in their multi-year capital budgeting surveys. To ensure consistent understanding,
the survey instrument described DCF methods as nancial-based techniques, including net
present value, internal rate of return, and the protability index, and nonnancial techniques
as considerations of rm strategy, growth potential, and competition in capital budgeting.
As reviewed in the previous section, the literature has called for integrating nonnancial
considerations into the capital budgeting process, but does not produce a standard denition
for what constitutes a nonnancial method. Instead, researchers often suggest taking several
common nonnancial aspects into consideration, including company strategy, growth po-
tential, and the impact of competition (e.g., Kaplan 1986; Klammer 1993; Myers 1984;
Shank and Govindarajan 1992). Consequently, these three aspects were explicitly cited in
the questionnaire to describe nonnancial techniques.
Participants responded using a ve-point scale anchored by 5 for extremely important
and 1 for not important. Since respondent ratings for the three investment categories
were highly correlated, a weighted average index was constructed for both DCF techniques
and nonnancial measures using the ratio of capital expenditure in each investment category
to total capital expenditure as a weighting factor.
8
Table 1 contains descriptive statistics for the DCF and the nonnancial index. The
actual scores achieved matched the theoretical minimum and maximum values (1 and 5).
8
The Cronbachs alpha values of 0.8656 for the DCF index and 0.8851 for the nonnancial index suggested a
high reliability for each variable. Further, a factor analysis of the three sub-ratings for DCF methods or non-
nancial measures resulted in a one-factor solution in each case. Both the DCF and the nonnancial factor scores
were highly correlated with the indexes (r 0.963 and r 0.897). Summated scores (indexes), rather than
factor scores, were used to facilitate interpretation of the survey results based on the original ve-point scale,
given the single factor solution for each index (Hair et al. 1998, 119120).
20 Chen
Behavioral Research in Accounting, 2008
The relatively high mean scores suggest widespread use of DCF techniques and nonnancial
measures.
Contingent Variables
The survey adopted existing instruments for both product standardization and rm
strategy. The respondents indicated the level of product standardization based on a four-
point, fully anchored single-item measure used by Brownell and Merchant (1990). A score
of 1 suggested the lowest level of product standardization (each unit custom-made) and 4
represented the highest product standardization (products completely standardized). The
theoretical and actual minimum and maximum values were found to be the same, and
the mean score of 2.461 is reported in Table 1.
To measure rm strategy, respondents were asked about the extent to which they agreed
with 12 statements concerning their rm strategy. These 12 statements described the un-
derlying characteristics of Prospector versus Defender, based on Miles and Snows (1978)
typology. The more strongly a rm agreed with these statements, the closer it was perceived
to position itself toward the Prospector end of the strategic continuum. Similar instruments
and/ or classications have been used in other accounting studies (Abernethy and Brownell
1999; Chong and Chong 1997; Haka 1987; Simons 1990). Since factor analysis resulted in
one common factor, an average of the 12 questions was computed as the measure for
strategy.
9
Although the theoretical range was from 1 to 5, the actual range was narrower,
from 1.67 to 4.42, as shown in Table 1.
While product standardization and rm strategy are two well-recognized contingent
variables in the literature (Chenhall 2003; Fisher 1998), they are not necessarily indepen-
dent. A signicantly negative correlation of 0.258 (p-value 0.005) between the two
variables is consistent with the notion that product standardization and rm strategy share
some common characteristics. However, such a relationship is unlikely to have posed a
serious threat in this study. Fisher (1998) suggests including multiple contingent variables
and addressing correlations among them.
Satisfaction
In a single question, respondents were asked directly to rate their rms overall satis-
faction with the capital budgeting process, from 1 (not satised) to 5 (extremely satised).
This approach was employed because of the lack of any established multiple-item measure
for this variable (Shields 1995). Similar one-item questions have also been used in other
studies to measure satisfaction with product-costing and performance measurement (Howell
et al. 1987; Swenson 1995). As shown in Table 1, the mean value of 3.487 and the standard
deviation of 0.842 suggest a sufcient level of variation in this variable.
