Вы находитесь на странице: 1из 26

Volume 27 Number 4 2001 3

The Role of Cash Flow Information in Predicting


Corporate Failure: The State of the Literature
by Divesh S. Sharma, Senior Lecturer, School of Accounting, Banking & Finance, Fac-
ulty of Commerce & Management, Griffith University, Nathan, Queensland 4111, Aus-
tralia

Abstract

The role of cash flow information in predicting corporate failure remains a contentious is-
sue. The literature tends to report that cash flow information does not add value to accrual
failure prediction models. This position is further strengthened following brief and un-
critical literature reviews on the subject. The purpose of this paper is to provide a compre-
hensive review of the cash flow failure prediction literature since Beaver’s (1966)
pioneering paper. The review and subsequent critique highlight problems with the litera-
ture. The identified problems are offered as explanations for the position of the value of
cash flow information for predicting failure and not necessarily due to its potential infor-
mation content. Theoretical arguments justifying the relevance of cash flow information
for predicting failure are presented prior to discussing avenues for future research.

I. Introduction

The role of cash flow information in discriminating between bankrupt and non-bankrupt
companies remains a contentious issue. In a number of literature reviews on bankruptcy
prediction (e.g. Zavgren, 1983; Jones, 1987; Neill et al. 1991; Watson, 1996) the com-
mon view is that cash flow information does not contain significant incremental informa-
tion content over accrual information in discriminating between bankrupt and
non-bankrupt firms. This observation is surprising given the logical relationship between
cash flow and bankruptcy. However, the literature has not given justice to the issue by in-
adequately reviewing cash flow failure prediction studies. The literature reviews to date
have tended to make an almost passing comment only without delving into the problems
with prior research. Therefore, the first purpose of this paper is to inform the position of
cash flow information for predicting bankruptcy through a comprehensive review of the
literature since the pioneering work of Beaver (1966). In reviewing the literature, the re-
sults of one of the most cited study (Casey & Bartczak, 1985) are re-interpreted. Discus-
sion of problems and limitations of prior cash flow failure prediction research follow the
review. These problems and limitations are offered as explanations for the current state of
the literature. The second purpose of the paper is reflected in the theoretical argument that
cash flow information does contain potentially significant information content over ac-
crual information in discriminating between bankrupt and non-bankrupt firms, particu-
larly in the determination of the probability of bankruptcy. None of the prior literature
reviews nor studies on bankruptcy prediction have thoroughly discussed the merits of
cash flow information for predicting bankruptcy. The paper concludes with directions for
future research.
Managerial Finance 4

II. Review of Cash Flow Failure Prediction Studies

Review of Cash Flow Failure Prediction Studies

The relevance of cash flow information for predicting bankruptcy was highlighted by
Beaver (1966). Beaver (1966) reported that cash flow from operations (CFFO), proxied
by net income plus depreciation, depletion and amortisation, to total debt had the lowest
misclassification error relative to common accrual measures of financial health. How-
ever, his univariate approach to analysing financial distress was seldom followed because
while one ratio would indicate failure another could indicate non-failure. Altman (1968)
overcame this problem through the use of multiple discriminant analysis (MDA) that si-
multaneously considers financial ratio indicators of corporate health. Altman (1968) did
not examine the value of cash flow information though. Deakin (1972) demonstrated em-
ploying MDA that cash flow to total debt was a significant predictor up to three years
prior to failure. Like Beaver (1966), Deakin (1972) defined cash flow as net income plus
depreciation, depletion and amortisation. While other researchers (e.g. Blum, 1974; Nor-
ton and Smith, 1979; Mensah, 1983) have shown that cash flow information contains in-
formation content in predicting corporate bankruptcy, they are not frequently cited. The
rare citation is probably due to the measure of cash flow employed by these researchers.
These researchers used Beaver’s (1966) definition of CFFO. Table 1 summarises the re-
search findings of the cash flow failure prediction literature to date.

Largay and Stickney (1980) recognised the limitation of using net income plus de-
preciation, depletion and amortisation (NIDEP) as a measure of CFFO. They demon-
strated based on the infamous W.T. Grant Company bankruptcy that NIDEP more
correctly reflected working capital from operations. To determine CFFO one had to ad-
just for changes in current assets and current liabilities other than cash. Largay and Stick-
ney’s (1980) more refined measure of CFFO indicated that W.T. Grant’s cash from
operations was negative in eight of the ten years prior to failure while NIDEP was rela-
tively steady until the year immediately prior to its demise. This finding renewed interest
in CFFO as an indicator of corporate failure.

Lee (1982), a strong advocate of cash flow reporting, showed that the fall of Laker
Airways was foreseeable on a cash flow basis. His analysis of CFFO revealed that Laker
Airways was in financial trouble three years prior to failure while profits were increasing
as failure approached. A stream of multivariate modelling studies began investigating the
information content of this refined measure of CFFO. Using an estimation sample and a
validation sample comprising 60 bankrupt and 230 non-bankrupt firms and three meas-
ures of cash flow information, viz. CFFO, CFFO/Current Liabilities and CFFO/Total Li-
abilities, Casey and Bartczak (1984; 1985) concluded that none of the three operating
cash flow variables significantly improved the classification accuracy of the six accrual
ratios model. Casey and Bartczak (1985) reached this conclusion on the basis of the
number of firms correctly classified into their respective groups. While their conclusion
is valid, there are grounds to argue that cash flow information has significant information
content over accrual information in assessing the predicted probability of failure. Predict-
ing the probability of failure extends the mere classification into either the failed or non-
failed group and is practically more useful. For instance, the classification of a company
into a non-failed group does not provide information on the likelihood of this group mem-
bership. Perhaps the likelihood of the membership into this group is 51%. Knowing the
Table 1: Summary of Studies Investigating Cash Flow Information as Predictors

Study Method Cash Flow Variables Results

Beaver (1966) Univariate test of various accrual cashflow/total debt cashflow/total debt best predictor
(USA) and cash flow ratios. Matched 79 cashflow/total assets with a 13% misclassification error
failed with 79 non-failed firms cashflow/net worth one year prior to failure and 21%,
based on size and industry. cashflow/sales 23%, 24% and 25% for two to five
years prior.

Deakin (1972) Multiple discriminant analysis up to cashflow/total debt cashflow/total debt significant for
Volume 27 Number 4 2001

(USA) five years prior to failure of 32 years one, two and three prior to
failed and 32 matched on industry failure in estimation sample.
and size non-failed firms. Validation Validation showed similar results.
performed on 11 failed and 23
non-failed firms.

Blum (1974) 21 multivariate discriminant models cashflow/total debt cashflow/total debt significant
(USA) up to six years prior to failure. predictor.
Paired sample design by industry,
size, employees and fiscal year.
Validation sample used.

Norton & Smith (1979) Linear multiple stepwise cashflow/sales cashflow/total assets and
(USA) discriminant analysis for four years cashflow/total assets cashflow/total debt in best
prior to failure. Paired sample cashflow/net worth discriminant model three years prior
design based on size and industry. cashflow/total debt to failure.

Largay & Stickney (1980) Comparisons and trend analysis of cash flow from operations cash flow from operations more
(USA) cash flow from operations and other accurately indicated impending
accrual variables including stock failure up to 10 years prior to WT
price for single case study of W.T. Grant’s demise. Profitability and
Grant Company. stock prices began showing signs
only as early as two years prior to
failure.
5
Table 1: Summary of Studies Investigating Cash Flow Information as Predictors (Cont.)

Study Method Cash Flow Variables Results

Lee (1982) Compared profitability and operating cash flow from operations cash flow from operations indicated
(UK) cash flow over a five year period for a failure three years prior to failure but
single case study of Laker Airways profitability did not.
company.
Managerial Finance

Mensah (1983) Matched pair design up to five years cashflow/total liability cashflow/net worth most significant
(USA) prior to failure for 30 bankrupt and 30 cashflow/total assets variable in historical cost model.
non-bankrupt firms. Validation sample cashflow/sales Cashflow/net worth ranked second in
composed of 11 bankrupt and 35 cashflow/net worth specific price level model.
non-bankrupt firms. Used MDA and
logit. Means and standard deviations
used instead of raw ratios.

Taffler (1984) 24 failed (widely defined) and 49 cashflow/total liability cashflow/total liability was the second
(UK) non-failed companies of the most significant predictor in the model.
distribution/retail sector for one year Application to UK companies three
prior. No validation sample. Used years later showed good performance.
MDA and adjusted model for prior
probabilities and misclassification
costs.

