Different approaches have been applied on the classification problem of the financial ratios. The first approach could be called a pragmatic or an authoritative approach. In this approach the classifications of financial ratios have largely developed from established business practices and personal views of eminent financial analysts. Many standard text-books present material from this approach. (See e.g. Aho 1981, Bernstein 1989, Brealey & Myers 1984: Ch. 25, Foster 1986, Fridson & Marocco 1986, Kettunen & Mäkinen & Neilimo 1976 and Lev 1974.
The second approach has been more deductive. In this approach the classification of the financial ratios has been based on the technical relationships between the different financial ratios. The "Du Pont triangle" from the beginning of the century is a classic in this respect. (See Horrigan 1968.) The modern papers using this "pyramid" approach include Courtis (1978), Laitinen (1983), and Bayldon & Woods & Zafiris (1984).
The third approach has been inductive, empirical classification of financial ratios using statistical techniques, factor analysis in particular. In this approach, factor analysis is used to reduce a (large) number of financial ratios into a smaller number of mutually exclusive categories covering the various aspect of the firm's activities. (See e.g. Salmi & Dahlstedt & Luoma & Laakkonen 1986 for a brief summary of the objectives of factor analysis in ratio analysis.) Methodologically, this means reducing a large number of measured variables into a smaller number of latent variables, and then giving interpretative names to these latent variables.
Since the paper by Pinches & Mingo & Caruthers (1973) conventional financial ratios have been factorized in many research papers including Pinches & Eubank & Mingo & Caruthers (1975), Laurent (1979), Johnson (1979), Aho (1980), Chen & Shimerda (1981), Cowen & Hoffer (1982), Yli-Olli & Virtanen (1985), and Ezzamel & Brodie & Mar-Molinero (1987), and using a confirmatory test method Kanto & Martikainen (1989).
Using an inductive approach for classifying financial ratios raises the question of the stability of the results between the different studies, and even between different years within the same study. This fact has been pointed out, and tested for, in several of these studies, Pinches & Mingo & Caruthers (1973) included. They concluded a reasonable stability of their results. On the other hand, the results of Polhman & Hollinger (1981) can be interpreted as a caution against drawing generalized conclusions from the Pinches & Mingo & Caruthers (1973) classification. Yli-Olli & Virtanen (1985, 1989, 1990) introduced using transformation analysis to test for the stability between different years, and data-sets from different countries. It is evident from these studies that the stability of the results is an important, and even a critical issue.
Another issue is the coverage of the selected financial ratios. The financial ratios have usually been selected from the traditional, accrual ratios. It has been put forward, e.g. by Artto (1978), that cash flows contain such information about the activities of the firm which is not present in the accrual-based financial statements. Gombola & Ketz (1983), and Yli-Olli (1983) observed that cash flow ratios produce an independent and persistent factor.
Three facts, which will be relevant for our research problem, are evident in the earlier research on classifying financial ratios using factor analysis. First, earlier research has been largely inductive. A hypothesis approach has been conspicuously scanty. Second, market data has not featured in these studies. Third, the methods of selection of the original financial ratios to be factored has strong ad-hoc features.
First, the need for rules and legislation for security trading conduct has been very much in evidence in Finnish public discussions. Second, the volume of trading grew explosively towards the end of the 1980's. Third, and most importantly for our current focus, the amount publicized of security trading and financial statement information has been steadily increasing. This has been paralleled by a growing number of academic Finnish (mostly empirical) research projects in modern finance theory.
As was discussed in the previous section, the formulas and classifications of analyzing the firms' financial statement numbers have become more or less established in financial statement analysis practices, and much research has been done on categorizing financial ratios.
Likewise, the efficiency of the security markets, and the determinants of security prices have been much researched (especially in the framework of the capital asset pricing model CAPM, and arbitrage pricing theory APT). (For CAPM see Sharpe 1963, Sharpe 1964, Lintner 1965, Mossin 1966, Black 1972, Black & Jensen & Scholes 1972, and Foster 1978. For APT see Ross 1976, Roll 1977, Dhrymes & Friend & Gulletin 1984, and Elton & Gruber 1987.) Despite this fact, both the practice, and theory of classifying the numerical security characteristics information is at its infancy.
With the increasing number of numerical indicators on securities (such as price/earnings, efficient yield, betas, etc.) it is important both from the viewpoint of practice and research to establish what the factual informational content of these numerical security characteristics is. What information is overlapping, are there distinct classes of numerical security characteristics, and what are their relationships to firms' financial statement numbers.
