An Empirical Analysis of Useful Financial Ratios

Author(s): Kung H. Chen and Thomas A. Shimerda

Reviewed work(s):
Source: Financial Management, Vol. 10, No. 1 (Spring, 1981), pp. 51-60
Published by: Wiley-Blackwell on behalf of the Financial Management Association International
Stable URL: http://www.jstor.org/stable/3665113 .
Accessed: 17/11/2012 10:27

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Wiley-Blackwell and Financial Management Association International are collaborating with JSTOR to
digitize, preserve and extend access to Financial Management.

http://www.jstor.org

This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
An Empirical Analysis
of Useful Financial
Kung H. Chen and Thomas A. Shimerda
KungH. Chenis AssociateProfessorof Accountingat the
Universityof Nebraska-Lincoln,and ThomasA. Shimerdais
Assistant Professorof Accountingat CreightonUniversity.
* Financialratios have playedan importantpart in
evaluatingthe performanceandfinancialconditionof
an entity. Over the years, empirical studies have
repeatedlydemonstratedthe usefulnessof financial
ratios. For example, financially-distressed
firms can
be separatedfrom the non-failedfirms in the year
before the declarationof bankruptcyat an accuracy
rate of betterthan 90%by examiningfinancialratios
[1]. In determiningbondratings,whenfinancialratios
werethe only variablesused,the resultingratingswere
virtuallyidenticalwith institutionalratings [21].
There is one recurringquestionwith the use of fi-
nancialratios:whichratios, amongthe hundredsthat
can be computedeasily from the availablefinancial
data, shouldbe analyzedto obtainthe informationfor
the task at hand?We hope here to help resolve the
problemof ratio selectionby examiningratios found
useful in recent empirical studies, reconcilingthe
differencesin the ones found useful in these studies,
and categorizingthem by seven factors suggestedin
the literature.
There are many useful ratios reported in the
literature. Discriminationis needed to identify a
This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
Ratios
limited set of financial ratios. Naturally, different
researchersoften include different ratios. Conse-
quently, results on the usefulnessof specific ratios
vary. Exhibit1 summarizesa numberof such studies
and the ratios they employed.The 26 studiesanalyze
more than 100 financialitems, of which 65 are ac-
counting ratios. Forty-one of these are considered
usefuland/or are used in the final analysisby one or
more of the researchers.Givensuch a heterogeneous
set of usefulfinancialratios,the decision-maker has to
be at a loss in selectingwhichratiosto use for the task
at hand. Conceivably,41 ratios cannot all be signifi-
cant or equallyimportantin a multi-ratiomodel.The
decision-makermay hesitateto omit a ratio if it has
been found useful in one or more of the empirical
studies.
Yet, it is impossibleto includemost of the useful
ratios found in the literature.Which ratios, then,
should be deleted, and which should be included?
Should the results from only one study, the results
from a combinationof studies,or the resultsfrom all
the studiesbe used?If only one study is to be used,
whichone shouldit be?
51
 a

Exhibit 1. Financial Ratios Incorporated in Predictive Studies

z z 0 Z] z:2: rC) ?2 Zz I z' Z u o z Ir- ca 0') ca Id 2;
> i-- 0:: 0
0
I -~q~ a
O
H-~d H. +
a
c or)
: ; -3)t
I-- >
Co HH-]
C: toJ>>
~3
A rt a) <
\ rr
t>
H ~3
CI
r-f
C?1 rt
, I
FIRM I--,
N3
FAILURE
--i- J, J I I
1966 Tamari ~/?

1966 Beaver * *X
*** * X X X XXXXX X X X X XXXXXXXX
1968 Altman X X X X X X X XX X XX X XX
1972 Deakin * * * * * * , * * *
1972 Edmister X *X XX X X X X
1974 Blum * * *
1975 Libby * XXXX XX X XX X
1975 Elam * * ** X X "*' * X X XX X
BOND RATINGS
1966 Horrigan X X XX * X X * X X X X X
1969 Pogue &
Soldofsky *
1970 West
1973 Pinches & Mingo X X X X * X X
MARKE' T RETURN
1971 Martin * X *
1973 O'Connor * X * X * *X
MERGERS
1973 Stevens X X X X X X X*X X X X X
BETA
1973 Breen & Lerner X X
Sub- Found Useful* 6 6 3 5 3 4 5 1 131 2 5 1 2 1 2 112 2 1 1 143 2 1 1 3 1 2 11 11 1
Total Mentioned 10 7 6 7 7 6 8 1 4 8 1 31' 1 5 1 6 1 1 3 3 2 2 6 4 5 2 4 1 7 1 3 1 4 1 6 3 2 1 3 5 5 1 1 4 2 3 1 2 1 1
PRIOR TO 1965
1932 Fitzpatrick X X
1935 Winakor & Smith X X X
1942 Merwin X XX
1945 Chudson X X X X
1958 Hickman X X
1958 Saulnier X X
1961 Moore & Atkinson X X X X
1962 Jackendoff X X X X
1962 Wojenlower X X X
1963 Jen X X
? I - 7 i 1 .........
Subto tal
-- -- I Found
---
- - 1- 1 -i i, 7
D 4 / I I I I I I1 - 1 I I -I - I , r _ - , ,- - - - ??
Total Useful
--
----- - I , ii11i
-- i 3
- - - - - I -
-.
3 4 - -
1 - - ,
I
111
L I
-

