Timo Salmi Professor of Accounting and Business Finance A Review of the Theoretical and Empirical Basis of Financial Ratio
Analysis Financial ratios are widely used for modelling purposes both by practitioners and researchers. The firm involves many interested parties, like the owners, management, personnel, customers, suppliers, competitors, regulatory agencies, and academics, each having their views in applying financial statement analysis in their evaluations. Practitioners use financial ratios, for instance, to forecast the future success of companies, while the researchers' main interest has been to develop models exploiting these ratios. Many distinct areas of research involving financial ratios can be discerned. Historically one can observe several major themes in the financial analysis literature. There is overlapping in the observable themes, and they do not necessarily coincide with what theoretically might be the best founded areas, ex post. The existing themes include
• • • • • • • • • •
the functional form of the financial ratios, i.e. the proportionality discussion, distributional characteristics of financial ratios, classification of financial ratios, comparability of ratios across industries, and industry effects, time-series properties of individual financial ratios, bankruptcy prediction models, explaining (other) firm characteristics with financial ratios, stock markets and financial ratios, forecasting ability of financial analysts vs financial models, estimation of internal rate of return from financial statements.
The history of financial statement analysis dates far back to the end of the previous century (see Horrigan, 1968). However, the modern, quantitative analysis has developed into its various segments during the last two decades with the advent of the electronic data processing techniques. The empiricist emphasis in the research has given rise to several, often only loosely related research trends in quantitative financial statement analysis. Theoretical approaches have also been developed, but not always in close interaction with the empirical research.
A financial ratio is of the form X/Y. pp. In this paper we review the existing trends in financial statement analysis literature by focusing primarily on the theoretical and empirical basis of financial ratio analysis. In traditional financial ratio analysis both the X and the Y are based on financial statements. One way of categorizing the ratios is on the basis where X and Y come from (see Foster. In our opinion the primary areas of the literature concerning the theoretical and empirical basis of financial ratio analysis are the functional form of the financial ratios. financial ratios can be divided into several. By critically considering financial ratio literature. If both or one of them comes from the income statement the ratio can be called dynamic while if both come from the balance sheet it can be called static (see ibid. sometimes overlapping categories. This area.Technically. 1990. 36-37. This is an important task to carry out since the ratios are often used intuitively. We review four of the research areas listed above. and classification of financial ratios. 10-11). Virtanen and Yli-Olli. The concept of financial ratios can be extended by using other than financial statement information as X or Y in the X/Y ratio. All the major financial ratio research avenues cannot be tackled within the limited space of this paper. where X and Y are figures derived from the financial statements or other sources of financial information. distributional characteristics of financial ratios. financial statement items and market based figures can be combined to constitute the ratio. pp. we also aim to help the decision makers to use ratios in an efficient way. reviewed in Section 3. Therefore. we select the estimation internal rate of return from financial statements as the fourth area.
. 1978. and Salmi. without sufficient consideration to their theoretical meaning and statistical properties. In doing this it is our purpose to pinpoint the different directions taken in quantitative ratio based research. For example.). A fundamental task of financial analysis is evaluating the performance of the business firm. These three research avenues are reviewed in Section 2. directly concerns profitability measurement.
linearity. While mostly tackling the former Whittington (1980) independently presents illustrative results finding the ratio specification inappropriate in a sample of U. In the discussion on Barnes's paper (Horrigan. Whittington also discusses the usage of a quadratic form in FRA. Two interrelated trends are evident. Significant instability in the results was reported. The usually stated requirement in controlling for size is that the numerator and the denominator of a financial ratio are proportional. 1983. They point out. existence of the intercept. Furthermore. Barnes. Theoretical discussions about the ratio format in FRA and empirical testing of the ratio model. 1983). Lev and Sunder bring up the problems caused in multiple regression models where the explaining variables are ratios with the same denominator. using theoretical deduction. Horrigan puts forward that financial ratio research should be more interested in the role of the financial
. He is thus able to tie in the ratio format aspects with the distributional properties of financial ratios (to be discussed later in this review). The proportionality considerations have implications on various facets of FRA. that in order to control for the size effect. It is shown that the choice of the size deflator (the ratio denominator) is a critical issue.K. and dependence on other variables in the basic financial variables relationship models Y = bX + e and its ratio format Y/X = b + e/X). This is a fact that has been discussed earlier in statistics oriented literature like in Kuh and Meyer (1955). the financial ratios must fulfill very restrictive proportionality assumptions (about the error term. in this paper). firms. Barnes (1982) shows how the non-normality of financial ratios can result from the underlying relationships of the constituents of the financial ratios. This basic assertion gives rise to one of the fundamental trends in financial ratio analysis (or FRA for short. The seminal paper is this field is Lev and Sunder (1979).Teppo Associate Professor of Accounting and Business Finance Basic Properties of Ratios Functional Form of Financial Ratios
The traditionally stated major purpose of using financial data in the ratio form is making the results comparable across firms and over time by controlling for size.