The second group of questions concerns the accrual and cash flow ratios. Does the introduction of the market based variables influence the familiar factor patterns? In other words are the earlier results corroborated, or have they been influenced by the limited selection of variables? In addition to the implications on the traditional accrual ratios, we are interested in whether the results on the distinct nature of the cash flow ratios are corroborated.
In tackling our research problem, we shall use a hypothesis approach rather than just observing and reporting the emerging classifications. The statistical methods will be factor analysis, and transformation analysis.
In tackling our research questions special attention must be given to stability, and avoidance of definitional correlation. One of the pitfalls of inductive methods, such as factor analysis, is whether the results are a consequence of a coincidence, and thus unstable, or do they result from true underlying factors, which would mean better stability. Hence we shall test the stability of our factor analysis results with transformation analysis.
Definitional correlation between financial ratios can easily arise if they include, either directly or indirectly, the same components. (E.g. net profit/total assets and net profit/sales are related by definition.) We strive to avoid this pitfall by a judicious selection of the original variables.
Technically, financial ratios can be divided into several, sometimes overlapping categories. A financial ratio is of the form X/Y, where X and Y are figures derived from the financial statements. One way of categorizing the ratios is on the basis of which statements X and Y come from. The most important sources are the income statement, the balance sheet, the funds flow and/or the cash flow statements.
Consider a few examples. The current ratio (current assets / current liabilities) which reflects liquidity, is a typical balance sheet / balance sheet ratio, and thus often considered static (i.e. a stock variable). Profit margin (net income / sales), reflecting profitability, is a typical income statement / income statement ratio, and thus often considered dynamic (i.e. a flow variable). Return on investments (ROI, i.e. net income / total assets) which reflects profitability, is a typical mixed income statement / balance sheet ratio. The examples could be easily extended into ratios involving funds or cash flows.
The information used in financial statement analysis is not limited to ratios nor, for that matter, numerical information. (In security market analysis, especially, it seems that the non-numerical information has a definite role. Nevertheless, we shall only consider quantifiable, numerical information from here on.) In financial statement analysis some non-ratio indicators are frequently used, and important. Examples of such information are total sales, number of employees, market shares, and so on. (For discussions and results on the ratio format and proportionality in financial statement analysis see Whittington 1980, Barnes 1982, Horrigan 1983, Barnes 1983, Barnes 1987, Fieldsend & Longford & McLeay 1987, and Perttunen & Martikainen 1989.)
An extension of the set of financial ratios is to use market based information in the numerator or the denominator. A typical example of such a ratio is the P/E (share price/earnings) ratio, which is much used in security analysis practice.
Taking the idea one step further we come to the case where both the numerator and the denominator are market based. Furthermore, there are important market based indicators which are not calculated as ratios. A security's beta (the systematic risk) is a prime example.
From now on we shall use the term market-based ratio both for the ratios with a market based component, and for the other market based indicators. The term financial ratio will be used for ratios and other similar data derived from financial statements, both with and without the market based element.
In the theory of finance the most common equilibrium model is the Capital Asset Pricing model. One of the basic assumptions of CAPM is that the investors are price takers, and have homogenous expectations about asset returns. The asset returns have a joint normal distribution in the CAPM assumptions. The implication of the normality assumption is that only two parameters, the mean and the variance, are needed to completely describe the distribution. It follows that our first hypothesis will be:
In defining the firm's performance in terms of profitability in the theory of the firm, two different valuations can be used. These are the economist's valuation and the accountant's valuation. The former is based on cash flows while the latter is based on accrual concepts. It can be shown theoretically that the two will be compatible only in the very special case of applying the annuity method of depreciation in the accountant's evaluation of profits. (See e.g. Salmi & Luoma 1981 for a proof, and a discussion.)
Furthermore empirical evidence from factor analysis studies corroborate that cash flow ratios involve information that is uniquely distinct from accrual ratios. (See Gombola & Ketz 1983, and Yli-Olli 1983.) Also the time-series behavior of cash-based accounting figures and accrual-based accounting figures show a fundamentally differing behavior. (See Kinnunen 1988.) These considerations lead to our second and third hypotheses.
itv@uwasa.fi pyo@uwasa.fi [ts@uwasa.fi] [Photo] [Programs] [Research] [Lectures] [Department] [Faculty] [University]