*Ratio found useful in study; (X) Ratio mentioned in study; (1) Net Income plus Depreciation, Depletion, Amortization; (2) No Credit Interval = Quick A
Expense minus Depreciation, Depletion, Amortization; (3) Quick Flow = C+MS+AR+ (Annual Sales divided by 12)/[CGS-Depreciation+Selling and
divided by 12];(4) Cash Interval = C+-MS/Operating Expense minus Depreciation, Depletion, Amortization; (5) Defensive Interval = QA/Operating Ex
Depletion, Amortization; (6) Capital Expenditure/ Sales; (7) Non-operating Income before Taxes/ Sales.

This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
CHEN AND SHIMERDA/USEFUL
FINANCIALRATIOS
Most of the studiesreportedhighpredictivepowers
for theirratios.This wouldseemto suggestthat good
resultscan be attainedby usingthe usefulratiosfrom
any study. The researchercan afford the luxuryof
studying a selected group of firms whose fates are
known and then searchingfor the set of financial
ratios that has the highest predictivepower for the
known results. Those whose businessit is to make
predictions,though, cannot affordsuch a procedure.
Without further testing and re-examinationof the
findings,the result from any one study is applicable
onlyto firmswiththe samecharacteristicsas thosein-
cludedin that study.
Ideally, the financial ratios analyzed should be
selected on some theoretical basis, coupled with
demonstratedempiricalevidenceof their usefulness.
An acceptabletheoreticalfoundationfor the selection
of ratiosfor decision-makinghas yet to be found,and
the scattered heterogeneousempirical evidence in
publishedstudiesdoes not identifya completeset of
useful ratios. Althoughin this paperan approachto
obtain an efficient set of financial ratios will be
presented,we do not presentan absolutemodel for
selecting specific ratios. The result will suggest,
however,that a usefulset of ratios can be developed
from seven basic financialfactors.
Overlapping of Ratios
Manyof the ratiosincludedin the studiesarehighly
correlatedwith one other. Jackendoff [14] demon-
stratesthis overlapping:
Another type of redundancy arises from the use of
ratios which are easily derived from one another,
althoughthe componentsare not identicalas is truein
inversions. ... One of the most obvious sets of such
relatedratios includes:
1. Worthto total debt (or its inverse,total debt
to worth).
2. Worthto total assets.
3. Total debt to total assets.
These are simplyvariantsof the equation:
Exhibit 2. Data Reductionin Factor-AnalyzedFinancialRatio Space
Study VariableSpace
PinchesandMingo(1973) 35
Pinches,Mingo,andCaruthers(1973) 48
Stevens(1973) 20
Libby(1975) 14
Pinches,Eubank,Mingo,and
Caruthers(1975) 48
This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
53
Total Assets = Total Debt + Net Worth
[14, p. 7].
In spite of his extensive study, such overlapping can
still be foundin most of the recentstudies.For exam-
ple, the 56 items used in the computation of the 28
ratios included in the Elam study [9] are derived from
only 18 different pieces of financial data, and the 28
items for Deakin's ratios [7] consist of only 10
separate pieces of data. The elimination of such
overlapping would aid in the development of a useful
set of financial ratios.
Uses of Principal Component Analysis
Not all overlapping ratios, however, can be elim-
inated by visual inspection. A statistical tool designed
to summarize such inter-relationships is principal
component analysis. One of the functions performed
by principal component analysis is to group variables
into a few factors that retain a maximum of informa-
tion contained in the original variable set. This tool is
a useful first step for subsequent analyses. The use of
principal component analysis, along with other
statistical methods, produces a more powerful and
basic analysis [11, p. 319].
Five of the recent studies have employed principal
component analysis. The results of these analyses are
summarized in Exhibit 2. In each study, the number of
variables was significantly reduced from the original
set of variables, yet the reduced set of ratios still ac-
counted for the majority of the variance accounted for
by the original set of financial ratios. Pinches and
Mingo [21] reducedtheir data set for bond ratings
from 35 to 7 variables (an 80% reduction) and still ac-
counted for 63% of the variation in the original data
matrix. Stevens [27] reduced 20 variables to 6 (a 70%
reduction) and accounted for 82% of the total
variance. Libby [16] reducedhis 14-ratio set to 5
ratios (a 64% reduction)with very little loss in the
predictive ability of the model.
The attempt by Pinches, Mingo, and Caruthers [22;
%Reduction %Variation
FactorSpace In Space Still Explained
7 80 63
7 85 91, 92, 87, 92
6 70 82
5 64 Not Reported
7 85 92
54
hereafter, PMC] to develop an empiricallybased