Hypothesis Testing Results
Table 2 contains the results of testing the two baseline hypotheses (H1a and H1b) about
the use of DCF methods versus nonnancial measures. Panel A compares the mean values
of the two capital budgeting indexes. As predicted in H1a, both the parametric and the
nonparametric tests show that DCF techniques are signicantly more important than non-
nancial measures. Panel B shows a signicantly negative correlation between the two
capital budgeting methods, which provides supporting evidence for H1b. While both DCF
9
The correlation between the summated and factor scores was very high at 0.915. There were two additional
reasons for choosing the summated score. First, a summated score facilitates interpretation (Hair et al. 1998,
119120). Second, testing the interaction approach hypothesis becomes statistically simpler using the original,
rather than standardized, measurement scale (Hartmann and Moers 1999).
DCF Techniques and Nonnancial Measures in Capital Budgeting 21
Behavioral Research in Accounting, 2008
TABLE 2
Capital Budgeting Methods: Mean Comparison and Correlation
(n 115)
Panel A: Pairwise Comparison
Mean
Difference
t-test
t-(p-value)
a
Wilcoxon Rank Test
z-(p-value)
a
DCF versus Nonnancial 0.401 2.492
(0.014)
2.568
(0.010)
Panel B: Correlation
Pearson Spearman
DCF versus Nonnancial 0.203
(0.029)
a
0.186
(0.047)
a
a
Two-tailed signicance.
DCF average score of discounted cash ow techniques used in three categories of capital
budgeting projects;
Nonnancial average score of nonnancial measures used in three categories of capital budgeting
projects; and
Mean Difference the difference between the mean scores of DCF and Nonnancial in Table 1.
techniques and nonnancial measures are important for capital budgeting, there is a partial
substitution effect between the two, as hypothesized.
10
To examine variations in the use of DCF techniques and nonnancial measures, cor-
relation and regression analyses were conducted. As shown in Panel A of Table 3, rms
with high product standardization appear to make more use of DCF methods, but
rms with low product standardization place more focus on nonnancial measures. While
the relationship between product standardization and nonnancial measures is signicant
using both the Pearson and Spearman correlations (two-tailed p-values 0.05), the rela-
tionship between product standardization and DCF techniques is insignicant (two-tailed
p-values 0.10).
11
These results provide some support for H2a. However, the univariate
results do not support the hypothesized effect of rm strategy on capital budgeting methods,
as stated in H2b. While the Pearson correlation shows a signicant relationship between
strategy and DCF method, the coefcient does not possess the predicted sign.
Regression analysis was also conducted. Panel B of Table 3 contains two regressions
with the same independent variables, including product standardization, rm strategy, and
total assets, but separate dependent variables, DCF techniques or nonnancial measures.
Given the substitution effect between the dependent variables, rather than treating the two
regressions independently, they were estimated jointly using a general linear model pro-
cedure that takes interdependency in the dependent variables into consideration. Overall,
10
Although the negative correlation supports a substitution effect, the relatively small correlation coefcient sug-
gests a partial substitution, at best. An examination of the survey responses reveals that some rms rely on both
DCF analysis and nonnancial consideration to a large degree, indicating a complementary relationship. Future
research may examine why the two types of capital budgeting techniques are substitutive in some rms but
complementary in others.
11
While two-tailed p-values are reported throughout the tables, one-tailed p-values may be used to interpret the
results where predictions are directional. In this connection, some of the insignicant results become marginally
signicant.
22 Chen
Behavioral Research in Accounting, 2008
TABLE 3
Capital Budgeting Methods, Product Standardization, and Company Strategy:
Correlations and Regressions
(n 115)
Panel A: Correlations
Standardization
Pearson Spearman
Strategy
Pearson Spearman
DCF 0.145
(0.123)
a
0.141
(0.133)
0.167
(0.074)
0.092
(0.328)
Nonnancial 0.243
(0.009)
0.186
(0.046)
0.058
(0.536)
0.082
(0.385)
Panel B: Multivariate Regressions with DCF and Nonnancial as Dependent Variables
Independent
Variables
Multivariate Tests
Hotelling-t F-(p-value)
DCF as
Dependent Variable
t-(p-value)
a
Nonnancial as
Dependent Variable
t-(p-value)
a
Standardization 0.089 4.919
(0.009)
0.260 2.291
(0.024)
0.319 2.547
(0.012)
Strategy 0.048 2.615
(0.078)
0.491 2.268
(0.025)
0.013 0.056
(0.956)
Assets 0.035 1.912
(0.153)
0.000 1.901
(0.060)
0.000 0.137
(0.891)
Adjusted R
2
0.071 0.034
a
Two-tailed signicance.
