Financial ratio variability and industry classification

The initial location at the University of Vaasa, Finland
http://lipas.uwasa.fi/~ts/publicat/FinRatioVarAndIndClassif.pdf

Salmi, Timo & Roy Dahlstedt & Martti Luoma & Arto Laakkonen (1986).
Financial ratio variability and industry classification. The Finnish Journal of
Business Economics 35:4, 333-356.

This paper is reproduced at the University of Vaasa in the electronic format with
the permission of The Finnish Journal of Business Economics. Copyright © 1986
by The Finnish Journal of Business Economics and the authors.

http://lta.hse.fi/

http://www.uwasa.fi/laskentatoimi/english/personnel/salmitimo/
http://www.hse.fi/FI/HKI/D/Roy_Dahlstedt/
http://lipas.uwasa.fi/~mjl/

---

LIIketaloudellinen
Aikakauskirja
Företagsekonomisk Tidskrift
The Finnish Journal of Business Economics
4·1986
Bertil Näslund
Some views on the current state of financial theory
F. H. Rolf Seringhaus
Market entry and the impact of export marketing assistance: A con·
ceptual approach to causal modelling
Karin Holstius
Sales response to advertising
Vesa Kanniainen
On the effects of inflation: The debtor-creditor
hypothesis recon-
sidered
Erkki K. Laitinen
Financial Ratio Model of a Public Enterprise
Timo Salmi -
Roy Dahlstedt -
Martti Luoma -
Arto Laakkonen
Financial ratio variability and industry classification
Erkki K. Laitinen
Suomalaisten ja ulkomaisten vuosikertomusten välittämä yrityskuva
(Summary on page 374)
Kauko Hahtola
Onko toiminta-analyyttinen
tutkimusote
törmännyt luonnontieteel-
lisen filosofian
muuriin?
(Summary on page 383)
Toimituskunta
JAAKKO HONKO (puh.joht. -
ordförande -
chairman), PERTII KETIUNEN,
Redaktionsråd
CAJ·GUNNAR LINDSTRÖM, VESA MÄKINEN, KALEVI PIHA, REIJO RUUHELA,
Board of Editors
FEDI VAIVIO, LARS WAHLBECK.
Toimitus
HUUGO RANINEN (päätoimittaja
-
huvudred. -
editor)
Redaktion
MIKA KASKIMIES (toim.siht. -
red.sekr. -
secretary)
Editors
The Finnish Journai of Business Economics
Address: Runeberginkatu 22-24,
00100 Helsinki 10, Finland

---

LIIketaloudellinen
Aikakauskirja
Företagsekonomisk
Tidskrift
The Finnish Journal of Business Economics
Toimituskunta
JAAKKO HONKO (puh.joht. -
ordförande-chairman),
PERTTI KETTUNEN, CAJ·
Redaktionsråd
GUNNAR LINDSTRÖM, VESA MÄKINEN, KALEVI PIHA, REIJO RUUHELA, FEDI
Board of Editors
VAIVIO, LARS WAHLBECK
Toimitus
HUUGO RANINEN (päätoimittaja -
huvudred.-editor),
Redaktion
MIKA KASKIMIES (toim.siht. -
red.sekr.-secretary)
Edltors
The FInnish Joumaf
of Business
Economics
Address: Runeberginkatu 22-24,
00100 Helsinki 10. Finland

---

TIMO SALMI, Professor of Accounting
and Business Finance
ROY DAHLSTEDT, Acting Professor
of Economics
MARTTI LUOMA, Associate Professor
of Statistics
ARTO LAAKKONEN, Lic.Sc. (Econ.),
Accounting and Business Finance
School of Business Studies
University of Vaasa
Financial ratio variability and industry c1assification *
In financial statement analysis it is common to compare the various financial
ratios of a business firm with industry averages or with other firms in the same
branch of industry.
This practice has direct implications on financial statement analysis. The eval-
uation of a firm's performance and financial status becomes strongly dependent
on the financial ratios of the industry, and particularly dependent on how these
ratios are distributed. In other words, this practice (implicitly or explicitly) em-
phasizes the relative position of a firm's financial ratio in the distribution of the
financial ratio in the relevant industry. The extreme would be to consider a firm
with e.g. a profitability
ratio below the industry average as poor, and vice
versa.O.5
Comparing the financial ratios between firms may, however, be quite mislead-
ing, especial1ywhen the structure of the firms is not homogenous. It is quite cam-
mon to assume that the differences of the firms are smal1er within an industry
than between industries. The comparison of firms within an industry is based on
the silent assumption or belief that the firms within an industry are similar enough
* Kiitämme Marcus Wallenbergin liiketaloudellista tutkimussäätiötä sekä Jenny ja Antti Wihu-
rin rahastoa saamastamme kannustuksesta.
0.5
See e.g. Barron (1986), p. 275.

