Several considerations prompted spanning the years from 1974 to 1984 in particular. The period is long enough to cover a complete business cycle. The oil-crisis had already culminated. The new Finnish accounting act, which influenced financial statement disclosure in a fundamental way, was passed in 1973. The level of security trading has exploded in the Finnish markets since 1984, and a significant number of mergers and acquisitions disrupting the comparability of the cross-sections have taken place since.
In a study applying data-oriented statistical methods such as factor analysis for categorizing ratios, the quality of the results is (along with other considerations) heavily dependent on a balanced choice of the ratios for the data base. Quite often the selections have, more or less, been based on an ad-hoc, stratified collection of a number of frequently used ratios. We shall use a more systematic procedure. The following interrelated criteria form the basis for our selection:
We introduce the three main categories of ratios, that is the accrual ratios, cash flow ratios, and market-based ratios.
As discussed in the previous chapter, we differentiate between the accrual ratios and the cash flow ratios because different principles of economic theory and accounting are involved. And, as stated, the behavior differences of the cash based accounting numbers from the accrual based has been the focus of interest in several studies.
The market-based ratios constitute the third ex-ante main category. This is natural, since containing market information they are distinct from the conventional financial ratios by definition.
Quick ratio is a prevalent indicator of liquidity with a long history of usage. As is well-known, it is defined as the ratio of the current assets less inventories to current liabilities.
Studies categorizing financial ratios often include the current ratio (current assets to current liabilities) in the data basis along with the quick ratio. We will, however, deliberately exclude the current ratio.
As stated in discussing our approach in the Introduction, definitional correlation between the preselected ratios should be avoided. The quick ratio and the current ratio are prime examples of this feature. Their denominators are the same, and their nominators are quite close definitionally. They only differ by the inventories.
It must be granted, of course, that avoiding definitional links altogether is impossible when selecting a proper set for the database for empirically categorizing financial ratios. This arises from the fact that the ratios must effect a wide coverage of the firms' characteristics and the securities' features.
This principle excludes the simultaneous inclusion of the quick ratio and the current ratio, but it does not indicate which of the two ratios is more suitable for our purposes. Our choice of the quick ratio over the current ratio is based on two considerations. First, inventories (which the current ratio includes) are not always truly liquid in nature. Second, many empirical results indicate that the quick ratio tends to be better behaved statistically than the the current ratio. This is revealed both in classification studies (see e.g. Yli-Olli & Virtanen 1985: 52), and in studies dealing with the distributional properties of these ratios (Virtanen & Yli-Olli 1989: 12-13).
The second included ex-ante liquidity ratio is the defensive interval measure. (See Davidson & Sorter & Kalle 1964.) This ratio is defined as the current assets less inventories per average daily expenditures to operations.
While the quick ratio is a balance sheet / balance sheet ratio, the defensive interval measure is a balance sheet / income statement ratio. We feel that, whenever possible and relevant, the ratios should have such differing bases to increase coverage, and to alleviate the definitional dependencies.
Defensive interval measure can be used to illustrate the fact that the ex-ante classifications of financial ratios are more or less arbitrary. Depending on the way it is looked at, the defensive interval measure could as well be deemed ex-ante a profitability measure. The common way of looking at this measure is that it indicates how well the liquid assets cover the expenditures needed to keep the operations running. But looking at it the other way round, profitability is dependent on the expenditures incurred. Comparing the operating expenditures to the capital base (in this the liquid assets) that is needed to create the activity, gives an inverse type of a profitability measure.
We are not claiming here an attempt to logically deduce which category of financial ratios the defensive interval measure best belongs to. On the contrary we put forward that this kind of duality is an important reason for making the empirical classification studies of financial ratios worthwhile endeavors.
The last of the financial ratios in the liquidity ex-ante category is the relative net working assets. If the liquidity ex-ante category were under observation by itself, we would have opted for net working assets to sales instead of net working assets to total assets. As it is, the selections in the other ex-ante categories have to be taken into account. In this case the total liabilities to sales in the capital adequacy ex-ante category is definitionally too close to net working assets per sales.
As to capital adequacy, we have deliberately used this title instead of the more common financial leverage, since the former is less restrictive.
The financial ratios in the capital adequacy ex-ante category are total liabilities to sales, long term debt to equity, and times interest earned.