Casey & Bartczak For 60 bankrupt and 230 non-bankrupt operating cash flow (OCF) defined as C&B concluded that none of the cash
(1984) firms up to five years prior to failure, working capital from operations flow ratios were strong indicators of
(USA) univariate analysis of three cash flow adjusted by non cash working capital overall classification of
and six accrual ratios. accounts, OCF/current liabilities and failure/non-failure. Results however
OCF/total liabilities. showed cash flow variables were better
than accrual variables in identifying
failures.
6
Table 1: Summary of Studies Investigating Cash Flow Information as Predictors (Cont.)

Study Method Cash Flow Variables Results

Casey & Bartczak (1985) For 60 bankrupt and 230 operating cash flow (OCF) defined cash flow ratios did not
(USA) non-bankrupt firms up to five as working capital from operations significantly increase the
years prior to failure, cash flow adjusted by non cash working predictive ability of the accrual
and accrual ratios were used to capital accounts, OCF/current MDA and logit models. On
classify the firms using MDA and liabilities and OCF/total liabilities. re-interpretation of their results,
logit. the cash flow variables
Volume 27 Number 4 2001

significantly increased explanatory


power and predicted probabilities
of failure/non-failure of the
accrual model.

Gentry et al (1985a) Logit analysis of 33 bankrupt and cash variables extracted from Dividend component was the only
(USA) 33 non-bankrupt firms three years Helfert’s (1972) model and scaled significant cash flow variable.
prior to failure. Validation sample using total net flow. Cash flow from operations not
consisted of 23 weak companies significant.
matched on size and industry with
23 non-weak companies.

Gentry et al (1985b) Probit analysis of 33 bankrupt and cash variables extracted from None of the accrual ratios were
(USA) 33 non-bankrupt firms three years Helfert’s (1972) model and scaled significant predictors. Dividends,
prior to failure. Assessed the using total net flow. investment and receivables cash
incremental predictive ability of variables were reliable predictors
accrual ratios and cash variables. one year prior to failure.

Viscione (1985) Trend analysis of 24 bankrupt cash flow from operations. cash flow from operations was not
(USA) firms up to five years prior to a strong indicator of financial
failure. Compared cash flow from distress.
operations with selected accrual
ratios.
7
Table 1: Summary of Studies Investigating Cash Flow Information as Predictors (Cont.)

Study Method Cash Flow Variables Results

Gombola et al (1987) Computed 21 accrual ratios and cash flow from operations/sales None of the cash flow ratios were
(USA) three cash flow ratios for 77 failed cash flow from operations/assets significant predictors of failure.
and matched non-failed firms. cash flow from operations/debt.
Managerial Finance

Data collected for at least one year


for each firm up to four years prior
to failure. Employed linear MDA.

Gahlon & Vigeland (1988) Comparison of cash flow profiles cash flow from sales activity, cash On average, cash flow from
(USA) with selected accrual ratios for 60 flow from operations, cash flow operations, cash flow after debt
bankrupt and 204 non-bankrupt after debt retirement, cash retirement, and cash coverage
firms five years prior to failure. coverage ratio. ratios indicated failure as early as
the fifth year prior to failure.

Dambolena & Shulmen (1988) Recomputed logit model net liquid balance which equals net liquid balance improved the
(USA) equivalents for Altman’s 1968 operating cash flows minus predictive accuracy of both
model and Gentry et al 1985b. increases in cash investments, plus models especially for
Used 25 bankrupts matched with increases in long term financial non-bankrupt firms. Improvement
25 non-bankrupts. A similar flows. in predictive accuracy was greater
sample size was used for for the Gentry et al model than for
validation. Tested the marginal Altman’s model.
predictive ability of a funds flow
ratio.

Aziz et al (1988) Investigated the potential of cash various cash flow variables cash tax paid was the most
(USA) flow variables extracted from extracted from the Lawson cash consistent variable with operating
Lawson’s cash flow identity flow identity model. cash flow ranking as the second
model to classify firm failure. For most significant predictor. Vast
49 failed firms matched with improvement in predictive
non-failed firms MDA and logit accuracy of failed firms.
models were developed.
8
Table 1: Summary of Studies Investigating Cash Flow Information as Predictors (Cont.)

Study Method Cash Flow Variables Results

Aziz & Lawson (1989) 49 bankrupt firms matched with Various cash flow variables The cash flow model was more
(USA) 49 non-bankrupt firms up to five extracted from the Lawson cash accurate in predicting
years prior to failure. Compared flow identity model. bankruptcies. Operating cash flow
cash models with Altman’s Z and and lender cash flow were the two
Zeta models, and a mixed model most significant cash variables.
comprising cash and accrual
variables.
Volume 27 Number 4 2001

Gilbert et al (1990) Investigated the predictive abilities cash flow from operations/current CFFO/TL significant in
(USA) of models based on two types of liabilities (CFFO/CL) classifying bankrupts and
samples: 52 bankrupt and 208 cash flow from operations/total non-bankrupts and CFFO/CL
non-bankrupt firms and 52 liabilities (CFFO/TL) significant in classifying bankrupt
bankrupt and 208 distressed firms. cash flow from operations/total and distressed. Concluded that
Holdout sample used to test assets. cash flow ratios add significantly
accuracy of model. Used 14 ratios to prediction accuracy of accrual
of which three were cash flow models.
ratios. Applied stepwise logit.
9
Managerial Finance 10

probability of failure enables the assessment of the degree of distress and the risk associ-
ated with a particular company. Bankers may lend at premium interest rates to companies
that are classified in the failed group that have a marginal probability of failure. Further-
more, bankers usually assess credit risk, which is a continuous measure, of a loan appli-
cant rather than merely classifying the applicant into a dichotomy of risky versus
non-risky. To illustrate the usefulness of cash flow information in predicting the probabil-
ity of failure, figures 3 and 4 (p.397 and p.399 of Casey and Bartczak, 1985) showing the
histogram of predicted probabilities of failure for companies in the validation sample are
reinterpreted.
Figure 3 showed the predicted probabilities for bankrupt and non-bankrupt groups
based on the accrual only model. Figure 4 also showed the same information following
the addition of operating cash flow information to the accrual model. A comparison of
figures 3 and 4 reveals the following. Following the addition of operating cash flow vari-
ables to the accrual model, the number of bankrupt companies with a probability of fail-
ure greater than 50% increased from 32 to 38. Moreover, there is a tendency of firms to
now cluster between the median and the third quartile which now has a narrower spread
than without the cash flow variables. It would be interesting to know which companies
changed probabilities and in which direction - it appears that almost all companies
showed increased probabilities of failure. For the non-bankrupt sample, the accrual only
model predicts that 27 companies have a zero probability of failure while the cash flow
inclusive model predicts 48 companies with a zero probability of failure. This is almost a
100% increase in correctly assigning probabilities of non-failure. If it is accepted that the
cost of misclassifying a bankrupt company is greater than the cost of misclassifying a
non-bankrupt company, then there is further evidence in favour of cash flow information.
According to Altman et al (1977) the cost to a lender of misclassifying a bankrupt com-
pany is far greater than the cost of misclassifying a non-bankrupt company. Casey and
Bartczak (1984) report classification accuracies for the bankrupt, non-bankrupt and the
overall sample based on the accrual model and each of the three cash flow variables, viz.
operating cash flow (OCF), operating cash flow to current liabilities (OCF/CL), and oper-
ating cash flow to total liabilities (OCF/TL). For the bankrupt group, the accrual only
model had the following accuracies: 83% in year one, 63% in year two, 57% in year
three, 30% in year four and 73% in year five. Respective accuracies for the cash flow
models were: OCF model: 90% in year one, 92% in year two, 83% in year three, 88% in
year four and 85% in year five; OCF/CL model: 83%, 70%, 60%, 72% and 63% and for
the OCF/TL model: 82%, 72%, 58%, 70% and 53%. Except for the fifth year before fail-
ure, the cash flow models outperform the accrual model in each year. The most notable
difference is between the OCF model and the accrual model. On the basis of these obser-
vations and the re-interpretation of Casey and Bartczak’s (1985) results, it is argued that
cash flow information has potentially significant information content in predicting prob-
abilities of failure particularly for the bankrupt group of companies.
Conclusions supporting Casey and Bartczak (1985) were reached by Gentry et al
(1985a) and Gombola et al (1987). Although Gentry et al (1985a) sought to investigate
the predictive ability of an alternative to accrual accounting information, they did not test
the ability of funds flow relative to the accrual model. In a subsequent publication, Gentry
et al (1985b) tested the incremental ability of funds flow variables over accrual variables
in predicting failure. They found that the addition of funds flow variables to the accrual
model significantly improved the classification accuracy of the accrual only model. Un-
like Casey and Bartczak (1985), Gentry et al (1985b) tested the incremental ability of
Volume 27 Number 4 2001 11