classification of financial ratios resulted in seven
classificationsof ratios acrossindustries.Theseseven
patternsoccurredin each year examined,accounting
for a consistentlyhigh amountof the variancecon-
tainedin the originaldata matrix.The compositionof
these classificationsremainedrelativelystable over
the 19-year period studied. A subsequentstudy by
Pinches, Eubank, Mingo, and Caruthers[23; here-
after,PEMC]showedthe short-termstabilityof these
factors. They also demonstratedthat a hierarchical
classificationof empiricalfinancialratioscan be con-
structed.
Diversity of Factors
The PMC and PEMC studiessuggestthe existence
of commonratioclassificationsandofferan empirical
basis for groupingfinancialratios.Accordingto their
findings,financialratios can be representedby seven
factors- Returnon Investment,FinancialLeverage,
CapitalTurnover,Short-TermLiquidity,Cash Posi-
tion, InventoryTurnover,and ReceivablesTurnover.
A slightlydifferentset of factorsis foundin an earlier
study by Pinches and Mingo [21]. The studies by
Stevensand Libby mentionedabove also attemptto
derivea reducedset of financialratiosto representthe
originalset of financialratios, using differentsets of
factors from those in the PMC and PEMC studies;
these can be seen in Exhibit3. A total of 12 factorsis
suggestedin these studies,with each studyseemingly
proposinga differentset of factors to representthe
variablespace portrayedby the financialratios. For
example,of the four factorsfoundin Stevens'sstudy
only one is includedin the PMC and PEMC factors.
Similarly,only one of the five factors suggestedby
Libby is found among the factors in the PMC and
PEMC studies.
It seems that the resultsfromprincipalcomponent
analyses are as diverse as the financial ratios
themselves.Differentsets of factorscan be foundin
differentstudieswith very little commonalityamong
any of them. Using such results as a guide for the
selectionof financialratiosis hardlya viablesolution,
if not actuallya waste of time. Inevitably,beforethe
results from principal component analysis can be
appliedto studies on financialratios, a satisfactory
answerto the questionof what are the commonfac-
tors representingthe financial ratios must be for-
mulated.A detailedanalysisof the five studiesreveals
that some of the twelve factors vary in name only.
Theycan be describedaccordingto the sevencommon
factors suggestedin Exhibit4.
This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
FINANCIALMANAGEMENT/SPRING1981
A Reconciliation of Factors
To a great extent, the diversityof factorsreported
in the literaturecan be attributedto the differencein
variables included in the principal component
analyses, as indicated in the following detailed
analysis.
Profitability and
Return on Investment
Stevens [27] found five ratios - earningsbefore
interest and taxes (EBIT)/total assets, EBIT/sales,
earnings before taxes (EBT)/sales, net income/net
worth,and net income/totalassets- with highload-
ings' on his profitabilityfactor. Each of these ratios
also had a high loadingon the returnon investment
factor in the PMC and PEMC studies. The net in-
come/total assets ratio was the most representative
ratioamongthe ratiosin Libby's[16]profitabilityfac-
tor. The profitabilityfactorfoundby bothStevensand
Libbywas the sameas the returnon investmentfactor
in the PMC and PEMC studies.
Activity, Receivable Turnover,
and Capital Turnover
A slightly more confusingpictureis found in the
activity factor of the Stevens [27] study. Two ratios
havehighloadingson this factor- sales/quickassets
(S/QA) has a loadingof 0.794, and sales/total assets
(S/TA) has a loadingof 0.85. The S/QA ratio in the
PMC and PEMC studieshas a highfactorloadingon
the receivablesturnoverfactor,whilethe S/TA ratio
has a highloadingon the capitalturnoverfactor.Such
diverseresultscan be explainedby differencesin the
originalvariablesincludedin the principalcomponent
analysisin these studies.
Principalcomponentanalysisis variable-sensitive:
differentfactors may be obtainedif differentsets of
variables are fed into the principal component
analysis.Substantiallymoreratiosare includedin the
PMC and PEMC studies than in the Stevens study
[27]. Among the ratios included in the PMC and
PEMC studies are five ratios relatingto receivable
turnover,includingthe S/QA ratio. But Stevensin-
cludes only one of the five ratios in his analysis.
Becausethe variancesof these5 ratiosaresimilar,and
5 of a total of 48 ratios in a 7-factor resultwill cer-
'A factor loading represents the extent to which the variable is
related to the factor and is commonly thought of as the correlation
between the variable and the factor. For a more detailed discussion
on the meaning of factor loadings, see William D. Wells and J. N.
Sheth, "Factor Analysis in Marketing Research" [29].
 FINANCIALRATIOS
CHEN AND SHIMERDA/USEFUL 55