Denote balance sheet items by b for short, and income statement items by i. The first of the ratios is a b/i concept, the second b/b, and the third an i/i concept. This variety is intended to give an improved coverage.
Profitability is best regarded as earnings generated in relation to the resources invested in a firm's activities. There are two major ways of looking at profitability. The shareholders are per definition interested mainly in the return on their investment. On the other hand, taking a more managerial oriented view, the focus of interest becomes the productivity of the firm's capital resources. These views are well reflected in including as profitability ratios the return after interest and taxes on equity, and the return on total assets. The empirical classification of these financial ratios in the profitability ex-ante category is naturally of particular interest in our study, where the relation to security market based data is at issue.
The third financial ratio in our profitability ex-ante category is the operating margin to sales. It is selected as an i/i ratio to complement the two b/b ratios.
In calculating the turnover ratios the practice varies between giving the turnover as a period or as a rate (these are inverses of each other). In business practice turnover periods may be easier grasp than turnover rates. Turnover periods also have better statistical properties, especially, a smaller variance and other higher moments, than do the turnover rates. (Virtanen & Yli-Olli 1989). Using these two criteria, we have decided between the two alternatives.
We have here called this ex-ante category efficiency rather than turnover ratios, since the former is more generic. Furthermore, the latter terminology refers to a method of calculation rather than to a concept.
We include two financial ratios in this ex-ante category. (See Appendix A.) These are the labor intensiveness (see Salmi & Dahlstedt & Luoma & Laakkonen 1986: 337-338), and the variable costs to fixed costs. The former is an i/b concept, while the latter is an i/i concept.
Labor intensiveness is defined as the personnel expenditures divided by the adjusted real-term fixed assets. Real-term fixed assets are used here rather than the book value because the former better reflects the technology at the firms disposal.
The empirical availability of the ratios has been one of our criteria in the selecting the ratios to be included in this study. Measuring variable costs to fixed costs is not unproblematic in this respect, because variable and fixes costs are not always easily available separately in Finnish financial statement analysis data. On the other hand variable costs to fixed costs is an established concept in measuring operating leverage.
In measuring the operating leverage as the proportion of the variable costs there are alternative definitions. The denominator of this ratio can be selected in several ways. Operative fixed costs alone can be used (as we are doing). The second alternative is using the operative fixed plus variable costs. The figures produced by the second alternative are easier to interpret, but the first alternative produces a better variation. The third alternative would be to include depreciation charges into fixed costs, but this aspect is already covered in our first measure of the operating leverage.
Considering earnings variability rises both principles and technical issues not present in the financial ratios we have presented this far. The financial ratios included so far are (more or less) easily defined and obtained for annual data as befits a cross section study. Riskiness, however, obviously is a longer-term characteristic of the firm. What is more, devising an annual measure of risk would involve unnecessary practical difficulties. And, as is recalled from our discussion at the beginning of Chapter 2, relevance in financial statement analysis practice, as well as availability and unproblematic calculation are among our criteria in selecting the financial ratios. Hence we shall not include earnings variability in our data base.
Even if we omit earnings variability it is useful briefly to outline measuring this aspect of the firm. Obviously, any growth rate would first have to be eliminated lest it cause variability in itself. Furthermore, a model would be needed to counter the fact that, ceteris paribus, fluctuations in later years would be wider simply because of greater volumes caused by growth. (As is well known, in statistics the phenomenon is called heteroscedasticy.) The earnings variability could be measured as the relative variance of the residual around the earnings trend. But, this would mean using a rather involved model for obtaining financial statement analysis information, and in actual practice calculating financial ratios is usually kept quite simple.
If earnings variability were included, earnings would have to be defined for this purpose. Using earnings after interest and taxes would then be the choice. It would mean using a shareholder's point of view rather than a managerial point of view. The latter is already reflected in the operating and financial leverages when considered in terms of riskiness.
Gordon's well-known dividend growth model postulates a link between a firm's cost of capital and security prices. In other words a link between a firm's features and its securities' features. (See Martikainen 1989: 36-39.) This fact, of course, further prompts including growth in our data basis.
We shall use the long-term growth rate of deflated sales as the indicator for growth. The same difficulty of cross sections versus time series applies as in measuring riskiness. Growth could, technically, be easily calculated in terms of annual growth rates. The possibility of mergers and acquisitions in the data make this alternative more unstable than using a longer term growth rate. Recall, however, that the most problematic years after 1984 are not included.