funds flow information both ways, that is, adding funds flow variables to the accrual
model and adding accrual variables to the funds flow model. However, Gentry et al
(1985a) measured CFFO based on Helfert’s (1972) funds flow model. Helfert’s model
does not include CFFO but refers to funds from operations that more closely represents
working capital from operations. Gombola et al (1987) conducted a factor analysis of 21
accrual ratios and three cash flow ratios and found that cash flows loaded on a separate
factor in the later years being 1973-1981 but did not load separately during the earlier
years, 1967-1972. Based on this finding, Gombola et al (1987) argued that if cash flow
has information content then it should be more salient based on data from the later years.
They used Casey and Bartczak’s (1985) measure of CFFO. Their results, based on one
cash flow ratio CFFO/SALES, did not reveal any significant information content in cash
flows for predicting failure. It is interesting to note that four of Gombola et al’s (1987)
coefficients did not have a logical sign. The following coefficients had counter-intuitive
signs, specifically negative signs: cash flow from operations to total assets in years two
and four prior to failure, current assets to sales in years one, two and four, sales to total as-
sets in year four, and income to assets in years one and two. The sign of these coefficients
are considered counter-intuitive because the sign for these ratios in Gombola et al’s
model implies, for example, that as return on assets increases, the likelihood of failure in-
creases.

Based on a trend analysis of 24 bankrupt firms over a five year period, Viscione
(1985) compared CFFO measured following Casey and Bartczak (1985) to selected ac-
crual ratios. He found that CFFO was not significant in identifying financial failure. Vis-
cione (1985, p.54) argued that CFFO as a variable may be misleading because of
management’s manipulation of the timing of cash flows:

The effect of not paying bills on time or reducing inventory below desired levels
is to increase the measure of cash flow from operations derived from income
statements and balance sheets...Such an increase is obviously not a good sign,
and it is probably safe to assume that these kinds of distortions arise most often
for firms experiencing financial distress.

Upon further investigation, Viscione (1985) found that 50% of his sample with posi-
tive CFFO drastically reduced inventory and accounts receivables and had high levels of
accounts payable. He subsequently cautioned the use of levels of CFFO and suggested
employing cash flow ratios.

In a purely descriptive study, Gahlon and Vigeland (1988) conducted univariate


non-parametric tests of 16 cash flow variables for 60 bankrupt and 204 non-bankrupt
firms for the 1973-1985 period. They computed cash flow variables using the Uniform
Credit Analysis method “adopted by numerous banks in their many commercial lending
activities” (Gahlon and Vigeland, 1988, p.5). One cash flow ratio, cash coverage ratio1,
was significantly (p<0.05) different for bankrupt and non-bankrupt firms for each of the
five years prior to failure. Four cash flow variables (cash operating income, cash income
taxes, cash flow after debt retirement and cash flow before financing) were also signifi-
cantly (p<0.05) different in each of the five years prior to failure. In addition, for each of
the three years immediately prior to failure, again four variables (net cash flow from op-
erations, cash net income, mandatory debt retirement, and total financing activity cash
flow) were significantly different between bankrupt and non-bankrupt firms.
Managerial Finance 12

Dambolena and Shulman (1988) developed a variant of CFFO called net liquid bal-
ance. Net Liquid Balance (NLB) was derived by subtracting increases in cash invest-
ments and adding increases in long term financial flows to CFFO disclosed in the
statement of cash flows. Dambolena and Shulman (1988) found that when NLB was
added to Altman’s (1968) Z score model and to Gentry et al’s (1985b) model it markedly
and consistently improved the predictive performance of both these models. In each case
the NLB variable was a statistically significant predictor suggesting that it signified a di-
mension different to any other variable in both models. One year prior to failure, the uni-
variate NLB variable misclassified only one out of the 50 failed firms while eight
non-failed firms were classified as failed.

Aziz, Emanuel and Lawson (1988) and Aziz and Lawson (1989) used Lawson’s
cash flow identity as the framework for their study investigating the ability of cash flow
to predict financial distress. They used this identity since they perceived corporate bank-
ruptcy to be closely related to firm valuation which in turn is closely related to Lawson’s
identity of cash flows.2 In the former study, the authors sought to “determine the suitabil-
ity of Lawson’s identity components for discrimination between bankrupt and non-
bankrupt firms” (Aziz, Emanuel and Lawson, 1988, p.422). Using both MDA and logistic
regression they found that cash flow variables correctly classified bankrupt and non-
bankrupt firms with a high degree of accuracy up to five years prior. In comparison to the
recomputed Altman’s (1968) Z model their cash flow model performed comparably. Aziz
et al’s (1988) cash flow model outperformed the Z model in four out of five years prior to
bankruptcy. While the Zeta model performed well in the second year prior to bankruptcy,
its performance in other years was similar to the cash flow model. In identifying potential
bankruptcies, the cash flow model outperformed both the Z and Zeta models. In the study
of 49 bankrupt and 49 non-bankrupt firms over the 1973-1982 period, Aziz and Lawson
(1989) obtained results similar to those described above. Aziz and Lawson (1989, p.62)
consequently concluded that cash flow failure prediction models:
exhibit a superior predictive accuracy and superior early-warning capabilities...
With the cost of incorrectly classifying a potentially bankrupt firm in mind, it
appears that C+B’s (Casey and Bartczak, 1984) suggestion to de-emphasise
cash flows in tracking the financial health of firms would be a retrogressive step.
However, Aziz and Lawson (1989) do not fully discuss the significance of the indi-
vidual variables of the cash flow model. They merely state that “TAXP is the only vari-
able that is significant (a = 0.05) for all five years, while each of the others is significant
(0.05 £ a ³ 0.10) for two of the five years” (Aziz and Lawson, 1989, p.59). They do not
disclose the coefficients nor the signs of their cash flow model. Hence, it becomes diffi-
cult to evaluate the significance of the variables in their model. It is possible for the model
to be highly accurate without many significant variables since as the number of independ-
ent variables increase, the total explanatory power also tends to increase (Hosmer and Le-
meshow, 1989).

One salient distinction from their 1988 study was that Aziz and Lawson (1989) also
tested a combined predictive model comprising cash flow variables and accrual ratios of
the Z model. They found that the mixed model performed as well as the cash flow only
model. Aziz and Lawson’s (1989) mixed model contained five accrual ratios and five
cash flow variables. Only three variables (operating cash flow, earnings before tax and in-
terest to total assets, and retained earnings to total assets) exhibited consistent signs of the
Volume 27 Number 4 2001 13

coefficients over the five year period. These three model variables had negative signs
which are intuitively appealing. It should be noted though that neither of these two stud-
ies employed cash flow ratios and hence is not comparable to prior cash flow studies.

In examining whether distressed firms filing bankruptcy could be distinguished


from those that avoid filing, Gilbert et al (1990) observed that contrary to the findings of
Casey and Bartczak (1985), cash flow ratios (cash flow from operations to total liabilities
and cash flow from operations to current liabilities as defined by Casey and Bartczak
(1985)) were significant predictors of distress. In both their logit models, the cash flow
ratios had intuitively appealing signs with the coefficients being significant at p=0.001.
The two cash flow ratios however did not simultaneously occur in any one model. This
suggests that the stepwise logit methodology adopted by Gilbert et al (1990) recognised
and appropriately treated the correlation between the two cash flow ratios.

Laitinen (1994) compared the predictive ability of traditional cash flow defined as
net income plus depreciation and CFFO defined as adjusting net income for accruals and
deferrals. Using 40 failed and 40 matched non-failed small and medium size Finnish
firms, Laitinen demonstrated through discriminant and logit analyses that traditional cash
flow was a more stable and reliable predictor of failure than CFFO. However, like Aziz et
al (1988) and Aziz and Lawson (1989), Laitinen did not employ cash flow ratios nor did
he attempt to investigate the predictive abilities of accrual versus cash flow ratios.