a + with bX(i) + intercept e(i). To extrapolate from Horrigan's critique. management. 1991).. The third model applies a (Box-Cox) transformation on the first model to tackle non-linearities. 1985) present the first extensive empirical studies of the statistical validity of the financial ratio method. in inter-industry comparisons proportionality of financial ratios is not supported. The first model is the traditional model for replacement of financial ratios by bivariate Y(i) = regression. On the other hand.
.. To illustrate. consider the potential impact of economics of scale. in our own interpretation the validity of financial ratio analysis should be determined by its usefulness to the decision making process of the different interested parties (owners. To assess the efficiency of management a direct comparison of financial ratios of small and big firms would have to be adjusted for the size effect. personnel. For a discussion also see Garcia-Ayuso Y(i) (1994). It is characteristic that the testing for proportionality is considered in terms of testing the hypothesis H0: a = 0.ratios themselves than in "the nature of the ratios' components or to the ratios' incidental role as data size deflators". McDonald and Morris (1984.). one with a single industry the other with one randomly selected firm from each (four-digit SIC) industry branch to investigate the implications of homogeneity on proportionality. = The second model b'X(i) in McDonald + and Morris is e'(i)
that is without the intercept to tackle heteroscedasticity. an investor evaluating different investment targets might be more interested in the level of profitability regardless whether or not it is a result of the size effect. While they find support for financial ratio analysis for comparisons within industry branches. The authors use three models with two samples. Barnes (1986) points out for statistical testing that the residual is typically heteroscedastic. Dropping the intercept from the model is not always enough to treat the heteroscedasticity (see Berry and Nix.
The above model is central in this area..
It is interesting to recognize that all ingredients of modern distribution analysis already appear incumbent in Mecimore's paper. This is because the significance tests in parametric methods prevalent in FRA research. Concomitant results are obtained by Lee (1985) using a stronger test (Kolmogorov-Smirnov) for a different set of data. In fact some of the papers reviewed tackle both the areas either separately or within the same framework.Distributional Characteristics of Financial Ratios It is typical of FRA research that there are several distinct lines with research traditions of their own. An early paper on financial ratio distributions was published in Management Accounting by Mecimore (1968). while not statistically significantly. Bird and McHugh (1977) adopt an efficient Shapiro-Wilk small-sample test for the normality of financial ratios for an Australian sample of five ratios over six years. such as regression analysis and discriminant analysis. but part of this research is intertwined with the proportionality research discussed above. the much later published article by Deakin (1976). The distributional characteristics of financial ratios have induced a research line of their own. In some cases there is little link to the other FRA fields. His chi-square findings reject (with one exception) the normality of eleven financial ratios in a sample of 1114 Compustat companies for 1954-72. Using descriptive statistical measures (average and relative deviations from the median) he observes cross-sectional non-normality and positive skewness for twenty ratios in a sample of randomly selected forty-four Fortune-500 firms. rely on the normality assumption. Likewise. but normality was still not supported. In the history of FRA it is common that professional journals and academic papers do not recognize each other. The paper most often referred to in literature as the seminal paper in this field is. Less extreme deviations from normality were observed when square-root and logarithmic transformations were applied. The recurring motivation for looking into the distributional properties of financial ratios is that the normal distribution of the financial ratios is often assumed in FRA. They also study the adjustment of the financial ratios towards industry means which is a
. industry grouping made the distributions less non-normal. however. Like Deakin they find in their independent study that normality is transient across financial ratios and time.