---

for comparison, Le. that their financial ratios belong to distributions which are
common to the industry branch. This, however, is more or less an untested
assumption.1
The object of this study is to establish whether it is a reasonable practice for
a firm to compare its central financial ratios to industry averages in order to ac-
quire information on the performance and the state of the firm. Firms are offi-
ciaIly c1assifiedinto standard industry branches according to the nature of their pro-
duction processes and old conventions. It is therefore obvious that the feasibility
of these comparisons and the validity of such information is dependent on the
homogeneity of the industry branch as measured by the financial ratios. It is there-
fore the problem of this study to establish whether, and to what extent, the standard
industry c1assification involves branches of industry with dynamicaIly homoge-
neous central financial ratios. The relevancy of standard c1assifications in the face
of central financial ratios is at issue at the same time.1.5
If homogeneity of financial ratios prevails in an industry branch, comparisons
with industry averages or other firms within the industry will give valuable infor-
mation. On the other hand, if heterogeneity prevails, comparisons to the industry
averages are ill-advised because then the standard classification must be assumed
to have produced groupings of firms which differ in their characteristics measured
by the financial ratios. Firm-to-firm comparisons are subject to the same risk of
ill-advised comparisons.
In cases of heterogeneity comparisons to the industry average, as weIl as to
the other firms within the industry, mainly reflect the differences of the struc-
tures of the firms rather than the differences in their financial performance.
Intra-industry comparisons between firms may be better warranted in some
branches of industry than in others. Our results, in fact, indicate that this is the
case for the publicly traded Finnish firms.
The problem of the study can be crystaIlized as foIlows. How homogeneous
is any branch of industry with respect to the covariation of the central financial
ratios of the firms comprising that industry. To find an answer we proceed in
three major stages. First, a dynamic measure of financial ratio closeness between
pairs of firms is developed. Second, the firms under observation are grouped into
homogeneous clusters on the basis of this closeness measure. Third, the congruence
I
See, however, Foster (1978), Section 3.3.
1.5 The research problem tackled in this report can be found in its embryonic form in Gupta &
Huefner (1972).

---

of this clustering with the standard industry classification is measured. As a re-
sult of this research approach we are able to draw conclusions on 1) the degree
of homogeneity of the different industry branches with respect to the central fi-
nancial ratios, 2) the ideal groups of reference for firms' financial ratio com-
parisons, 3) the variability of homogeneity from branch to branch, and 4) the over-
all congruence of the standard industry classification system with financial ratio
homogeneity.
To start with, decisions on the following three questions should be made:
1) Selection of the sample firms
2) Selection of the central financial ratios
3) Selection of the time-span of observation
The firms selected for our research project are forty-two publicly traded Finn-
ish industrial firms with financial statements available for 1974-1984.
This se-
lection has several advantages. First, the selected firms play a major role in Finn-
ish industry. This enhances the generality of the results. Second, the quality of
the financial statement disclosure by these firms is above average in most cases.
Third, the financial statements are readily available in our computerized data-
base. Fourth, the size of the sample stays manageable.
The firms included in our sample can be seen in Appendix B (and Appendix C).
As a pilot study, our method was tried out with data from four graphical in-
dustry and four data-processing firms. The results of the pilot study are not in-
cluded. Such pilot data, separate from the actual sample, was used to enhance
the independence between our method development and our empirical findings.
In general, much research has been carried out on factoring financial ratios
into mutually exclusive categories. The aim of trying to find independent cate-
gories of financial ratios is to develop sets of ratios which would cover the firms'
various activities as effectively as possible. The references in Yli-Olli (1983) and
Yli-Olli and Virtanen (1985) cover the work in this area rather comprehensively.
They are not repeated in this paper.
The central financial ratios for our study will be selected from earlier litera-
ture and experience in financial statement analysis in order to cover profitability,
liquidity, capital structure, and turnover aspects of the firm. Furthermore, a mea-
sure of operating leverage will be included. This kind of measure is needed, since,
a priori, capital and labor intensity varies from industry to industry. Our choice
of the financial ratios will be discussed in greater detail in the next chapter.
In Chapter 3 we develop a measure of financial ratio closeness between busi-
ness firms. Our principal measure is based on covariation of financial ratios
between firms.
In Chapter 4 the firms are categorized into homogenous groups based on our
covariance measure using Ward's minimum variance cluster analysis. For com-
parison, we also perform cluster analysis based on the common approach of using