Size is another non-ratio measure which has to be considered in this connection. Granted, there are research results with Finnish data which do not lend clear-cut support for a size effect on the firms financial ratios, but more research on that particular subject is needed. (See e.g. Lehtinen 1989, and Lukkaroinen 1988. Also see Buckley & Dunning & Pearce 1984.) On the other hand, size effects on risk adjusted returns of NYSE stocks have been reported. (See Banz 1981, and Reinganum 1981.) We shall include total assets in proportion to the whole sample as a measure of size.
Size is measured for each firm for each year by dividing the total assets (adjusted for appreciation) by the maximum observed total assets over all the firms and years. Thus the range of the resultant ratios will be between 0 and 1.
Let us review two of the earlier results on factoring financial ratios when cash flow ratios are included in the data. It is important to note in the ensuing discussion that cash flow ratios have often been regarded as profitability ratios.
Gombola & Ketz (1983: 113) concluded in their empirical study that cash flow ratios do load on separate and distinct factors, "when cash flow is measured as cash revenues from operations less cash expenses for operations". On the basis of their results they emphasize the need of including cash flow ratios "in predictive or descriptive studies involving financial ratios".
Yli-Olli (1983) studied the same problem with a special emphasis on observing the stability of cash flow ratio loadings. He concluded that the cash streams (as calculated in Finland) do not load on the same factor as profitability. He also demonstrated applying transformation analysis that the temporal stability of the loading of the cash flow ratios on different factors is very poor. He concluded that the cash flow ratios measure different aspects of the firms' performance at different stages of business cycles.
The first of our cash flow ratios is cash net income (cash margin II) divided by cash from sales. The second of our cash flow ratios is cash operating income (cash margin Ib) divided by total assets. The cash net income is defined as follows. (For more details see Kinnunen 1988, Appendix 4.)
cash from sales less cash based direct materials less cash based direct labor plus other cash based net (non-operating) income = cash operating income (cash margin Ib) less cash based interest less cash based direct taxes less cash based dividends = cash net income (cash margin II)The choice of the two different cash flows again also reflects the distinction of the ownership and the managerial function. As will be recalled, a similar aspect was taken up in discussion the operating leverage ratios.
The third of the cash flow ratios takes up a different aspect of the firms activities, that is its investment intensity. It is defined as cash flow to capital investments divided by cash based sales. The reason for including this cash flow ratio is twofold. On one hand we wish to include a cash flow ratio which is not a profitability (ex-ante) type of ratio as are the other two. On the other hand a measure of investment activity is clearly relevant in trying to have a set of ratios that covers well the different aspects of a firms activities.
Capital investments are at the very heart of a firm's success. Major investments are expected to be reflected on the firm's security prices. If the (discounted) after-tax cash flow from a capital investment at the weighted cost of capital is positive, then the firm's security prices are expected to increase, and vice versa. This fact further emphasizes the need of including a ratio involving the capital investment activity of the firm. Here cash basis is particularly relevant, because then the initial investment outlay is directly involved. Accrual basis smooths the capital investment into a long series of depreciation.
In this light it is surprising that capital investment intensity ratios have not always been included in studies factoring financial ratios. There are, of course, exceptions. The domestic study presented in Aho (1981: Ch. 7) is one example.
The fourth of the cash flow ratios is a cash-based indicator of a firms operating leverage. Cash outflow to materials & supplies and wages & salaries divided by cash from sales indicate the (cash based) structure of the firms expenses. In accounting theory there has been discussion about the order in which expenditures should be deducted from revenues. It has been put forward (Saario 1949, see Salmi 1978 for a review in English) that there is a strict priority order of costs based on the divisibility of the costs. Direct labor has the highest priority of all costs, and direct materials have the second highest priority. This priority order of costs is present in the fourth cash flow ratio, which thus well reflects the operating operating leverage (production technology) of the firm.
The fifth cash flow ratio measures the ability of the firm to meet its financial obligations arising from its debt financing by financial institutions and investors. Cash net income (cash margin II) divided by interest bearing debt indicates the burden caused by the debt financing taken by the firm. Thus this cash flow ratio reflects the financial risk of the firm.