Exhibit 3. Factors Found in Five EmpiricalAnalysesof FinancialRatios

Pinches,Mingo, Pinches,Eubank,
Pinchesand andCaruthers Mingo,and
Factors Mingo(1973) (1973) Stevens** Libby(1975) Caruthers(1975)

Asset Balance X
Activity X X
Profitability X X
Liquidity X X
Cash Position X X X
Short-Term Liquidity X X
Receivables Turnover X X
Inventory Turnover X X
Return on Investment X X X
Short-Term Capital Intensiveness X X* X*
Long-Term Capital Intensiveness X X* X*
Financial Leverage X X X

*Thesestudieshad one factor,capitalturnover,whichincludedthe ratios fromthe abovetwo factors.

**Sevenfactorswerefoundin the PinchesandMingostudy.Variablesin threeof thefactors- size,debtanddebtcoveragestability,andearn-
ings stability- werenot ratiomeasures.Thustheywereexcluded.For the samereason,two factorsfoundin the Stevensstudy- dividend
policy and priceearnings- wereexcluded.

Exhibit 4. A Reconciliationof Factors DepictingFinancialRatios

SevenBasicFactors
Financial Capital Returnon Inventory Receivables S-T Cash
Study Leverage Turnover Investment Turnover Turnover Liquidity Position

PMC X X X X X X X
PEMC X X X X X X X
Stevens X Liquidity Profitability Activity
Libby Asset Profitability Activity Liquidity X
Balance

tainly account for a significantportion of the total
variance defined by the 48 ratios, the S/QA ratio
formsa separatefactorwithothersimilarratiosin the
PMC and PEMC studies.
On the otherhand,the S/QA ratiois the only ratio
out of the five ratiosthat is usedin the Stevensstudy.
The other four ratios representedby the receivable
turnoverfactor in the PMC and PEMC studies are
not included in Stevens's principal component
analysis. Consequently,the S/QA ratio will either
correlatewith one otherratio in a differentfactor or
not be representedat all in the factorsselected.Never-
theless, this evidencesuggeststhat the activityfactor
of the Stevensstudyis representedby factorsfoundin
the PMC and PEMC studies.
Liquidity, Asset Balance,
and Capital Turnover
Similarreasoningcan be appliedto the importance
of the current assets/total assets (CA/TA) ratio.
Stevensdoes not includeCA/TA as one of his ratios,
but he does use net working capital/total assets
(NWC/TA). This ratio loads heavily on the factor
Stevensdescribesas his liquidityfactor.This ratio is
mentionedbut is not includedin the final factors in
both the PMC and PEMC studies. A separate
analysis,whichwill be discussedin detaillater,reveals
that the NWC/TA ratiois highlycorrelated(r = 0.80)
with the currentassets/total assets ratio (CA/TA).
The resultof a principalcomponentanalysisindicates
that bothNWC/TA andCA/TA ratioshavehighfac-
tor loadings,0.85 and 0.91, respectively,on the same
factor. This indicatesthat the NWC/TA ratio could
be an important ratio in the same factor as the
CA/TA ratio were it includedin the same analysis.
In the PMC and PEMC studies,the CA/TA ratio
is amongthe ratioswith high loadingson the capital
turnover factor. Thus, the liquidity factor in the
Stevensstudyis similarto the capitalturnoverfactor
in the PMC and PEMC studies.Libbyalso uses the
CA/TA ratio to representone of his five factors.He