Ward (1994) investigated why traditional cash flow is thought to be a strong predic-
tor of financial distress. By adding the traditional cash flow variable to a model compris-
ing six accrual ratios and CFFO, Ward observed that the significance of cash flow from
operations was not affected. Rather he found that the net income to total asset variable be-
came insignificant. Ward (1994, p.53) subsequently concluded that traditional cash flow
“is a significant predictor of financial distress because NOF is a better measure of eco-
nomic income than NITA (net income to total assets), not because NOF is a naive meas-
ure of operating cash flow.”

Ward’s (1994) findings are inconsistent with Laitinen (1994). Several possibilities
for the differences exist. First, Laitinen’s study was based on 40 small and medium sized
bankrupt and non-bankrupt Finnish firms, while Ward’s (1994) sample consisted of 164
healthy and 63 non-healthy US firms for the estimation sample and 111 healthy and 47
non-healthy firms for the hold out sample. Ward’s firms are presumably larger than Laiti-
nen’s since they were listed US firms. Secondly, Ward selected his sample from a 12
month time period only while Laitinen does not disclose the time period of his sample.
Thirdly, Ward actually tested the incremental predictive power of traditional cash flow by
adding it to a model containing six accrual ratios and the CFFO variable. Laitinen how-
ever simply compared the univariate discriminant and univariate logit performances of
the two cash flow variables. Finally, the definition of CFFO may be different between
Finland and USA since the computations disclosed to arrive at CFFO by adjusting net in-
come for accruals and deferrals were not identical between the two studies.

The literature examining the value of cash flow information for predicting failure
can be summarised as inconsistent and inconclusive. While Casey and Bartczak (1984;
1985), Gombola et al (1987) and Viscione (1985) demonstrated the CFFO does not pos-
sess incremental information content over accrual information in predicting failure, com-
parable studies by Gahlon and Vigeland (1988), Gilbert et al (1990) and Ward (1994)
Managerial Finance 14

show that CFFO adds significant predictive power to accrual models. There are several
explanations why the literature is in such a state. The following section offers possible
explanations for the state of the literature.
III. Limitations of Prior Cash Flow Failure Prediction Studies
Measurement Problems
The construct validity of operating cash flows is a significant factor explaining the vari-
ability in the results of prior studies. Gombola et al (1987) identified eight studies that
measured operating cash flow as net income plus depreciation, depletion and amortisa-
tion. Early research (Beaver, 1966; Deakin, 1972; Blum, 1974) and some later research
(Norton and Smith, 1979; Mensah, 1983) measured CFFO as net income plus deprecia-
tion, depletion, and amortisation. This formulation did not correctly measure CFFO. Con-
sequently, Largay and Stickney (1980) suggested a more complete formulation of CFFO.
Largay and Stickney’s (1980) method for measuring CFFO was subsequently used as the
basis for computing CFFO in later research (Casey and Bartczak, 1984 and 1985; Vis-
cione, 1985; Gilbert et al, 1990; Laitinen, 1994). Other researchers (Gentry et al, 1985a
and 1985b; Aziz et al, 1988; Aziz and Lawson, 1989; Gahlon and Vigeland, 1988) have
utilised fund flow models3 to define their cash flow variables.
Gombola and Ketz (1983) and Bowen, Burgstahler, and Daley (1986) demonstrated
that CFFO and its proxies were distinct and that each had different information content.
This generated research investigating the information content of cash flow and accrual in-
formation in a number of traditional areas such as the capital market studies. While these
studies show that cash flow and accrual information are not highly correlated, that is, they
do not measure the same construct, results on the predictive ability and incremental con-
tent are mixed. For instance, Bowen et al (1987) found that when cash flow data was
added to an accrual model, cash flow data enhanced the ability of accrual data in explain-
ing security price changes. When accrual data was added to a cash flow model, they ob-
served that it also improved the explanatory power of the cash flow model in explaining
security price changes. Murdoch and Krause’s (1989) investigation of the predictive
power of accrual and cash flow data in forecasting operating cash flow indicated that ac-
crual data possess incremental predictive ability beyond that provided by cash flow data
and that cash flow data is not useful for forecasting cash flow. In a subsequent study em-
ploying time series data, Murdoch and Krause (1990) found support for cash flow data
containing incremental information content over accrual data in predicting future cash
flows. In explaining dividend changes Simons (1994) concluded that cash flow data do
not contain information beyond that provided by accrual information. Most of these com-
parative studies however relied on reconstructed measures of CFFO. Reconstructing
CFFO may not capture the real CFFO. Recently, Austin and Bradbury (1995) demon-
strated that even the best mechanical rules for reconstructing CFFO produced large er-
rors. Therefore, these studies and in particular, prior failure prediction studies that have
employed mechanical reconstructions of CFFO may not have tested the power of real
CFFO. This contention was raised earlier by Watts and Zimmerman (1986) in relation to
capital market research and by Gombola et al (1987) in relation to predicting company
failure. These authors questioned the validity of prior studies employing proxies that
were remote from the “real” thing. In acknowledgement of the possibility of errors in
their estimation of CFFO, Gombola et al (1987) encouraged replication of their study
when cash flow data is available through company annual reports. Another implication of
Volume 27 Number 4 2001 15

mechanical reconstruction is that prior studies have employed a diversity of methods for
measuring CFFO therefore no firm cross-sectional conclusions can be made regarding
the ability of cash flow information to predict corporate failure.

Model Validation

A further limitation of prior cash flow studies is that not all studies have validated their
models. Model validation is essential because it is known that performance in the deriva-
tion sample will generally exhibit unrealistically low misclassification rates (Marais et al,
1984). Once again, this questions the validity and conclusions of some cash flow failure
prediction studies. It also questions the generalisability of the models derived.

Cash Flow and Accrual Information Multicollinearity

Gombola et al (1987) suggested that cash flow information would be more salient in the
late years relative to early years. Late years included the period 1973-1981 while the
early years covered the period 1967-1972. Through a factor analysis they found that cash
flow information did not load on a separate factor in the early years but loaded on a sepa-
rate factor in the late years period. This suggests that cash flow failure prediction studies
employing data from the early years would not detect significance in cash flow informa-
tion in a multivariate model because of multicollinearity between cash flow and accrual
information. The significance would tend to increase between the early and late years. Of
the multivariate studies that did not find cash flow information to be significant in pre-
dicting failure two, of which one is the most widely quoted study, combined early and
late years data (Casey and Bartczak, 1984 and 1985; Gentry et al, 1985a). On the other
hand, studies that found cash flow information to be significant used late years data (Aziz
et al, 1988; Aziz and Lawson, 1989; Gilbert et al, 1990). This suggests that cash flow in-
formation was becoming more salient with time perhaps due to inflation (Hawkins,
1977), the increasing number of corporate collapses (Altman, 1983), and the movement
of income measurement from a cash basis to an accrual basis (Hawkins, 1977). Now that
there is widespread recognition that cash flow information is distinct from that provided
by the income statement and balance sheet, and the consequential introduction of a cash
flow accounting standard in the USA, New Zealand, Canada, UK, and Australia, it would
appear that, cash flow information would be more salient and hence not exhibit high mul-
ticollinearity with accrual data. This suggests that cash flow information may provide
added predictive ability to that provided by accrual information, and the data saliency
confounding in past research can now be properly addressed.