While improving the statistical results trimming and transformations can pose a problem for the theoretical rigor in FRA research. Also his empirical results confirm non-normality. The list includes deleting true outliers. however. Ezzamel. the square root transformation makes the distribution approximately normal. He seeks for a best fitting t-distribution for a cross-section of 1634 UK and Irish firms. McLeay (1986b)
. There is. adjusting the underlying financial data. Foster (1978) points out the outlier problem in FRA.different area of FRA research. winsorizing that is equating the outliers to less extreme values. Foster also puts forward accounting. Kolari. One of the avenues taken is to study new industries. The results indicating non-normality of financial ratio distributions have led researchers into looking for methods of restoring normality to warrant standard parametric statistical analyses. a certain degree of circularity in their approach. Mar-Molinero and Beecher (1987) and Ezzamel and Mar-Molinero (1990) review and replicate the earlier analyses on UK firms with a particular emphasis on small samples and outliers. The empirical results are supported by later papers such as So (1987). The best-fitting (in the maximum-likelihood sense) t-distribution varies across financial ratios (the t-distribution can be considered a family of distributions defined by its degrees of freedom). A varying and often a considerable number of outliers has to be removed for different financial ratios in order to achieve normality. They point out that if the ratios follow the gamma distribution. respectively. economic and technical reasons for the emergence of outliers in FRA. McInish and Saniga (1989) take on the distribution of financial ratios in banking. since instead of identifying the underlying causes of the outliers they employ a mechanistic statistical approach to identify and remove the outliers from the tails of the financial ratio distributions. and trimming by dropping the tails. he presented in Foster (1986) a list of alternatives for handling outliers in FRA. Frecka and Hopwood (1983) observe that removing outliers and applying transformations in a large Compustat sample covering 1950-79 restored normality in the same financial ratios as tackled by Deakin (1976). retaining the outlier. Bougen and Drury (1980) also suggest non-normality based on a cross-section of 700 UK firms. The gamma distribution is compatible with ratios having a technical lower limit of zero. Buckmaster and Saniga (1990) report on the shape of the distributions for 41 financial ratios in a Compustat sample of more than a quarter million observations. Instead of deleting or adjusting the observations McLeay (1986a) proposes using a better fitting distribution with fat tails for making statistical inferences in FRA. Later.
but the general inference is that "normality will be the exception rather than the rule". immediately evident. The interpretation in terms of implications to financial ratio distributions are not. Martikainen (1992) uses a time-series approach to 35 Finnish firms in turn observing that controlling for the economy factor improves normality. The question of the distribution of a ratio format variable (financial ratio) has been tackled mathematically as well as empirically. Virtanen and Yli-Olli (1989) studying the temporal behavior of financial ratio distributions observe in Finnish financial data that the business cycles affect the cross-sectional financial ratio distributions. however. many later papers tackle the same basic question of ratio distributions using different samples and expanding on the methodologies. In a sample of 35 Finnish firms. Tippett (1990) models financial ratios in terms of stochastic processes. Buijink and Jegers (1986) study the financial ratio distributions from year to year from 1977 to 1981 for 11 ratios in Belgian firms corroborating the results of the earlier papers in the field. Also the results by Martikainen (1991) demonstrate that normality can be approached by other procedures than removing outliers.
.also tackles the choice between equally weighted and value weighted aggregated financial ratios in terms of ratio distributions on a sample of French firms. Barnes (1982) shows why the ratio of two normally distributed financial variables does not follow the normal distribution (being actually skewed) when ratio proportionality does not hold. They also point to the need of studying the temporal persistence of cross-sectional financial ratio distributions and suggest a symmetry index for measuring it. Refined industry classification brings less extreme deviation from normality. Typically. four ratios and fifteen years about half of the non-normal distributions became normal if economy-wide effects were first controlled for using the so-called accounting-index model.
Eubank. They find that the a priori categories are correlated with each other. The other three dimensions turn out more commonly to be interrelated.
. They use the redundancy indexes produced by canonical correlation analysis to evaluate how well financial ratios fit the relevant factor. deduction and visual approximation of data. Rather than using cross-sections across firms their data consist of time series of 40 Finnish firms for 1974-84. As noted earlier Courtis (1978) presents a pyramid scheme of financial ratios based on a mix of experience. and compare the empirical clusters with the a priori dimensions. He compares his results with the deductive classification by Courtis (1978) and finds a good correspondence. financial leverage. profitability. at least beyond three to five factors. Consequently a number of later studies hypothesize an a priori classification and then try to confirm the classification with empirical evidence. liquidity. With the exception of administration Laurent identifies and locates each of his ten empirical factors in Courtis's framework. They apply cluster analysis to group the 15 initial ratios separately for each firm in the sample. Profitability and revenue liquidity appear almost invariably as distinct clusters.Confirmatory Approach It seems that despite the initial optimism the inductive studies have been unable to agree on a consistent classification of financial ratio factors. A tentative emergence of this idea can be detected in Laurent (1979). Such a comparison has the hallmarks of the basic idea of the confirmatory approach. Laurent performs a standard principal component factorization for a set of 45 financial ratios presumably for a single year of 63 Hong Kong companies. Pohlman and Hollinger (1981) test two a priori classification schemes based an a sample of Compustat firms for 1969-78. Thus they caution against using too few financial ratios in FRA. Mingo and Caruthers (1975) with seven factors. and revenue liquidity. This can be considered an a priori classification. working capital. (It practically is Lev's (1974) categorization. Luoma and Ruuhela (1991) present five a priori "dimensions" for the financial ratios. They call the first the "traditional" scheme.) The second is not actually a priori classification but the empirical classification by Pinches.