The sixth of the cash flow ratios also measures the firm's ability to meet its financial obligations, but from a slightly different angle. Cash outflow to interest payments divided by cash operating income (cash margin Ib) reflects the firms financial risk based on the interest payments the firm must make in relation to the cash flows the firm is able to generate.
Technically we subdivide the market-based ratios ex ante into three categories. The first category includes ratios which are directly based on financial statements. The second category includes ratios where the numerator comes from financial statements, and the denominator from market based information, or vice versa. The third category includes market based indicators.
In selecting the market-based ratios their prevalence in security analysis practice was primary. (Naturally all the criteria presented at the beginning of Chapter 2 still apply.) For this purpose several foreign and domestic publications directed at the investing public were scanned for eligible frequently used ratios.
Dividend payout ratio is considered an indicator of the firm's dividend policy. Theoretically, the question of the dividend payout policy, and its effect (or irrelevance) on the value of the firm is one of the classical topics of finance. By including this variable we do not directly take a stand about the influence of the payout policy on the value of the firm. We include it to see its empirical behavior in relation to the other ratios of our study.
Nevertheless, there are arguments for the relevance of the dividend payout decision which should be revisited here. (The generic interpretation is that a relevance of dividend decisions reflects market imperfections.) Most importantly, the signaling approach view should be taken up here. According to this important view, changes in the dividends are signals to investors from the management indicating a long-term shift in the firms profitability and/or financial position. Another important view on the relevance of dividend policy is the so-called clientele effect. This effect is based on the idea that firms with different payout policies attract different kind of investors because of the difference in the tax treatment of personal capital gains and dividends. For an interesting survey about management views on the tenets of finance discipline on these issues see Baker & Farrelly & Edelman (1985).
Technically, calculating the dividend payout ratio can be problematic in firms with unprofitable years. This is because dividends my be distributed even in years with negative earnings (i.e. losses). The statutory limit on dividends is set (in Finland) by retained earnings, not the periods earnings alone. Thus the data must be checked for extreme or even negative values.
Dividend yield percentage at the year end prices is readily available for Finnish data.
The P/E ratio, that is price per earnings ratio, is perhaps the most prevalent market-based ratio. Interpreting what this ratio factually means is, however, somewhat ambiguous. Often it is described in the well-known terms of P/E = (D/E)/(k-g), that is payout ratio capitalized by return and growth. This definition has been used to interpret e.g. high P/E ratios. According to this view a high P/E ratio may indicate high dividend growth expectations from the part of the investors, or low risk so that a low return is considered sufficient, and so on.
There is also another way of looking at the P/E ratio. Consider its inverse E/P. Now it is earnings relative to capital invested (capital in the form of price prince per share). Looked at this way, the P/E ratio resembles profitability ratios.
We choose to use the E/P format, since this way round the problem caused by potential near-zero earnings is avoided.
The third ratio in this ex ante category is the market to book ratio. It is calculated as the stock price per share divided by the book values per share.
The first of these ratios is the return on the security. The return on a security has two main components, that is the capital gain (or loss) and dividends. As discussed earlier, there has been much debate whether these two components are valued differently by the investors or not. Including the return on the security and dividend yield as separate variable in our data base covers the potential difference.
Security beta is the second variable in this ex ante category. The concept of beta is central in capital market theory, and more specifically in the capital asset pricing model (CAPM). Expressing the riskiness of a security in terms of its beta can be interpreted as follows. The investor is not interested in the properties of a single security per se. Rather, the investor evaluates a security on the basis of its influence on the risk-return characteristics of his/her portfolio should the security be included in his/her portfolio.
In empirical testing the explanatory power (coefficient of correlation) of CAPM has been quite low. There has been much debate as to the reasons for this state of matters. (See Roll 1977 in particular.) One of the (many) potential explanations is that investors might after all consider a security's variance rather than its beta in assessing its riskiness. According to this view the investors consider securities individually rather than as parts of their portfolios, or the investors' portfolios are quite non-diversified (in other words consist very few securities.) Results regressing security returns with security variances are consistent with this view. (See Levy & Sarnat 1986: Ch. 13.)
Consequently, the third variable we include in this ex ante category is the security's total risk. It is calculated as the variance of the security's returns.
itv@uwasa.fi pyo@uwasa.fi [ts@uwasa.fi] [Photo] [Programs] [Research] [Lectures] [Department] [Faculty] [University]