Research Diversity

The diversity of operating cash flow variables investigated also inhibit drawing any firm
conclusions because researchers have sought to define and derive their own cash flow ra-
tios. While this extends the range of operating cash flow variables tested, it does not pro-
vide adequate replication in what was, and may still be considered, a fairly new research
paradigm. According to Kuhn (1970), repeated confirmations is essential for reaching a
consensus on a phenomenon in a given research paradigm. This diversity issue is com-
pounded by the diversity of research approaches and statistical techniques employed in
the paradigm. For instance, leading studies (Largay and Stickney, 1980; Lee, 1982)
adopted a case study approach that conducted a more detailed analysis of their failing
firm’s cash flow. Other studies (Casey and Bartczak, 1984 and 1985; Gombola et al,
Managerial Finance 16

1987; Gilbert et al, 1990) have conducted large matched (bankrupt and non-bankrupt)
sample multivariate investigations of up to three cash flow variables in conjunction with
accrual variables. Viscione (1985) examined cash flows, its proxies and accrual ratios for
13 bankrupt firms only. The variety of statistical techniques used include simple univari-
ate ratios (Beaver, 1966; Gahlon and Vigeland, 1988), simple multiple-discriminant
analysis (MDA) (Deakin, 1972; Blum, 1974; Casey and Bartczak, 1984; Gombola et al,
1987), stepwise MDA (Norton and Smith, 1979), logit (Gentry et al, 1985a; Dambolena
and Shulmen, 1988) stepwise logit (Gilbert et al, 1990), logit and MDA to detect stastical
technique sensitivity (Mensah, 1983; Casey and Bartczak, 1985; Aziz et al, 1988; Aziz
and Lawson, 1989; Laitinen, 1994), and probit (Gentry et al, 1985b). Different statistical
techniques may produce different results. This occurs because the functional relationship
between the independent (predictor) variables and the dependent variables may be suited
more to a particular technique than to another. The results achieved may be due to statisti-
cal artifact rather than the manipulation of the independent variables. This artifact could
be detected through the application of two or more statistical techniques. If the two tech-
niques produce very similar results then method variance is said to be low with the results
being attributed largely to the manipulation of the variables.

Narrow Focus

Most of the prior cash flow failure prediction studies investigating the predictive content
of cash flow have focused too narrowly on operating cash flow. These studies have ig-
nored the potential of other cash flow variables, in particular the role of cash flow compo-
nents in predicting financial failure. Thus, these studies may have prematurely and
unjustly dismissed the value of cash flow information in predicting financial failure.

IV. The Concept of Cash Flow

Cash flow is not a new concept. Up to the 18th century cash flow accounting was the
main reporting system (Winjum, 1972). There was relatively little importance attached to
accounting allocations and profit measurement since most transactions were conducted
and recorded in cash terms. However, the industrial revolution precipitated business
growth and continuity which required more funds than the business generated. The com-
plexity and number of transactions exploded giving rise to the need for periodic financial
statements measuring profit and financial position. Consequently, accounting allocations
became necessary (Littleton and Zimmerman, 1963). The accrual measures were an ap-
proximation of future cash flows. Over the life of the business, it was expected that even-
tually profits and the value of net assets would equal net cash flow. Thus, the cash flow
concept formed the core of the allocation-based accrual system of accounting as we know
it today.

The one consistent argument amongst proponents of cash flow reporting is that cash
flow reporting avoids the frailties associated with the allocation system that constitutes
the core of our conventional accounting system. Consequently, these proponents (Tho-
mas, 1969; Heath, 1978; Jones, 1975; Lawson, 1968 and elsewhere; Lee, 1971 and else-
where) recommended abandonment of income measurement in favour of measuring cash
flows. Much of the development in cash flow accounting is attributable to Lawson and
Lee. These authors viewed the entity as a total financial system. This system represents
the cash inflows an entity generates and the cash outflows it makes with any residual be-
Volume 27 Number 4 2001 17

longing to owners or shortfall indicating borrowings. The essence of their system was
captured by the following equations where O = net cash flow from operations, R = cash
outflow for replacement and growth investment, T = taxation cash flow, I = cash outflow
for interest on borrowings, E = cash injection from owners, B = borrowings, C = change
in cash resources of the entity, and D = distributions to owners:
O-R-T-I+E+B-C=D (Equation 1: Lawson)

O-R-I-T-D+E+B=C (Equation 2: Lee)

O-I-T-D+E+B-R=C (Equation 3: Lee’s revised version of Equation 2)


The difference in these equations reflects the order and priority of cash inflows and
cash outflows. Lee’s first formulation suggests that replacement and growth investment
to ensure continuity of the entity had the highest priority claim to any net operating cash
flow surplus followed by interest, taxation, and distributions to owners. In his revision
however, he ranks replacement and growth investment as having the lowest priority. His
revision was justified on grounds that it was more neutral and did not suggest as his first
equation did that borrowings and equity inflows were being used to meet the operating
cash flow shortfall for replacement and growth investment, interest, taxation, and distri-
butions to owners. Lee’s revision remains valid today. It is valid because interest and
taxation are considered non discretionary cash outflows while cash outflows for reinvest-
ment are considered discretionary (Sondhi, Sorter and White, 1987; Stancill, 1987; Ste-
phens and Govindarajan, 1990).
In a financial analysis context it has been argued that cash flows provide value rele-
vant information complementary to accrual information (Charitou and Venieris, 1990).
As early as the sixties, Coughlan (1960, 1962, 1964), Edey (1963) and Staubus (1961;
1966) advocated the use of cash flows for performance measurement primarily because
of distortions to accounting income caused by conventional allocations. None of these
authors have developed their work on cash flows any further. Later, researchers such as
Ijiri (1979), Heath (1978), Climo (1976), Lawson (1968 and elsewhere) and Lee (1971
and elsewhere) strongly argued that cash flow information was more relevant than ac-
crual information for evaluating an entity’s performance, liquidity and solvency. The ra-
tionale for the relevance of cash flow information for financial analysis is discussed in the
following section.
V. The Relationship Between Cash Flow and Corporate Health
Cash Flow: The Key to Solvency
Cash flow is a critical business solvency measure. Heath (1978, p.20) relates cash flow to
financial flexibility when he states that financial flexibility is the capacity of a firm “to
control cash receipts and payments to survive a period of financial adversity.” Further,
Heath and Rosenfield (1979, p.48) contended that:
Solvency is a money or cash phenomenon. A solvent company is one with ade-
quate cash to pay its debts; an insolvent company is one with inadequate cash.
Evaluating solvency is basically a problem of evaluating the risk that a company
will not be able to raise enough cash before its debts must be paid.
Managerial Finance 18

Solvency analysis is not simply a matter of evaluating a company’s so-called


current assets and liabilities...
From a lenders perspective, cash is what services debt and pays the debt. From a sur-
vey of 266 US commercial loan officers, Fulmer, Gavin, and Bertin (1991) observed that
for new commercial loan applications, 90% of lending officers require a statement of
cash flows. 80% of these lenders also require a cash flow schedule specified by the
lender. In terms of lending risk factors, Fulmer et al (1991) found that respondents rated
stability of cash flows over time as the most important factor followed by, in rank order of
importance, stability of net income, key ratios specified by the lender, subjective factors
(size, reputation, past relationship with bank), industry trends, credit agency rating, fit of
the loan in portfolio, and credit scoring models (z score). More recently, Rizzi (1994)
provided a framework for measuring debt capacity which included default risk and de-
fault loss. His proposed measure of default risk was cash flow coverage ratios. His ration-
ale for using cash flow coverage ratios was that debt is primarily serviced from internally
generated free cash flow. Past cash flow failure prediction research implicitly assumed
the relationship between cash flow and debt servicing and debt paying capacity when for-
mulating a relationship between cash flow and financial failure. This is a very reasonable,
perhaps more reasonable than the relationship between asset values and income and debt
servicing and debt paying capacity, assumption indeed and one which is grounded in a
reasonable theory.
Cash Flow and Dividend Policy
It is widely recognised that the level of dividend payout is related to the capacity of the
firm to make such payments. The ability of an entity to make dividend payments there-
fore reflects its financial state of affairs. Whether the capacity to make dividend payout is
empirically related to cash remains a contentious issue. To shed further light on this issue,
Simons (1994) investigated the incremental predictive ability of cash flow over accrual
information in predicting dividend changes. She observed that cash flow information did
not possess significant incremental content over accrual information. However, in the
case of financially distressed firms it is plausible that a relationship between cash flow
and dividend payout or dividend changes may exist. This argument is plausible for two
reasons. First, dividend payment is made out of cash flows and under financial distress
where cash flows are poor, it would be expected that dividend payments would be re-
duced or omitted. Secondly, in the long term it is cash flows that are important in sustain-
ing dividend payments. Consequently, shareholders pay considerable attention to future
cash flows in the form of future dividend flows when ascertaining the value of the firm.
DeAngelo and DeAngelo (1990, p.1430) suggested that managers of distressed firms
with long dividend histories are reluctant to omit dividend payments “because they would
be the first managers in many years whose policies have generated insufficient cash to
pay dividends.” Without adequate cash flow the alternative is to reduce cash dividend
payments, resort to bonus share or rights issues or to borrow to meet dividend payments.
Any one or a combination of these alternatives may be an indication of insufficient ongo-
ing operating cash flows and consequently financial distress.
Cash Flow and Creative Accounting
It is widely recognised in the literature that creative accounting goes against the spirit of
accounting regulation and its primary objective is to mislead and even defraud users of fi-
Volume 27 Number 4 2001 19