This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
56
denotesthe factorbest representedby CA/TA as his
asset balance factor. This means that one ratio,
CA/TA, is usedto describea capitalturnoverfactor,
an asset balance factor, and, if employed in the
Stevensstudy, could even describea liquidityfactor.
At first it appearssomewhatconfusingthat one
ratio can representthree basicallydifferentfactors.
This again can be reconciledif one examines the
originalratio samplefor the threestudies.Stevensin-
cludes only two ratios that are slightlyrelatedto li-
quidity- NWC/TA andNWC/S. As no otherratios
could representliquidity in his study, an obvious
choicefor the factorcontaininghighloadingson these
ratioshas to be liquidity.If Stevenshadstartedwitha
largervariableset, his liquidityfactorcouldverywell
not haveincludedNWC/TA. The ratiosStevenscalls
liquiditycould actuallyrepresentcapital turnoveras
describedby PMC and PEMC.
Libbystartedwith a 14-variableset. Six ratioshad
total assets as their denominator,and four of those
ratios had a currentasset item for their numerator.
Three other ratios related current asset items to
currentliabilities,while four ratios used sales as the
denominatorbase. Because of the restrictedset of
ratios employed and the predominateuse of total
assets as a denominatorbase, Libby'sresultwas not
surprising.Why he chose "asset balance"to describe
the factor representedby CA/TA was not reported,
nor were any of the ratios' factor loadings.Without
detailedinformationaboutthe ratiosand the factors,
furtheranalysisis impossible.
Giventhe set of ratiosincludedin both the Stevens
andLibbystudies,it is possiblefor the threefactors-
capital turnover,liquidity, and asset balance - to
capturebasicallythe same informationeven though
theirtitlesdiffer.The ratioCA/TA is commonto two
and significantlycorrelatedwith a ratio in the third
factor.Withthis commonbond,it is possibleto recon-
cile Stevens'sliquidityand Libby'sasset balancefac-
tors withPMC'sand PEMC'scapitalturnoverfactor.
Other Factors
Two other factors are quite easily reconciled.The
leverage factors in the PMC, PEMC, and Stevens
studiesare virtuallyidentical.In all threestudies,the
long-term debt/net worth, long-term debt/total
assets,andtotal liabilities/totalassetsratioshavehigh
loadingson the leveragefactor.The ratio with a high
loading on Libby's liquidity factor is current
assets/currentliabilities.This same ratio also has a
high loadingon the short-termliquidityfactorin the
This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
FINANCIALMANAGEMENT/SPRING1981
PMC and PEMC studies.
The final factorto reconcileis Libby'sactivityfac-
tor. The ratio designatedby Libby to representthis
factor is currentassets/sales. This same ratio has a
high loading on the inventoryturnoverfactor in the
PMC and PEMC studies.The cash positionfactoris
commonto the PMC and PEMCstudiesas well as to
Libby'sstudy. The ratio in Libby'sstudy,cash/total
assets,is also highlyloadedon the cashpositionfactor
in both the PMC and PEMC studies.Results of the
above analyses suggest that factors found in the
Stevens and the Libby studies, althoughdifferentin
name, are includedin the PMC and PEMC factors.
The evidencewe havedescribedso farconfirmsthat
financialratioscan be groupedandrepresentedby the
seven common factors defined in the PMC and
PEMC studies. Each factor represents a unique
dimension in the description of financial charac-
teristicsof a businessfirm.This findingoffersa possi-
ble reconciliationof the diverseresultsof variousem-
piricalstudies.This resultcan be appliedto reconcile
the seeminglydisorganizedstate of useful financial
ratios in the severalstudieson the predictionof firm
failurereportedin the literature.
Important Factors in
Firm-Failure Prediction
Thirty-fourfinancial ratios have been found by
researchersto be significantvariablesin the prediction
of firmfailurein recentstudies.If a smallernumberof
ratios could still convey substantially the same
amountof information,the task of usingthem would
be easier.
If the seven-factorspacefromthe PMC andPEMC
studies is used as the basis for classifyingfinancial
ratios,all but ten of the ratiosfoundusefulin the firm
failurepredictionstudiescan be classifiedby one of
the seven factors. Exhibit 5 summarizesthe seven
studies and their ratios. Ten ratios could not be
classifiedwithoutfurtheranalysisbecausethey were
not includedin the finalfactorsof the PMC or PEMC
study. They are quick assets/inventory, net in-
come/common equity, quick flow ratio,2 funds
flow/current liabilities, net income/sales, funds
flow/total debt, working capital/total assets,
long-termdebt/currentassets,no-creditinterval,and
retainedearnings/totalassets.
2Blum [4, p. 16] defines quick flow as [Cash + Notes Receivable +
Market Securities + (Annual Sales - 12)] - [(Cost of Goods Sold
- Depreciation Expense + Selling and Administration Expense +
Interest) - 12].
CHEN AND SHIMERDA/USEFUL
FINANCIALRATIOS 57