---

the average levels of financial ratios instead of their covariation. It should be
stressed, however, that this common approach based on averaging financial data
reduces the informational content of the ratios. Our covariation measure avoids
the loss of information caused by averaging. Thus our analysis will be carried out
separately on the basis of the covariation and on the basis of the average levels
of the central financial ratios.
When the firms have been categorized into clusters, a further measure based
on conditional probability is developed in our paper to assess the congruence of
the (official) classifications of the firms into industry branches and the clustering
of the firms along their observed financial ratios.
In this chapter we present the selection of the financial ratios used in our study.
We proceed from the general principles of the selection to the specific details of
calculating the ratios selected.
In principle, financial ratios are calculated by the interested parties of the firm
(debtors, creditors, management, competitors, investors, etc.) The fundamental
aim of using financial statement analysis is making better decisions. More accu-
rately, the purpose of using information (such as financial ratios) by the decision
maker is to gain enough benefit from the better decision to warrant the cost of
acquiring and using the information.2 In financial statement analysis this goal is
naturally furthered by a wellfounded selection of the financial ratios.
In our study, we first select five broad categories of financial ratios. The ob-
jectives and research results on factoring financial ratios3 constitute the premises
of our selection of financial ratios. Thus, our selection is based on the following
four considerations.
1) Maximum coverage of firms' various aspects and activities.
2) Minimum overlapping.
3) Sparsity.
4) Relevance in the actual practice of financial statement analysis.
The categories we select are
1) Operating leverage
2) Profitability
3) Financial leverage
2
Academic research can be considered an exception to the rule, since in academic research the
aim of financial statement analysis is not decision making directly, but the furthering of the stock
of knowledge of economic phenomena.
3
See Yli-Olli (1983), Yli-Olli and Virtanen (1985), and their references for research on factoring
financial ratios.

---

4) Liquidity
5) Turnover ratios
Second, we select a representative financial ratio in each category. Our selec-
tion of the financial ratios is a compromise between »pragmatical empiricism»\
earlier literatures, and theoretical consideration.
In practice, financial ratios are very often used for comparing firms with each
other (cross-sections) and/or assessing the development of the firm' s position over
time (time-series). As is recalled from Introduction, we are interested in establish-
ing to what extent the common practice of intra-industry comparison rests on
a valid basis. Therefore, our selection of the representative financial ratios and
their computation is closely connected to the conventions for calculating finan-
cial ratios as recommended by Business Research Council of Finnish Banks, Yri-
tystutkimusneuvottelukunta
(1983).6
The inclusion ofprofitability,
financialleverage, liquidity, and turnover ratios
is routine in financial statement analysis, and thus requires no further comments.
Operating leverage is included because of the nature of the hypothesis we are test-
ing (Le. whether firms form clusters along industry branches when grouped by
the financial ratios.) Operating leverage reflects the characteristics of the produc-
tion process of the firm (especially its capital vs. labor intensiveness).
The representative financial ratios selected, as wel1 as their empirical defini-
tions are given below. As stated in Intraductian,
and listed in the Appendices,
the ratias were calculated far 42 publicly traded Finnish firms for 1974-1984.
1) Labor intensiveness:
Personnel expenditures /
Adjusted real-term fixed assets
The ratia abave is a measure af aperating leverage. Persannel expenditures
per fixed assets are used as a surrogate of variable/fixed casts. Personnel expendi-
tures are calculated as the salaries, wages plus social expenditures on the income
statement. In standard Finnish financial statement analysis fixed assets and de-
preciatian are usual1y adjusted ta conform with the maximum geametric deprecia-
tian al10wed in Finland. We apply the maximum geometric depreciatian when cal-
culating the return on capital. In calculating labar (ar capital) intensiveness, haw-
ever, the labor cost must be compared with physical productian capacity rather
than the book value af the firm's fixed assets. Therefore, far this case, we apply
a much lawer depreciation percentage, Le. three per cent which has empirical1y
been estimated ta carrespond ta the average wear and tear of physical assets in
4
As defined by Horrigan (1968, p. 288).
5 Such as Aho (1981), Bernstein (1978), Brealey & Myers (1984), Foster (1978), Gibson (1978),
Kettunen & Mäkinen & Neilimo (1976), Lev (1978), Yli-Olli (1983), and so on.
6
The ubiquitous source for practitioners of financial statement analysis in Finland.