DCF average score of discounted cash ow techniques used in three categories of capital budgeting
projects;
Nonnancial average score of nonnancial measures used in three categories of capital budgeting projects;
Standardization product standardization with a high score indicating high standardization; and
Strategy company strategy with a high (low) score indicating a Prospector (Defender) strategy.
controlling for size, the regression analysis in Panel B of Table 3 provides qualitatively
similar results as that in Panel A. The hypothesized effects of product standardization on
both DCF analysis and nonnancial considerations were supported according to both the
multivariate Hotellings t-test and individual t-tests in each equation. Although rm strategy
was signicant in the DCF regression, the sign of the coefcient was not in the predicted
direction as hypothesized in H2b. Consequently, the multivariate regressions provide sup-
port for H2a, but not for H2b.
To test the interaction approach hypotheses (H3a and H3b), Table 4 presents a mod-
erated regression analysis of whether the interaction term between one capital budgeting
variable and one contingent variable signicantly explains satisfaction with the capital bud-
geting process. This moderated regression analysis was originated by Southwood (1978),
recently reviewed by Hartmann and Moers (1999), and has been frequently applied in
contingency-based accounting studies (e.g., Abernethy and Brownell 1999; Brownell and
Merchant 1990; Govindarajan and Gupta 1985). Since there were two capital budgeting
and two contingent variables, Table 4 contains four separate regressions, with satisfaction
as the dependent variable. In addition, total assets are included in each regression as a
DCF Techniques and Nonnancial Measures in Capital Budgeting 23
Behavioral Research in Accounting, 2008
TABLE 4
Impact of Fit on Satisfaction: Moderated Regression Analysis
(n 115)
Independent
Variables
DCF
t-(p-value)
a
Independent Variables
Nonnancial
t-(p-value)
a
Panel A: Standardization
DCF 0.411 2.295
(0.024)
Nonnancial 0.327 1.421
(0.158)
Standardization 1.363 4.588
(0.000)
Standardization 0.465 1.575
(0.118)
Assets 0.000 1.046
(0.298)
Assets 0.000 1.159
(0.249)
DCF * Standardization 0.299 4.280
(0.000)
Nonnancial * Standardization 0.165 2.076
(0.040)
Adjusted R
2
0.256 Adjusted R
2
0.056
F-Test (p-value) 10.818
(0.000)
F-Test (p-value) 2.685
(0.035)
Panel B: Strategy
DCF 1.027 2.397
(0.018)
Nonnancial 0.278 0.649
(0.518)
Strategy 1.471 2.291
(0.024)
Strategy 0.241 0.411
(0.682)
Assets 0.000 0.858
(0.393)
Assets 0.000 1.321
(0.189)
DCF * Strategy 0.275 1.814
(0.072)
Nonnancial * Strategy 0.054 0.379
(0.706)
Adjusted R
2
0.172 Adjusted R
2
0.069
F-Test (p-value) 6.920
(0.000)
F-Test (p-value) 3.102
(0.018)
a
Two-tailed signicance.
DCF average score of discounted cash ow techniques used in three categories of
capital budgeting projects;
Nonnancial average score of nonnancial measures used in three categories of capital
budgeting projects;
Standardization product standardization with a high score indicating high standardization;
Strategy company strategy with a high (low) score indicating a Prospector (Defender)
strategy;
DCF * Standardization interaction between DCF and Standardization;
DCF * Strategy interaction between DCF and Strategy;
Nonnancial * Standardization interaction between Nonnancial and Standardization; and
Nonnancial * Strategy interaction between Nonnancial and Strategy.
control variable. To test the interaction approach hypotheses, we examined both the value
and the signicance of the interaction term in each equation.