Exhibit S. Factor Classification of Important Ratios for Predicting Firm

Failure as Found in Recent Empirical Studies

Studyby
Factor Ratio Beaver Altman Deakin Edmister Blum Elam Libby

Return Net Income/Sales* x

on Funds Flow/NW x
Investment Funds Flow/TA x
Net Income/TA x X X
Net Income/NW x
EBIT/Sales x
EBIT/TA x
NI/Common Equity** x
Capital QA/TA X
Turnover Funds Flow/Sales x
CurrentAssets/TA X X
Net Worth/Sales X
Sales/TA x x
WC/TA* x x X
Financial Total Liabilities/TA x X x
Leverage Total Liabilities/NW x x
Long-Term Debt/CA** x
Funds Flow/TD** X X x x
Funds Flow/CL** X
Retained Earnings/TA** x
Short-Term CurrentAssets/CL x x X X
Liquidity Quick Assets/CL x X X
Current Liabilities/NW X
Current Liabilities/TA x
Cash Cash/Sales x
Position Cash/Total Assets x x
Cash/Current Liabilities x x
No Credit Interval** x
Quick Flow** X
Inventory Current Assets/Sales x x
Turnover Inventory/Sales x
Sales/Working Capital x x
Receivables Quick Assets/Inventory** X
Turnover Quick Assets/Sales x
*Ratio not includedin the final factorsof the PEMC studies.
**Rationot in the 48 ratiosincludedin the PEMC study.

Empirical Results tor loadings of the paired ratios on a common factor

To study the relationship of the ten ratios to the are also reported.
other ratios included in the factors, we conducted a
Net Income/Sales,
principal component analysis of 39 ratios for the firms
included in the COMPUSTAT tape. A total of 1,053 Net Income/Common Equity,
firms with complete data for both total assets and net and Return on Investment
sales in 1977 was included. The results from the test The result of the analysis shows that two of the ten
are summarized in Exhibit 6. Each of the ten ratios is ratios correlate highly with ratios representing the
paired with the ratio from the PEMC study for which return on investment factor in the PEMC study. Net
its product moment correlation is the highest. The fac- income/sales is highly correlated (r = .98) with

This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
58 FINANCIALMANAGEMENT/SPRING1981

Exhibit 6. Factor Loading and Product-Moment Coefficient of Correlation of Selected Ratios in 1977

Factor Factor Correlation Levelof

UnclassifiedRatio Loading ClassifiedRatio Loading Factor Coefficient Significance

Net Income EBIT Returnon

Sales .90 Sales .86 Investment .98 .001
Net Income Net Income Returnon
CommonEquity .62 Net Worth .87 Investment .39 .001
WorkingCapital CurrentAssets Capital
TotalAssets .85 TotalAssets .91 Turnover .80 .001
FundsFlow Net Worth Financial
TotalDebt .92 Total Debt .89 Leverage .88 .001
FundsFlow Net Worth Financial
CurrentLiabilities .91 Total Debt .89 Leverage .84 .001
Long-TermDebt Long-TermDebt Financial
CurrentAssets -.81 TotalAssets .63 Leverage -.31 .001
RetainedEarnings TotalDebt Financial
TotalAssets .62 TotalAssets -.71 Leverage -.72 .001
Cash Cash
No CreditInterval .82 Sales .85 Position .80 .001
Cash Cash
QuickFlow -.86 Sales .85 Position -.58 .001
QuickAssets Receivables Receivable
Inventory .98 Inventory .97 Turnover .98 .001

EBIT/Sales with factor loadings of .90 and .86,
respectively, on the same factor. Net income/com-
mon equity and net income/net worth have high load-
ings (.62 and .87, respectively) on the same factor and
are significantly correlated (r = .39). This is not sur-
prising, since common equity and net worth differ
only by the amount of preferred stock outstanding.
The ratios EBIT/Sales and net income/net worth
have high loadings on PEMC's return on investment
factor. Thus the two ratios that correlate with these
ratios could be classified as ratios exhibiting return on
investment characteristics.3
Working Capital/Total Assets
and Capital Turnover
The ratio, workingcapital/total assets, does not
have a high factorloadingscore on any of the factors
in the PEMC study. The ratio is significantly corre-
lated with the ratio, currentassets/total assets. The
current asset/total assets ratio is represented in the
'In the present analysis, all but net income/common equity and net
income/net worth load on the same factor. These two ratios load on
a separate factor. This slight difference can be attributed to
differences in the original variable space.
capital turnover factor in the PEMC study. The ratio
WC/TA is thus classified as a capital turnover ratio.
Ratios Pertaining to Financial Leverage
Four of the unclassifiedratios correlateand load
highly with ratios classifiedin the financialleverage
factor of the PMC and PEMC studies.Two of these
ratios, funds flow/total debt and funds flow/current
liabilities,are highly correlated(.88 and .84, respec-
tively)withthe net worth/totaldebtratio.Theirload-
ings on a commonfactorare .92 for fundsflow/total
debt, .91 for fundsflow/currentliabilities,and .89 for
net worth/total debt. The two ratios, long-term
debt/currentassetsandretainedearnings/totalassets,
can also be regardedas representative of the financial
leveragefactor.Theyload and are significantlycorre-
lated with ratios that are part of PEMC's financial
leveragefactor.
The ratio,retainedearnings/totalassets,was one of
Altman's [1] most significant ratios in predicting
bankruptcy,but it has not been incorporatedin any
other study on firm failure. The results of factor
analysis applied to the 1977 data confirm its impor-
tance in the financial leverage factor.