nancial reports in extreme circumstances. The literature (e.g. Tweedie and Whittington,
1990) also argues that creative accounting is largely possible and encouraged by the flexi-
bility of accounting standards and generally accepted accounting principles. Given that
the objective of creative accounting is to portray the favourable by distorting and sup-
pressing unfavourable accounting information, it is not surprising that distressed compa-
nies are more likely than non-distressed companies to engage in such practices. The
literature (e.g. Schwartz, 1982; Lilien et al, 1988; Dharan and Lev, 1993; Sharma and
Stevenson, 1997) consistently reports a high propensity for distressed firms to engage in
creative accounting.
Consequently, no matter how sophisticated the statistical technique, if financial data
is not adjusted for the effects of creative accounting, failure prediction models will pro-
duce erroneous results. Zavgren and Freidman (1988) attribute lack of significance of two
of their model variables to noisy measurements and the possibility that the profit figures
used were manipulated. In fact Ohlson (1980) found that of the 13 misclassified bankrupt
firms, 11 reported profits and displayed no signs of distress. Ohlson (1980, p.129) pointed
out that
the reports of the misclassified bankrupt firms seem to lack any “warning sig-
nals” of impending bankruptcy. All but two of the thirteen companies reported a
profit. The two losses were relatively minor...and these two companies had
strong financial positions...Other ratios analyzed showed the same “healthy”
patterns. It is not surprising that these firms were misclassified...
Ohlson (1980) further explained that alternative estimation techniques would also
produce results similar to his logit model. Upon testing this proposition, his results did
not refute his argument. Therefore, he concluded that “more than anything else, signifi-
cant improvement probably requires additional predictors” (Ohlson, 1980, p.130). Given
that Ohlson’s (1980) six of the nine predictors were most frequently mentioned in the lit-
erature and were accrual financial ratios, it is not unreasonable to suggest cash flow vari-
ables as the “additional predictors” that may improve prediction accuracy.
In addition, where financial statement users are aware of creative practices and those
that have low confidence in the integrity and reliability of reported accounting informa-
tion, these users will at least be skeptical of the results produced by failure prediction
models. This is one possible explanation why failure prediction models are not widely
used in practice. Gadenne and Iselin (1995) noted that human decision makers may not
use failure prediction models because model builders have tended not to adjust for the
window dressing effect on financial information. Gadenne and Iselin (1995) suggested
that future research should seek to adjust for the window dressing effects prior to model
construction and seek predictor variables which are more difficult to window dress. The
former suggestion may be difficult to implement because window dressing effects are not
easily discernible from a firm’s financial statements. The second suggestion is more fea-
sible using cash flow information because, as explained below, it is less susceptible than
accrual information to window dressing and its scope for window dressing relative to ac-
crual information is limited.
Lee proposed (1981) that cash flow information has numerous advantages over ac-
crual information. Because cash flow accounting avoids dubious accounting allocations
present in the accrual system, because it produces more objective financial information,
because it provides users with fundamental and critical financial data, because cash flow
Managerial Finance 20

data is comparable across entities since subjective accrual allocations are excluded and
because cash flow data are unambiguous measures of financial performance, Lee (1981)
argued that cash flow information is more meaningful, relevant and reliable than accrual
based accounting data particularly in the context of financial analysis. If these arguments
are valid, then one would expect that cash flow data is less susceptible to manipulation in
comparison to accrual financial information. This proposition was specifically investi-
gated by Sharma (1995).
Sharma (1995, p.56) argued that “if cash flow reporting avoids dubious allocations
and subjectivities of accrual accounting, it will not be affected by accounting policy
changes in the same way conventional measures of performance are affected.” He dem-
onstrated on the basis of a hypothetical company that cash flow information does tend to
lessen the effects of accrual window dressing. His analysis showed that a company en-
gaged in window dressing through accounting policy changes was classified as healthy
using Altman’s (1968) failure prediction model but on a cash flow basis, the company’s
financial performance and financial condition indicated financial distress. Sharma’s
(1995) analysis implies that cash flow information may introduce a degree of objectivity
in the performance evaluation process currently not provided by accrual based perform-
ance analysis. Other studies indirectly supporting the objectivity of cash flow information
include Largay and Stickney (1980) and Kochanek and Norgaard (1988) and more re-
cently Sharma (1996). In particular, Kochanek and Norgaard (1988) found that the Char-
ter Company’s earnings were inflated by the inclusion of nonrecurring and noncash
items. Consequently, Charter’s financial demise was obscured. On a cash basis though,
because cash flow information captures information relevant to financial distress and it
avoided the subjectivity and manipulations impounded in Charter’s earnings. Charter’s
financial inflexibility indicated financial distress. Sharma (1996) demonstrated similar
effects for the case of Brash Holdings, a large music and electronics goods retailer in
Australia. Since the analysis of Kochanek and Norgaard (1988) and Sharma (1996) were
based on a single company, further investigation through large-scale generalisable em-
pirical research is necessary before any firm conclusions are made.
Furthermore, because accrual accounting policies tend to vary between firms within
an industry and across industries, comparative financial performance analysis may be
limited by the extent of differences in the accounting policies adopted. Consequently, the
financial performance analysis may not reflect the actual economic performance of the
firm but reflect the effect of the accounting policies adopted. Some users, especially
bankers (Dietrick and Stamps, 1981), attempt to adjust the financial information due to
accounting policy differences prior to performing financial analysis. The adjustment pro-
cess may be problematic because some firms may not fully disclose the accounting poli-
cies adopted and computing the financial effect for disclosed policies may also present
difficulties. This problem is aggravated when users do not have sufficient knowledge of
accounting policies, principles, procedures, and terminology. Thus, even if adjustments
were attempted some inherent differences would remain suggesting that the problem lies
with the current reporting system (Tweedie and Whittington, 1990). Therefore, a report-
ing system that enhances comparability between firms is required. The FASB in Amer-
ica, AASB in Australia and other global accounting standard setting bodies that have
introduced a requirement that public companies report a statement of cash flows argue
that this new accounting report will improve comparability amongst firms. This is so be-
cause the cash flow statement is based on a single accounting policy, that is, to report all
cash inflows and cash outflows. A difference that will need to be adjusted relates to the
Volume 27 Number 4 2001 21

classification of items into operating, investing, and financing. However, given the extent
of disclosure this adjustment will be a relatively less difficult process. Uncontrollable in-
dustry differences and management manipulation of the timing of cash flows will remain.
VI. Summary, Conclusion and Future Directions
Despite numerous failure prediction studies investigating the ability of cash flow infor-
mation to predict corporate failure, their results are mixed and hence inconclusive for the
following reasons:
(1) Cash flow from operations has not been properly measured.
(2) Some researchers have not validated their model using a validation sample.
(3) Studies did not take note of Gombola et al’s (1987) caution on conducting time
series analysis using cash flow information because in the early years cash flow
data and accrual data were correlated thus the significance of cash flow data
would not have emerged. Most of the studies that rejected cash flow as a useful
predictor used data from a time period that Gombola et al (1987) described as
early years data where cash flow and accrual data were highly correlated.
(4) Researchers have tended to focus primarily on cash flow from operations thus ig-
noring other potentially important cash flow variables and components of vari-
ous cash flows.
(5) The large variety of cash flow ratios investigated with varying measurements
and the diverse range of research methods and statistical techniques employed in
the paradigm makes cross comparisons amongst studies difficult. This also does
not provide sufficient evidence of replication of past studies which is essential to
establish the existence of a phenomena in a given paradigm.
Several research opportunities are evident from the identified limitations of the criti-
cal literature review. The most obvious opportunities relate to addressing these limita-
tions. A fruitful avenue for future research is to investigate the information content of
cash flow information contained in the statement of cash flows. Using cash flow from op-
erations reported in the statement of cash flows avoids the problem associated with using
proxies. Research could investigate the predictive ability of reported and computed val-
ues of cash flow from operations. The effects of total cash flow from operations (aggre-
gated CFFO) and its components (disaggregated CFFO) on failure prediction accuracy
may provide further insight on the relationship between liquidity and financial distress.
Examining the relationship between the components of operating cash flow would high-
light the causes of the variability in operating cash flow. This can lead to identifying com-
panies that are reducing inventory and receivables and increasing accounts payable to
manipulate cash flow from operations. Investigating the relationships between investing,
financing, and operating cash flows would indicate the major source of cash for the com-
pany. If a company is on the verge of bankruptcy for financial reasons, then one would
expect to observe declining operating cash flows and increasing cash inflows from fi-
nancing and investing. Probing which components of financing and investing cash flows
is providing the cash would reveal information not discernible from the income statement
and the balance sheet. An example of the usefulness of reported cash flow information for
financial analysis is provided in Sharma (1996). The extent to which cash flows vary
Managerial Finance 22