12
12
Lower-order variables (capital budgeting and contingent variables) were included in each model, but the main
effects are not interpretable, and hypothesis testing depends on the interaction term due to the interval scale of
variable measurement (Hartmann and Moers 1999; Southwood 1978). Furthermore, multicollinearity does not
affect either the value or the signicance of the interaction term, and a t-test is equivalent to a partial F-test of
incremental R
2
, as a result of adding the interaction term.
24 Chen
Behavioral Research in Accounting, 2008
Panel A of Table 4 contains the results of the impact on satisfaction of t between
product standardization and each capital budgeting variable. The interaction terms are sig-
nicant, with the expected signs. It appears that when product standardization is high, an
emphasis on DCF methods leads to higher satisfaction with the capital budgeting process;
when standardization is low, an emphasis on nonnancial measures is associated with higher
satisfaction. These results provide support for H3a. However, results for strategy, as shown
in Panel B of Table 4, are mixed. While, as hypothesized, the marginally signicantly
negative interaction term in the DCF regression suggests that an emphasis on DCF methods
by Defenders (Prospectors) leads to higher (lower) satisfaction, there is no evidence to
support the impact of t between strategy and nonnancial measures. Therefore, H3b is
conrmed for DCF methods, but not for nonnancial measures.
Additional Analysis
While the selection and interaction approaches both supported the hypothesized rela-
tionships between product standardization and capital budgeting techniques, the study found
no evidence for the impact of rm strategy, except in the case of the interaction effect
between strategy and DCF methods. Based on subsamples, an additional analysis was car-
ried out in an attempt to reconcile the results between testing the selection and interaction
hypotheses. The analysis followed a simple logic: since the selection approach hypothesizes
a correlation between a contingent variable and a capital budgeting variable, and the inter-
action approach predicts a positive impact of this correlation on satisfaction, empirically,
we should then observe a stronger relationship between the contingent and capital budgeting
variables for higher than for lower satisfaction rms. To carry out this analysis, the 115
sample rms were divided into two groups: 65 high-satisfaction rms, with a satisfaction
score of 4 or 5, and 50 low-satisfaction rms, with a score below 4. The multivariate
regression models in Panel B of Table 3 were then estimated separately for each subsample.
As presented in Table 5, two interesting ndings emerged from this analysis. First, it
was found that the hypothesized relationships between product standardization and DCF
methods or nonnancial measures (H2a) exist only for the high-satisfaction subsample,
which is consistent with the moderated regression results in Table 4. Second, the strategy
variable was found to be insignicant in both the high- and the low-satisfaction samples,
thus eliminating any confusion caused by the separate application of the selection and
interaction approaches. In sum, this analysis provided additional evidence that product
standardization is related to both capital budgeting techniques, and that an appropriate t
between them leads to higher satisfaction with the capital budgeting process.
CONCLUSIONS AND DISCUSSIONS
This study provides preliminary evidence for a contingency-based theory for capital
budgeting. The study advances the literature in three ways. First, instead of determining
use or non-use of DCF techniques as in most capital budgeting surveys, this study examined
the degree of DCF usage. Second, this study measured and compared the use of nonnancial
considerations to DCF analysis. Although both play an important role across all rms, the
study found that DCF techniques are overall more important than nonnancial considera-
tions. Consistent with theoretical discussion and anecdotal evidence, nonnancial measures
appear to be used as a partial substitute for DCF analysis when DCF techniques are less
effective. Third, this study applied contingency theory as a basis to explain variations in
the use of both DCF techniques and nonnancial measures. While the results generally do
not support the rm strategy variable, the study found consistent evidence for the hypoth-
esized role of product standardization on the capital budgeting methods. It appears that
DCF Techniques and Nonnancial Measures in Capital Budgeting 25
Behavioral Research in Accounting, 2008
TABLE 5
Impact of Fit on Satisfaction: Additional Evidence Based on Subsamples
Independent
Variables
Multivariate Tests
Hotelling-t F-(p-value)
DCF as
Dependent Variable
t-(p-value)
a
Nonnancial as
Dependent Variable
t-(p-value)
a
Panel A: Multivariate Regressions for High-Satisfaction Firms (n 65)
Standardization 0.405 12.13
(0.000)
0.389 3.445
(0.001)
0.527 3.377
(0.001)
Strategy 0.024 0.729
(0.487)
0.265 1.199
(0.235)
0.086 0.280
(0.780)
Assets 0.047 1.423
(0.249)
0.000 1.698
(0.095)
0.000 0.003
(0.997)
Adjusted R
2
0.144 0.131
Panel B: Multivariate Regressions for Low-Satisfaction Firms (n 50)
Standardization 0.002 0.049
(0.952)
0.000 0.003
(0.998)
0.063 0.302
(0.764)
Strategy 0.026 0.593
(0.557)
0.403 1.078
(0.287)
0.043 0.110
(0.913)
Assets 0.090 2.036
(0.142)
0.000 1.762
(0.085)
0.000 0.452
(0.653)
Adjusted R
2
0.046 0.059
a
Two-tailed signicance.