This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
CHEN AND SHIMERDA/USEFUL
FINANCIALRATIOS
No-Credit Interval, Quick Flow,
and Cash Position
Two previously unclassified ratios, no-credit
intervalandquickflow, correlatewithcash/sales.The
coefficients of correlation are .80 and -.58, respec-
tively. Furtherevidencethat these two ratios can be
grouped with ratios that measure cash position is
providedin their factorloadings.The factorloadings
of these two ratioson the cash positionfactorare .82
and -.86, respectively.
Quick Asset/Inventory
and Receivables Turnover
The last ratio to classifyis quick assets/inventory.
This ratio correlateshighly with receivables/inven-
tory (r = .98), and both ratioshave high loadingson
the same factor. The ratio, receivables/inventory,
loads heavilyon the receivablesturnoverfactorin the
PEMC study. Becauseof this relationship,the ratio
quick assets/inventoryis classified in Exhibit 6 as
belongingto the receivablesturnoverfactor.
Based on the PMC and PEMC resultsand supple-
mentedwiththe analysisof the 1977data,all 34 finan-
cial ratiosthat werefoundusefulin the variouspredic-
tive studieson bankruptcycan be assignedby one of
the seven major factors. Becauseratios belongingto
the same factor are highly correlated and reveal
primarilythe sameinformation,a decision-makercan
select an appropriateset of financialratios that best
representsthese seven factors for the predictionof
firm failure.
Conclusions
Thispaperdemonstratesthat the financialratiosin-
vestigated in previous predictive studies of bank-
ruptcy can be classified by a substantiallyreduced
numberof factors.The ratios classifiedby the same
factor are highlycorrelated,and the selectionof one
ratioto representa factorcan accountfor most of the
informationprovidedby all the ratios of that factor.
Inclusionof morethanone ratiofroma factorleadsto
multicollinearityamong ratios and distorts the rela-
tionship between the dependent and independent
variables.In addition,highcorrelationbetweenratios
causesthe resultsto be sample-sensitiveand possibly
misleading. It is important then that a minimum
numberof ratios, one ratio in most cases, be selected
to representeachfactorfor furtherstatisticalanalysis.
Still, the questionof whichratio shouldrepresenta
factor has yet to be resolved.The popularprocedure
This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
59
of selectingthe ratio with the highestabsolutefactor
loading makes the selectionsensitiveto the sample.
Such a proceduremay be satisfactoryfor data reduc-
tion purposes,but it is certainlynot satisfactoryfor
model buildingor theoryconstruction.Concertedef-
fort shouldbe appliedto the processof selectingthe
most representativeratios of these factors.
The selectionof the best representativeratio for a
factor is not independentof the ratios selected for
otherfactors.Eachratio containscommonas well as
unique information.The common informationcon-
tainedin a ratio is representedby factors.The unique
informationis not sharedby anyotherratioin the fac-
tor. Consequently,the set of financialratios used for
furtheranalysisshouldbe selectedin such a way that
the ratios capturemost of the common information
containedin their factors and, as a group, contain
more of the uniqueinformationthan any otherset of
ratios. Unfortunately,such a theory is yet to be
developed.Empiricalevidencehas so far been con-
cernedwith the extractionof common factors.
No study has reportedon the type and amountof
unique information contained in a ratio. Future
studies on this problem will certainly enhance the
usefulnessof financialratios and facilitatethe selec-
tion of appropriateratios for decision-making.
References
1. EdwardI. Altman, "FinancialRatios, Discriminant
Analysisandthe Predictionof CorporateBankruptcy,"
Journalof Finance(September1968),pp. 589-609.
2. WilliamH. Beaver,"AlternativeAccountingMeasures
as Predictors of Failure," The Accounting Review
(January1968),pp. 113-122.
3. WilliamH. Beaver,"FinancialRatios as Predictorsof
Failure,"EmpiricalResearchin Accounting:Selected
Studies, 1966,Journalof AccountingResearch(1967),
pp. 71-111.
4. MarcP. Blum,"TheFailingCompanyDoctrine,"Jour-
nal of AccountingResearch(Spring 1974),pp. 1-25.
5. WilliamJ. Breenand EugeneM. Lerner,"Corporate
FinancialStrategiesand MarketMeasuresof Risk and