---

Finnish industry. Furthermore, eumulative appreeiations of fixed assets made on
books are adjusted for.
2) Return on assets:
Adjusted net profit after taxes + interest expenses /
Adjusted balanee sheet total
Adjusted net profit is net profit before the effeet of ehanges in reserves. Ii
is also adjusted for depreciation (eL the diseussion above), ehanges in pensions
funds' deficit, and taxes transferred to equity. Adjusted balanee sheet total is eal-
eulated as balanee sheet less liabilities not bearing interest less cumulative appre-
ciations of fixed assets less eumulatively the difference from maximum geometrie
depreciation (30 per eent for maehinery and equipment, 9 percent for buildings
and eonstructions).
3) Leverage:
Liabilities bearing interest expenses /
Adjusted balance sheet total
4) Quick asset ratio:
Quick assets / Net sales
Quiek assets are financial assets less short term liabilities.
5) Inventory turnover in days:
360 * Average inventory /
(use of materials and supplies + direet wages)7
Next, a methodology is developed for our purposes. As is reealled, our pur-
pose is to estimate which firms belong to same categories on the basis of the be-
haviour of the selected financial ratios. The clustering method to be used is Ward's
minimum varianee cluster analysis.
Alternatively, diseriminant analysis eould have been considered. We chose
cluster analysis instead because it goes beyond testing the observed groupings. Ii
indicates the underlying ideal groupings.
Cluster analysis is a method to classify objects into groups (clusters).8 The
classification is based on a distanee measure or on a similarity measure for the
objects. Cluster analysis involves the ehoice of a distanee or a similarity measure,
the eriterion to form clusters, and the proeedure to define the number of clusters.
* Our thanks are due to Mr. Hannu Hirvonen for his skilful advice in special data processing
problems.
7
Le. direct costs of goods produced.
8
See e.g. Romesburg (1984).

---

In using financial ratios as the determinants of c1assifying firms into c1usters
three dimensions are involved. These are the forty-two firms, the five financial
ratios, and the eleven years. One approach to the excess dimensionality is to take
averages over the years. Research results using average financial ratios over the
eleven years are reported in Appendix C.
However, using the averages is insufficient. As stated earlier, considerable loss
of information ensues in averaging time-series. Furthermore, covariation between
financial statement analysis ratios within firms is ignored in averaging. Therefore,
we develop an alternative measurement technique, which covers the variation of
the financial ratios. By its nature, our covariation measure will be independent
of the levels of the ratios and thus ignore the levels.
For the rest of this chapter, consider measuring the covariation between
financial statement analysis ratios within and between firms.
First, consider just two firms and a single financial ratio (denoted by X be-
low).
Several alternatives are possible in trying to establish whether the time-series
of the financial ratios X(t, h) and X(t, k) come from equivalent stochastic pro-
cesses. (Recall, that if this is the case, the comparison of the two firms is then
well-founded for the feature which is measured by financial ratio X. For example,
if financial ratio X is a measure of profitability, such as return on total assets,
profitability comparison for the two firms is tenable.) For the measure of c10seness
between X(t, h) and X(t, k) we se1ect the correlation between the observed time-
series of the ratio.
Next, consider a general number 1 of financial ratios for just two firms h and
k. The correlations r(hk)(ij) between the ratios of the two firms can be presented
as a matrix. Indices i and j are used for the financial ratios.
(firm k)
j
1
r(hk)(ll)
r(hk)(lj)
r(hk)(lI)
(firm h)
r(hk)(il)
r(hk)(ij)
1
r(hk)(Il)
r(hk)(Ij)

---