by considering the relationship between the familiar accounting rate of return (the firm's annual profit in relation to its assets) and the internal rate of return. although Salamon (1973) casts doubt of this view. (The average of K(t) and K(t+1) is also often used. In terms of economic theory the profitability of a firm could be defined as the internal rate of return of the capital investments constituting the firm. D(t) is the depreciation in period t.
where F(t) is the funds flows from operations in period t. Using four different cases of accumulation of assets t
I(o) = sum R(t)/(1+r). British economists present one tradition of tackling the question of the divergence between the ARR and IRR since Harcourt (1965) put forward his position that the accountant's rate of return is "extremely misleading". = formal definitions. and n is the life-span of the capital investment. and K(t) is the net book value of assets at the beginning of year t. The ARR vs IRR discussion can also be deemed as seeking a reconciliation between accounting based measurement and the economic theory of income. There is a significant body of literature which considers profitability assessment. (The existence conditions for a rational solution for r. (F(t) ARR is defined in literature as D(t))/K(t). To recount the general.
.) IRR is naturally defined as r by n t=1 where I(o) is the initial capital investment outlay. The relationship has been considered both as a purely mathematical relationship and from the empirical estimation point of view. They will be called IRR and ARR below since there is some variance in the full terms in literature. There is a strong tradition in literature that seeks to estimate the internal rate of return. The measurement of profitability is intimately linked with both. either from a time series of the financial statements of the firm.Measurement of Profitability and Financial Ratios ARR vs IRR The fundamental task of accounting is income determination and the evaluation of the firm's assets. more narrowly. or. and the multiple solutions of the polynomial equation have been tackled in the relevant literature but are not reviewed in this paper). especially for the latter. R(t) the net cash flow in period t.
The relationship between the ARR and IRR is also indirectly involved in studies considering the relation between rules of thumb for capital investment decisions (payback reciprocal) and the ARR on the other hand and IRR on the other. He concludes by an explicit warning about profitability comparison between firms in different industries or different countries if accountants' measurements are used. Vatter (1966) ponders the content of Solomon's paper at great length. His paper shows that the difference between the two measures involves project lives. He questions both the realism of Solomon's assumptions and the validity of IRR as a practical measure of profitability. 483). Interestingly these two papers are practically devoid of references to other literature. the depreciation method. Stauffer (1971) presents a generalized analysis of the ARR vs IRR relationship using continuous mathematics under several cash profile assumptions. Further numerical examples to illustrate the disparity of accounting and economic profitability measurement are provided in Solomon and Laya (1967). and different depreciation methods. Livingstone and Salamon (1970) build on Solomon's model and conduct a simulation analysis of the ARR-IRR relationship by extending the assumptions of the previous models into more general cases. pattern of cash flows generated by the projects making up the firm. See Sarnat and Levy (1969. different growth rates. They also include the effect of growth. They observed under their assumptions that ARR shows a dampening cyclical behavior determined by the project life-spans. McHugh (1976) and Livingstone and Van Breda (1976) have an exchange of views about the mathematical derivations and the generality of the results of Livingstone and Salamon (1970).(growth) he asserts that it is not possible to develop rough rules of thumb to adjust ARR to reflect IRR under different life-spans of investments. The formal mathematical relationship between the ARR and IRR is independently considered by Solomon (1966). the net cash flow patterns generated by the investments. He demonstrates that the depreciation schedule affects the relationship. It can only be deduced that he implicitly gives very little value for the financial statements annually prepared by the accounting profession. p. and the level or IRR. Also he puts forward that the accounting and
. and the lag between the investment outlays and their recoupment. Using both a zero-growth and a growth model he demonstrated that the ARR (book-yield in Solomon's terms) is not a reliable measure of the IRR (true-yield in Solomon's terms). the reinvestment rate.