Return,"Journalof Finance(May 1973),pp. 339-351.
6. WalterA. Chudson,The Patternof CorporateFinan-
cial Structure, New York, National Bureau of
EconomicResearch,1945.
7. Edward B. Deakin, "A DiscriminantAnalysis of
Predictorsof BusinessFailure,"Journalof Accounting
Research(Spring 1972),pp. 167-179.
8. RobertO. Edmister,"An EmpiricalTest of Financial
Ratio Analysisfor Small BusinessFailurePrediction,"
Journalof Financialand QuantitativeAnalysis(March
1972),pp. 1477-1493.
60
9. Rick Elam, "The Effect of Lease Data on the Predictive
Ability of Financial Ratios," The Accounting Review
(January 1975), pp. 25-43.
10. Paul J. Fitzpatrick, Symptoms of Industrial Failures,
Washington, D.C., Catholic University of America
Press, 1931.
11. Richard Gorsuch, Factor Analysis, Philadelphia, W. B.
Saunders Company, 1974.
12. W. Braddock Hickman, Corporate Bond Quality and
Investor Experience, Princeton, N.J., Princeton Uni-
versity Press, 1958.
13. James O. Horrigan, "Some Empirical Bases of Finan-
cial Ratio Analysis," The Accounting Review (July
1965), pp. 558-568.
14. Nathanial Jackendoff, A Study of Published Industry
Financial and Operating Ratios, Small Business
Management Research Report, Philadelphia, Temple
University, 1962.
15. Frank Chifeng Jen, "The Determinants of the Degree of
Insufficiency of Bank Credit to Small Business," Jour-
nal of Finance (December 1963), pp. 694-695.
16. Robert Libby, "Accounting Ratios and the Prediction
of Failure: Some Behavioral Evidence," Journal of Ac-
counting Research (Spring 1975), pp. 150-161.
17. Alvin Martin, "An Empirical Test of the Relevance of
Accounting Information for Investment Decisions,"
Journal of Accounting Research, Supplement (1971),
pp. 1-31.
18. Charles L. Merwin, Financing Small Corporations,
New York, National Bureau of Economic Research,
1942.
19. Geoffrey H. Moore and Thomas R. Atkinson, Risks
and Returns in Small Business Financing, 41st Annual
Report, New York, National Bureau of Economic
Research, 1961.
20. Melvin C. O'Connor, "On the Usefulness of Financial
Ratios to Investors in Common Stock," The Account-
ing Review (April 1973), pp. 339-352.
WESTERN FINANCE ASSOCIA TION
1981 MEETINGS
Date: June 18-20, 1981
Place: Jackson Lodge, Jackson Hole, Wyoming, in the Grand Tetons
Program: Professor James C. Van Home
1981 WFA Program Chairman
Graduate School of Business
Stanford University
Stanford, California 94305
This content downloaded by the authorized user from 192.168.72.227 on Sat, 17 Nov 2012 10:27:57 AM
All use subject to JSTOR Terms and Conditions
FINANCIALMANAGEMENT/SPRING1981
21. George E. Pinches and Kent A. Mingo, "A Multivariate
Analysis of Industrial Bond Ratings," Journal of
Finance (March 1973), pp. 1-18.
22. George E. Pinches, Kent A. Mingo, and J. Kent
Caruthers, "The Stability of Financial Patterns in In-
dustrial Organizations," Journal of Finance (May
1973), pp. 389-396.
23. George E. Pinches, A. A. Eubank, Kent A. Mingo, and
J. Kent Caruthers, "The Hierarchical Classification of
Financial Ratios," Journal of Business Research (Oc-
tober 1975), pp. 295-310.
24. T. F. Pogue and R. M. Soldofsky, "What's in a Bond
Rating," Journal of Financial and Quantitative
Analysis (June 1969), pp. 201-228.
25. Raymond I. Saulnier, Neil H. Jacoby, and Harold G.
Halcrow, Federal Lending and Loan Insurance,
Princeton, N.J., Princeton University Press, 1958.
26. Raymond F. Smith and Arthur H. Winakor, Changes in
the Financial Structure of Unsuccessful Corporations,
Urbana, Ill., Bureau of Business Research, University
of Illinois, 1935.
27. D. L. Stevens, "Financial Characteristics of Merged
Firms: A Multivariate Analysis," Journal of Financial
and Quantitative Analysis (March 1973), pp. 149-158.
28. Meir Tamari, "Financial Ratios as a Means of Fore-
casting Bankruptcy," Management International
Review (1966), pp. 15-21.
29. William D. Wells and J. N. Sheth, "Factor Analysis in
Marketing Research," in Multivariate Analysis in
Marketing, David A. Aaker, ed., Belmont, California,
Wadsworth Publishing Company, Inc., 1971, pp.
212-227.
30. Richard R. West, "An Alternative Approach to
Predicting Corporate Bond Ratings," Journal of Ac-
counting Research (Spring 1970), pp. 15-21.
31. Albert M. Wojinlower, The Quality of Bank Loans: A
Study of Bank Examination Records, New York,
National Bureau of Economic Research, 1962.