The 65 high-satisfaction rms scored 4 or 5 in satisfaction with their capital budgeting process, while the 50
low-satisfaction rms reported a satisfaction score of 1, 2, or 3.
DCF average score of discounted cash ow techniques used in three categories of capital budgeting
projects;
Nonnancial average score of nonnancial measures used in three categories of capital budgeting projects;
Standardization product standardization, with a high score indicating high standardization;
Strategy company strategy, with a high (low) score indicating a Prospector (Defender) strategy; and
Satisfaction satisfaction with capital budgeting process.
rms with high product standardization tend to place more emphasis on DCF analysis,
while rms with low standardization tend to focus more on the use of nonnancial mea-
sures. Also, a stronger t between product standardization and the two capital budgeting
methods has a signicantly positive effect on satisfaction with the capital budgeting process.
The ndings of the study have implications for future research. This study shows that
both DCF techniques and nonnancial measures are widely used in practice and that their
use appears linked, in part, to contingent-based variables. These results suggest that future
capital budgeting studies should examine not only traditional DCF analysis but also non-
nancial considerations in relation to rms operating environment.
While this study presents new evidence about capital budgeting methods, several meth-
odological limitations should be noted when evaluating the ndings of this study, some of
which offer directions for future research. First, the relatively low response rate is still a
cause for concern, even though the nonresponse analysis did not reveal any signicant bias,
except for the effect of size. Second, measurement errors inherent in survey studies cannot
be ruled out. There can be no direct control over whether respondents interpret the survey
questions in the manner intended. The use of the single-item measurement for satisfaction
and the broad-based denition for nonnancial measures are a particular concern. Although
26 Chen
Behavioral Research in Accounting, 2008
the instrument for measuring product standardization was adopted from the literature, the
single-question measure makes the assessment of reliability impractical. Third, in addition
to the two contingent variables, other contextual factors may also affect DCF techniques
and nonnancial measures. Analyzing multiple contingent factors simultaneously, especially
those that make conicting demands on different capital budgeting techniques, will further
improve our understanding. Fourth, it is perplexing that the impact of the strategy variable
on DCF analysis was found to be contrary to the hypothesized effect. While the survey
instrument was based on the literature, the Miles and Snow typology (1978) may not fully
reect different aspects of competitive strategy. Given the importance of rm strategy in
contingency-based management accounting literature (Chenhall 2003; Fisher 1998), future
studies should pay particular attention to the relationship between strategy and capital bud-
geting techniques. Finally, a cross-sectional survey at the corporate level that queries man-
agers with nance or accounting backgrounds, such as CFOs, although common in the
literature, may not be sufcient to deal with heterogeneity in both investment projects and
operating environments. A more focused study of capital budgeting practice at lower levels
may produce additional insights.