To us this view appears logical because it is unlikely that profit oriented business firms could. Long and Ravenscraft (1984) present a critical view on Fisher and McGowan's claim of the prevalence of the IRR. Fisher (1984) discards the criticism insisting that ARR does not relate profits with the investments that produce it. Stephen (1976). claims that Gordon fails to resolve the difference between ARR and IRR. He also points out that the using the annuity method of depreciation makes ARR a more accurate reflection of the IRR. The crucial requirement is that the accountant's evaluation of the assets (their book value) and the economist's evaluation (the discounted net cash flow) are equal. Also Bhaskar (1972) arrives at the conclusion that "in general ARR does not perform satisfactorily as a surrogate for the IRR". the assumptions in their examples. Gordon (1974. He shows that ARR can be a meaningful approximation of the IRR when "the accountant's income and asset valuations approximate the economic income and asset values". but points out that the annuity method has undesirable side-effects for accounting measurement. Bhaskar augments his deductions with a statistical analysis of his simulation results on ARR and IRR levels. and that the situation is aggravated by the introduction of taxation into the analysis. the managers can be able to make sufficient adjustments. He also applies his results to estimate the profitability of the British manufacturing industry 1960-1969 from aggregate accountant's
. indulge in totally unsound measurement and management practices. They conclude that the accounting rate of return is a misleading measure of the economic rate of return and see little merit in using the former. and their mathematical derivations. in the long run. The central condition is linked to the depreciation method. for example. on the other hand. Gordon concludes by pointing out that even if no general "cookbook tricks" can be devised for converting the ARR to the IRR. From the accounting point of view it is interesting that he points to the task of estimating the real rates of return from historical accounting data. Fisher and McGowan (1983) consider economic rate of return (IRR) the only correct measure of economic analysis. The accountant's accumulated depreciation must approximate the accumulated economic depreciation for the ARR and IRR to converge.the economic measurements (ARR/IRR) are irreconcilable. Kay (1976) refutes Salamon's conclusion and contends that IRR can be approximated by the ARR irrespective of the cash flow and depreciation patterns. 1977) takes a more optimistic view on the potential reconciliation between ARR and IRR. Likewise.
Stark (1982) recounts Key's results by including working capital. He also studies the relationship when IRR is not equal to the growth rate of assets. Peasnell (1982b) goes on to consider the usefulness of ARR as a proxy of IRR for FRA. Peasnell (1982a) also considers economic asset valuation and yield vs accounting profit and return in a discontinuous (discrete time) mathematical derivation framework (while Kay. He presents an iterative weighting scheme for estimating the IRR from the ARR.data. and Salmi. Tamminen (1976) presents a thorough mathematical analysis (with continuous time) of ARR and IRR profitability measurement under different contribution distribution. loan financing and taxation. The IRR is a long-term. He points to the results stating that the annual ARR is a surrogate of the IRR only
. Wright (1978) considers Kay's (1976) view too optimistic and claims that one cannot easily translate ARR into IRR except under special circumstances. He also stresses the fact that the measures are conceptually different by nature. As one result he derives a growth-dependent formula for a conversion between IRR and ARR assuming realization depreciation. Salmi and Luoma (1981) demonstrate using simulated financial statements that applying Kay's results require more restrictive assumptions than originally indicated by Kay (1976). The analysis is conducted under steady-state growth conditions.g. and ARR is a constant. Bierman. already inherently is a valid and useful variable especially in the positive research approach. Luckett (1984) reviews and summarizes the ARR vs IRR discussion. Whittington (1979) points out that the ARR vs IRR discussion should also consider whether ARR. growth conditions. and depreciation methods. He proves that if there are no opening and closing valuation errors of assets with respect to their economic values. His main conclusions are pessimistic. then the constant ARR equals the IRR. average-type ex ante measure. 1961. while the ARR is a periodic ex post measure. Key and Mayer (1986) revisit the subject coming to the conclusion that "accounting data can be used to compute exactly the single project economic rate of return". then extended to structural changes and for under cyclical fluctuations. instead of IRR. 1978). He also studied the possibility of extenuating circumstances that could reduce the ARR vs IRR discrepancy in statistical analysis. Applying a standard variation measure he comes to the conclusion that the usage of ARR does not lead to serious valuation errors in FRA provided that the variations in the ARRs are not too great. 1976 used continuous time). Under constant growth equal to IRR he proves that IRR can be derived as the mean of ARRs. (For the definition of the realization depreciation see e.
Kelly and Tippett (1991) present a stochastic approach to estimating the IRR and ARR. He concludes that while ARR has serious limitations as a measure of business performance. Feng. He also claims that it estimating the IRR in actual practice from the annual ARRs is not generally practical.under very special circumstances. Dressler. He evaluates the validity of ARR as a proxy for IRR by examining the association between corporate level ARR and the stock return for 241 Compustat firms for 1963-82.S. companies for 1955-84. Shinnar. and find them significantly different in a sample of five Australian firms. ARR and the cash flow pattern for 38 U. claiming that ARR has no relevance is an overstatement. Jacobson (1987) takes another approach to the IRR vs ARR controversy.
. he does not take on examining the association between IRR and stock returns. and Avidan (1989) estimate the IRR. However. possibly because of the difficulty of estimating the IRR from the published data.