across industries and internationally is a subject worthy of research. An opportunity ex-


ists for future researchers to conduct a behavioural experiment investigating the effects of
reported cash flow information, accrual information, and a complementary information
set comprised of cash flow and accrual information in a financial distress prediction task,
a lending decision task and in an investment decision task. Only one research to date
(Sharma and Iselin, 2000) has considered the usefulness of cash flow information in such
contexts, the results of which indicate significant information usefulness in cash flow
data for assessing corporate solvency. Finally, conditions under which cash flow infor-
mation provides valuable information would be insightful. For example, under high task
uncertainty (where accrual information does not clearly indicate financial distress) the
provision of cash flow information may add value whereas under low task uncertainty its
incremental information content may be insignificant.
Volume 27 Number 4 2001 23

Endnotes

1. Gahlon and Vigeland (1988) do not disclose what this ratio represents.

2. For a description of Lawson’s identity in the context of corporate bankruptcy, see Aziz
and Lawson (1988).

3. Gentry et al, (1985a and 1985b) used Helfert’s funds flow model, Aziz et al, (1988)
and Aziz and Lawson (1989) used Lawson’s cash flow identity model, and Gahlon and
Vigeland used a credit agencies cash flow model.
Managerial Finance 24

References
Altman, E.I. (1968) ‘Financial Ratios, Discriminant Analysis and the Prediction of Cor-
porate Bankruptcy’, The Journal of Finance, Vol. XXIII, No. 4, pp.589-609.
Altman, E.I., Haldeman, R.G. and Narayanan, P. (1977) ‘Zeta Analysis: A New Model to
Identify Bankruptcy Risk of Corporations’, Journal of Banking and Finance, Vol. 1,
pp.29-54.
Altman, E.I. (1983) ‘Accounting Implications of Failure Prediction Models’, Journal of
Accounting, Auditing & Finance, pp.4-19.
Austin, L.M. and Bradbury, M.E. (1995) ‘The Accuracy of Cash Flow Estimation Proce-
dures’, Accounting and Finance, Vol. 35, No. 1, pp.73-86.
Aziz, A., Emanuel, D.C. and Lawson, G.H. (1988) ‘Bankruptcy Prediction - An Investi-
gation of Cash Flow Based Models’, Journal of Management Studies, Vol. 25, No. 5,
pp.419-435. Aziz, A. and Lawson, G.H. (1989) ‘Cash Flow Reporting and Financial Dis-
tress Models: Testing of Hypotheses’, Financial Management, Vol. 18, No. 1, pp.55-63.
Beaver, W.H. (1966) ‘Financial Ratios as Predictors of Failure’, Journal of Accounting
Research (Supplement), pp.71-111.
Blum, M. (1974) ‘Failing Company Discriminant Analysis’, Journal of Accounting Re-
search, Spring, pp.1-25.
Bowen, R.M., Burgstahler, D. and Daley, L.A. (1986) ‘Evidence on the Relationships
Between Earnings and Various Measures of Cash Flow’, The Accounting Review, Vol.
LXI, No. 4, pp.713-725.
Bowen, R.M., Burgstahler, D. and Daley, L.A. (1987) ‘The Incremental Information
Content of Accrual Versus Cash Flows’, The Accounting Review, Vol. LXII, No. 4,
pp.723-747.
Casey, C.J. and Bartczak, N.J. (1984) ‘Cash Flow - It’s Not The Bottom Line’, Harvard
Business Review, pp.61-66.
Casey, C. and Bartczak, N. (1985) ‘Using Operating Cash Flow Data to Predict Financial
Distress: Some Extensions’, Journal of Accounting Research, Vol. 23, No. 1, pp.384-
401.
Charitou, A. and Venieris, G. (1990) ‘The Need for Cash Flow Reporting: Greek Evi-
dence’, The British Accounting Review, Vol. 22, No. 2, pp.107-117.
Climo, T.A. (1976) ‘Cash Flow Statements for Investors’, Journal of Business Finance
and Accounting, Autumn, pp.3-14.
Coughlan, J.W. (1960) ‘Contrast Between Financial-Statement and Discounted-Cash-
Flow Methods of Comparing Projects’, NAA Bulletin, June, pp.5-17.
Coughlan, J.W. (1962) ‘Accounting and Capital Budgeting’, The Business Quarterly,
Fall, pp.39-48.
Volume 27 Number 4 2001 25

Coughlan, J.W. (1964) ‘Funds and Income’, NAA Bulletin, September, pp.23-34.
Dambolena, I.G. and Shulman, J.M. (1988) ‘A Primary Rule for Detecting Bankruptcy:
Watch the Cash’, Financial Analysts Journal, pp.74-78.
Deakin, E.B. (1972) ‘A Discriminant Analysis of Predictors of Business Failure’, Journal
of Accounting Research, pp.167-179.
DeAngelo, H. and DeAngelo, L. (1999) ‘Dividend Policy and Financial Distress: An Em-
pirical Investigation of Troubled NYSE Firms’, The Journal of Finance, Vol. XLV, No.
5, pp.1415-1431.
Dharan, B.G. and Lev, B. (1993) ‘The Valuation Consequence of Accounting Changes:
A Multi-Year Examination’, Journal of Accounting, Auditing, and Finance, Vol. 8, No.
4, pp.475-494.
Deitrick, J.W. and Stamps, J.L. (1981) ‘The Use of Accounting Information by Bank
Loan Officers’, The Journal of Commercial Bank Lending, November, pp.51-62.
Edey, H.C. (1963) ‘Accounting Principles and Business Reality’, Accountancy, Novem-
ber,pp.998-1002.
Fulmer, J.G., Gavin, T.A. and Bertin, W.J. (1991) ‘What Factors Influence the Lending
Decision: A Survey of Commercial Loan Officers’, Commercial Lending Review, Vol. 7,
No. 1, pp.64-70.
Gadenne, D.L. (1995) A Comparative Study of Accounting Information Users and Statis-
tical Models in a Financial Distress Prediction Task, Ph.D Dissertation, Griffith Univer-
sity.
Gadenne, D.L. and Iselin, E.R. (1995) ‘The Prediction of Corporate Financial Distress:
The Comparative Performance of Humans and Models, Accountability and Performance,
Vol. 1, No. 1, pp.1-18.
Gahlon, J.M. ad Vigeland, R.L. (1988) ‘Early Warning Signs of Bankruptcy Using Cash
Flow Analysis’, The Journal of Commercial Bank Lending, pp.4-15.
Gentry, J.A., Newbold, P. and Whitford, D.T. (1985a) ‘Classifying Bankrupt Firms with
Funds Flow Components’, Journal of Accounting Research, Vol. 23, No. 1, pp.146-161.
Gentry, J.A., Newbold, P. and Whitford, D.T. (1985b) ‘Predicting Bankruptcy: If Cash-
flow’s Not The Bottom Line, What Is?’, Financial Analysts Journal, pp.47-56.
Gilbert, L.R., Menon, K. and Schwartz, K.B. (1990) ‘Predicting Bankruptcy for Firms in
Financial Distress’, Journal of Business Finance & Accounting, Vol. 17, No. 1, pp.161-
171.
Gombola, M.J. and Ketz, J.E. (1983) ‘A Note on Cash Flow and Classification Patterns of
Financial Ratios’, The Accounting Review, Vol. LVIII, No. 1, pp.105-114.
Gombola, M.J., Haskins, M.E., Ketz, J.E. and Williams, D.D. (1987) ‘Cash Flow in
Bankruptcy Prediction’, Financial Management, pp.55-65.
Managerial Finance 26