APPENDIX
SURVEY QUESTIONS USED IN THIS STUDY
1. Techniques used in the project selection phase of capital budgeting are often re-
ferred to as either nancial or nonnancial. Within nancial-based techniques, net
present value (NPV), internal rate of return (IRR), and the protability index (PI)
are usually called discounted cash ow models. Nonnancial-based capital bud-
geting techniques include consideration of company strategy, growth potential, or
competition. A particular company may use only one of these techniques or may
combine them to provide information for decision making. For each of the follow-
ing three categories of investment, please circle the number indicating the relative
importance of each type of the techniques listed below. A technique is considered
extremely important if it is always relied upon by managers. If the technique is not
used or has no signicance in capital budgeting decisions, it is considered not
important.
Not Extremely
Important Important
For Replacement Projects:
Discounted Cash Flow models (NPV, IRR, or PI) . . . . . . . . . . . . . . 1 2 3 4 5
Nonnancial techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5
For ExpansionExisting Products:
Discounted Cash Flow models (NPV, IRR, or PI) . . . . . . . . . . . . . . 1 2 3 4 5
Nonnancial techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5
For ExpansionNew Products:
Discounted Cash Flow models (NPV, IRR, or PI) . . . . . . . . . . . . . . 1 2 3 4 5
Nonnancial techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5
DCF Techniques and Nonnancial Measures in Capital Budgeting 27
Behavioral Research in Accounting, 2008
Please estimate the average proportion of total capital expenditures you made in the last
ve years that should be classied within these three investment categories:
Replacement Projects
ExpansionExisting Products
ExpansionNew Products
Total 100%
2. Please circle the number that indicates your rms overall satisfaction with your
capital budgeting process.
Not Extremely
Satisfied Satisfied
The level of your satisfaction . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5
3. Please circle the number that characterizes your rms product standardization.
1 Each unit is custom-made.
2 Products differ but have common components.
3 Products are basically alike, with only minor differences, e.g., models, add-
on features.
4 Products are completely standardized.
4. Please circle the number that indicates your agreement with the following state-
ments about your companys strategy in comparison with those of your leading
competitors. Please note that none of the specic strategies is inherently good
or bad.
Strongly Strongly
Disagree Agree
You dene your business market very broadly . . . . . . . 1 2 3 4 5
Your customer base changes frequently . . . . . . . . . . . . . . 1 2 3 4 5
You monitor external environment very widely . . . . . . 1 2 3 4 5
Your product mix often changes . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5
You actively create changes in your customer base . . 1 2 3 4 5
You compete primarily on innovation rather than low
cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1 2 3 4 5
Your rm grows in spurts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5
You actively seek growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 3 4 5
You do not constrain the design of your products to
your production capacities . . . . . . . . . . . . . . . . . . . . . . . . . .
1 2 3 4 5
Your production employees skills are not
specialized . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1 2 3 4 5
You always seek exible instead of low-cost
production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1 2 3 4 5
Your R & D spending as a percentage of sales is
very high . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1 2 3 4 5
28 Chen
Behavioral Research in Accounting, 2008
REFERENCES
Abernethy, M. A., and P. Brownell. 1999. The role of budgets in organizations facing strategic change:
An exploratory study. Accounting, Organizations and Society 24: 189204.
Brownell, P., and K. A. Merchant. 1990. The budgetary and performance inuences of product stan-
dardization and manufacturing process automation. Journal of Accounting Research 28: 388
397.
Burns, R. M., and J. Walker. 1997. Capital budgeting techniques among the Fortune 500: A rational
approach. Managerial Finance 23: 315.
Carr, C., and C. Tomkins. 1996. Strategic investment decisions: The importance of SCM. A compar-
ative analysis of 51 case studies in U.K., U.S. and German companies. Management Accounting
Research 7: 199217.
Chenhall, R. H., and K. Langeld-Smith. 1998. The relationship between strategic priorities, man-
agement techniques and management accounting: An empirical investigation using a systems
approach. Accounting, Organizations and Society 23: 243264.
. 2003. Management control system designs within its organizational context: Findings from
contingency-based research and directions for the future. Accounting, Organizations and Society
28: 127168.
Chong, V. K., and K. M. Chong. 1997. Strategic choices, environmental uncertainty and SBU per-
formance: A note on the intervening role of management accounting systems. Accounting and
Business Research 27: 268276.
Fisher, J. 1998. Contingency theory, management control systems and rm outcomes: Past results and
future directions. Behavioral Research in Accounting 10 (Supplement): 4764.
Govindarajan, V., and A. K. Gupta. 1985. Linking control systems to business unit strategy: Impact
on performance. Accounting, Organizations and Society 10: 5166.
Graham, J. R., and C. R. Harvey. 2001. The theory and practice of corporate nance: Evidence from
the eld. Journal of Financial Economics 60: 187243.
Hair, J. F., R. E. Anderson, R. L. Tatham, and W. C. Black. 1998. Multivariate Data Analysis. 5th
edition. Upper Saddle River, NJ: Prentice Hall.
Haka, S. F. 1987. Capital budgeting techniques and rm specic contingencies: A correlational anal-
ysis. Accounting, Organizations and Society 12: 3148.
Hartmann, F. G. H., and F. Moers. 1999. Testing contingency hypothesis in budgetary research: An
evaluation of the use of moderated regression analysis. Accounting, Organizations and Society
24: 291315.
Hoque, Z., and W. James. 2000. Linking balanced scorecard measures to size and market factors:
Impact on organizational performance. Journal of Management Accounting Research 12: 117.
Howell, R. A., J. D. Brown, S. R. Soucy, and A. H. Sneed. 1987. Management Accounting in the
New Manufacturing Environment. Montvale, NJ: National Association of Accountants.
Ittner, C., and D. Larcker. 2001. Assessing empirical research managerial accounting: A value-based
management perspective. Journal of Accounting and Economics 32: 349410.
Kaplan, R. S. 1986. Must CIM be justied by faith alone? Harvard Business Review (MarchApril):
8794.
, and D. Norton. 1992. The balanced scorecardMeasures that drive performance. Harvard
Business Review (JanuaryFebruary): 7179.
Kim, S. H., T. Crick, and S. H. Kim. 1986. Do executives practice what academics preach? Manage-
ment Accounting November: 4952.
Klammer, T., B. Koch, and N. Wilner. 1991. Capital budgeting practices: A survey of corporate use.
Journal of Management Accounting Research Fall: 113131.
. 1993. Improving investment decisions. Management Accounting (July): 3543.
Miles, R. E., and C. C. Snow. 1978. Organizational Strategy, Structure, and Process. New York, NY:
McGraw-Hill.
Miller, P., and T. OLeary. 1997. Capital budgeting practices and complementarity relations in the
transition to modern manufacture: A eld-based analysis. Journal of Accounting Research 35:
257271.
DCF Techniques and Nonnancial Measures in Capital Budgeting 29
Behavioral Research in Accounting, 2008
Mukherjee, T. K. 1987. Capital budgeting surveys: The past and the future. Review of Business and
Economics Research 22: 3756.
Myers, S. C. 1984. Finance theory and nancial strategy. The Institute of Management Sciences 14:
126137.
Pike, R. 1996. A longitudinal survey on capital budgeting practices. Journal of Business Finance &
Accounting 23: 7992.
Ryan, P. A., and G. P. Ryan. 2002. Capital budgeting practices of the Fortune 1000: How have things
changed? Journal of Business and Economics 8: 355364.
Schall, L. D., G. L. Sundem, and W. R. Geijsbeck. 1978. Survey and analysis of capital budgeting
methods. Journal of Finance March: 281287.
Shank, J. K., and V. Govindarajan. 1992. Strategic cost analysis of technological investments. Sloan
Management Review Fall: 3951.
Shields, M. D. 1995. An empirical analysis of rms implementation experience with activity-based
costing. Journal of Management Accounting Research Fall: 148166.
Simons, R. 1990. The role of management control systems in creating competitive advantage: New
perspectives. Accounting, Organizations and Society 15: 127143.
Southwood, K. E. 1978. Substantive theory and statistical interaction: Five models. American Journal
of Sociology 83: 11541203.
Swenson, D. 1995. The benets of activity-based cost management to the manufacturing industry.
Journal of Management Accounting Research Fall: 167180.
Vaivio, J. 1999. Exploring a non-nancial management accounting change. Management Account-
ing Research 10: 409437.

Anda mungkin juga menyukai