Hawkins, D.F. (1977) ‘Toward An Old Theory of Equity Valuation’, Financial Analysts
Journal, November-December, pp.48-53.
Heath, L.C. (1978) Financial Reporting and the Evaluation of Solvency, New York:
American Institute of CPAs.
Heath, L.C. and Rosenfield, P. (1979) ‘Solvency: The Forgotten Half of Financial Re-
porting’, Journal of Accountancy, pp.48-54.
Helfert, E.A. (1972) Techniques of Financial Analysis, Irwin: Homewood, Illinois.
Hosmer, D.W. and Lemeshow, S. (1989) Applied Logistic Regression, John Wiley: New
York.
Ijiri, Y. (1979) ‘A Simple System of Cash-Flow Accounting’, in Accounting for a Simpli-
fied Firm Owning Depreciable Assets (eds.) Sterling, R.L. and Thomas, A.L. Houston:
Scholars Book Company.
Jones, C.J. (1975) ‘Accounting Standards: A Blind Alley?’, Accounting and Business Re-
search, Autumn, pp.273-279.
Jones, F.L. (1987) ‘Current Techniques in Bankruptcy Prediction’, Journal of Accounting
Literature, Vol. 6, pp.131-164.
Kochanek, R.F. and Norgaard, C.T. (1988) ‘Analyzing the Components of Operating
Cash Flow: The Charter Company’, Accounting Horizons, March, pp.58-66.
Kuhn, T.S. (1970) The Structure of Scientific Revolutions, The University of Liverpool
Press: Liverpool.
Laitinin, E.K. (1994) ‘Traditional Versus Operating Cash Flow in Failure Prediction’,
Journal of Business Finance & Accounting, Vol. 21, No. 2, pp.195-217.
Largay, III, J.A. and Stickney, C.P. (1980) ‘Cash Flows, Ratio Analysis and the W.T.
Grant Company Bankruptcy’, Financial Analysts Journal, pp.51-54.
Lawson, G.H. (1968) Applications of a Business Theory, University of Liverpool Press,
Liverpool.
Lee, T.A. (1971) ‘Goodwill - An Example of Will-o’-the Wisp Accounting’, Accounting
and Business Research, Autumn, pp.318-328.
Lee, T. (1982) ‘Laker Airways - The Cash Flow Truth’, Accountancy, pp.115-116.
Lee, T.A. (1981) ‘Cash Flow Accounting and Corporate Financial Reporting’, in Essays
in British Accounting Research, M. Bromwich and A. Hopwood (eds.), Pitman Publish-
ing Limited, London, Ch.3, pp.63-78.
Lilien, S., Mellman, M. and Pastena, V. (1988) ‘Accounting Changes: Successful versus
Unsuccessful Firms’, The Accounting Review, Vol. LXIII, No. 4, pp.642-656.
Littleton, A.C. and Zimmerman, V.K. (1963) Accounting Theory: Continuity and
Change, Prentice-Hall.
Volume 27 Number 4 2001 27

Marais, M.L., Patell, J.M. and Wolfson, M.A. (1984) ‘The Experimental Design of Clas-
sification Models: An Application of Recursive Partitioning and Bootstrapping to Com-
mercial Loan Classifications’, Journal of Accounting Research, Vol. 22, (Supplement),
pp.87-114.
Mensah, Y.M. (1983) ‘The Differential Bankruptcy Predictive Ability of Specific Price
Level Adjustments: Some Empirical Evidence’, The Accounting Review, Vol. LVIII, No.
2, pp.228-245.
Murdoch, B. and Krause, P. (1989) ‘An Empirical Investigation of the Predictive Power
of Accrual and Cash Flow Data in Forecasting Operating Cash Flow’, Akron Business
and Economic Review, pp.100-113.
Murdoch, B. and Krause, P. (1990) ‘Further Evidence on the Comparative Ability of Ac-
counting Data to Predict Operating Cash Flows’, The Mid-Atlantic Journal of Business,
Vol. 26, No. 2, pp.114.
Neill, J.D., Schaefer, T.F., Bahnson, P.R. and Bradbury, M.E. (1991) ‘The Usefulness of
Cash Flow Data: A Review and Synthesis’, Journal of Accounting Literature, Vol. 10,
pp.117-150.
Norton, C.L. and Smith, R.E. (1979) ‘A Comparison of General Price Level and Histori-
cal Cost Financial Statement in Prediction of Bankruptcy’, The Accounting Review, Janu-
ary, pp.72-87.
Ohlson, J.A. (1980) ‘Financial Ratios and the Probabilistic Prediction of Bankruptcy’,
Journal of Accounting Research, Vol. 18, No. 1, pp.109-131.
Rizzi, J.V. (1994) ‘Determining Debt Capacity’, Commerical Lending Review, Vol. 9,
No. 2, pp.25-34.
Schwartz, K.B. (1982) ‘Accounting Changes by Corporations Facing Possible Insol-
vency’, Journal of Accounting, Auditing & Finance, pp.33-43.
Sharma, D.S. (1995) ‘Accounting Policy Changes and Cash Flows’, Accounting Forum,
Vol. 19, No. 1, pp.54-63.
Sharma, D.S. (1996) ‘Analysing the Statement of Cash Flows’, Australian Accounting
Review, pp.37-44.
Sharma, D.S. and Stevenson, P.A. (1997) ‘The Impact of Impending Corporate Failure
on the Incidence and Magnitude of Discretionary Accounting Policy Changes’, The Brit-
ish Accounting Review, Vol. 29, pp.129-153.
Sharma, D.S. and Iselin, E. (2000) ‘The Decision Usefulness of Reported Cash Flow In-
formation Versus Accrual Information for Banker’s Assessment of Corporate Solvency’,
Western Decision Sciences Conference, Maui, Hawaii.
Simons, K. (1994) ‘The Relationship Between Dividend Changes and Cash Flow: An
Empirical Analysis’, Journal of Business Finance and Accounting, Vol. 21, No. 4,
pp.577-587.
Managerial Finance 28

Sondhi, A.C., Sorter, G.H. and White, G.I. (1987) ‘Transactional Analysis’, Financial
Analysts Journal, pp.57-64.

Stancill, J.M. (1987) ‘When Is There Cash in Cash Flow?’, Havard Business Review,
March-April, pp.38-48.

Staubus, G.J. (1961) A Theory of Accounting to Investors, Berkley: University of Califor-


nia Press.

Staubus, G.J. (1966) ‘Alternative Asset Flow Concepts’, The Accounting Review, July,
pp.397-412.

Stephens, R.G. and Govindarajan, V. (1990) ‘On Assessing a Firm’s Cash Generating
Ability’, The Accounting Review, Vol. 65, No. 1, pp.242-257.

Taffler, R.J. (1984) ‘Empirical Models for the Monitoring of UK Corporations’, Journal
of Banking and Finance, Vol. 8, pp.199-227.

Thomas, A.L. (1969) The Allocation Problem in Financial Accounting Theory, Studies in
Accounting Research No. 3, Evanston III: American Accounting Association.

Tweedie, D. and Whittington, G. (1990) ‘Financial Reporting: Current Problem and


Their Implications for Systematic Reform’, Accounting and Business Research, Winter,
pp.87-102.

Viscione, J.A. (1985) ‘Assessing Financial Distress’, The Journal of Commercial Bank
Lending, pp.39-55.

Ward, T.J. (1994) ‘An Empirical Study of the Incremental Predictive Ability of Beaver’s
Naive Operating Flow Measure Using Four-State Ordinal Models of Financial Distress’,
Journal of Business Finance & Accounting, Vol. 21, No. 4, pp.547-561.

Watson, I. (1996) ‘Financial Distress - The State of the Art in 1996’, International Jour-
nal of Business Studies, Vol. 4, No. 2, pp.39-65.

Watts, R. and Zimmerman, J. (1986) Positive Accounting Theory, New Jersey: Prentice-
Hall.

Winjum, J.O. (1972) The Role of Accounting in the Economic Development of England
1500-1750, Centre for International Education and Research.

Zavgren, C.V. (1983) ‘The Prediction of Corporate Failure: The State of the Art’, Journal
of Accounting Literature, Vol. 2, pp.1-37.

Zavgren, C.V. and Friedman, G.E. (1988) ‘Are Bankruptcy Prediction Models Worth-
while? An Application in Securities Analysis’, Management International Review, Vol.
28, No. 1, pp.34-43.

Вам также может понравиться