the magic in earnings: economic earnings versus accounting earnings

CFA Institute

The Magic in Earnings: Economic Earnings versus Accounting EarningsAuthor(s): Fischer BlackSource: Financial Analysts Journal, Vol. 36, No. 6 (Nov. - Dec., 1980), pp. 19-24Published by: CFA InstituteStable URL: http://www.jstor.org/stable/4478396 .

Accessed: 12/06/2014 14:19

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 ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

CFA Institute is collaborating with JSTOR to digitize, preserve and extend access to Financial AnalystsJournal.

http://www.jstor.org

This content downloaded from 194.29.185.216 on Thu, 12 Jun 2014 14:19:54 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=cfa

http://www.jstor.org/stable/4478396?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


by Fischer Black

The Mlagic in Earnings: Ecunumic Earnings versus

Accsu.nhing Earnings

Users of financial statements-analysts, stockholders, creditors, managers, tax authorities and even economists-really want an earnings figure that measures value, not change in value. Analysts, for example, want an earnings number they can multiply by a standard price-earnings ratio to arrive at an estimate of the firm's value. Accordingly, the ideal set of accounting rules is one that makes the price- earnings ratio as constant as possible. The main thing lacking in present accounting practice is the recognition that this has been the goal all along.

Ideally, the earnings figure will take account of everything observable about the firm, including past earnings growth and past earnings volatility. Because everything that bears on the future will be incorporated in the current earnings figure, estimates of future cash flow, or future earnings, will be of no help in estimating the firm's value.

To be sure, the accountant's earnings figure will often yield variable price- earnings ratios. In arriving at his earnings figure, the accountant can use much information about the firm that analysts and the public will never know. But he cannot use information external to the firm, accounting rules that have not yet been agreed upon or even information about the firm that came in after the financial statement cutoff date.

The surprising thing is that, even though accountants have not formally recognized the goal of having an earnings figure that measures value, they have done a remarkably good job of achieving this goal. In particular, the variability of book value to price ratios exceeds the variability of earnings to price ratios, both across the universe of stocks and over time, suggesting that the earnings figure is a better measure of value than the book value figure. That's the magic in earnings.

W HO are the users of financial statements, and what do they want earnings

to measure? The users of financial statements include financial analysts, stockholders, creditors, managers, tax authorities and even economists concerned with national income. (Earnings figures, somewhat modified, are used as the corporate component of national income.)

I claim that all these users want the same kind of earnings figure. They all want earnings to be a measure of value, not a measure of change in value.

Fischer Black is Professor of Finance at the Sloan School of Management, Massachusetts Institute of Technology, and an associate editor of this journal.

FINANCIAL ANALYSTS JOURNAL / NOVEMBER-DECEMBER 1980 O 19



Value or Change in Value? Security analysts are clearest in their thinking about earnings. They want an earnings figure they can multiply by a standard price-earnings ratio to get an estimate of value. Since they don't want to work at figuring out what the price- earnings ratio should be, they will clearly be happiest if it's always 10. Thus they would like the accounting process to give an earnings figure they can simply multiply by 10 to get an estimate of value.

Other users of financial statements are less clear in their thinking. Accountants, for example, sometimes say that a firm's earnings should represent its ability to generate cash. But one of the propositions of modern finance is that the best way to estimate a firm's ability to generate cash is to estimate its value. A firm with high value can always generate cash, if not through its operations, then through issuing securities or selling assets. The firm's ability to generate cash will change as its value changes.

Economists, on the other hand, usually say that earnings should be related to change in value, and this view has lately been gaining acceptance among accountants. The move to include gains and losses due to changes in currency values in full in the earnings statement for the year in which they occur represented acceptance of the economists' view of value. It was soon discovered, however, that inclusion made a firm's earnings figures more erratic and less useful as a measure of how the firm was doing. If you want an earnings figure you can multiply by 10 to get an estimate of value, you don't want to include the full amount of each year's currency gains and losses. Now we are seeing a retreat from that way of handling currency gains and losses. Maybe the gains and losses will flow through to the balance sheet, but not via the income statement, at least not all in the current year. This is a move in the right direction.

This suggests another way of handling the tension between the two views of earnings. The accountants seem to be taking earnings as change in value when they call for articulation of the earnings statement and balance sheet, since in the simplest case the change in a firm's net assets from the start to the end of a period will be set equal to the firm's retained earnings for the period. The system is saved, however, by the exceptions. Many items affect the balance sheet without affecting income, or at least without affecting income in the same amount. The earnings

figure ends up reflecting value more than change in value.

Thus financial statements give two estimates of the firm's value-book value and a multiple of earnings. In many cases the two estimates of value will be very similar. In some cases, the earnings figure will give a more reliable estimate. In other cases (such as firms showing negative earnings), the book value will give a more reliable estimate of value. The end result of the current system of accounting, however, is an earnings figure that usually gives a reliable estimate of value, plus other information that can be used to arrive at an even better estimate of value. The main thing that's lacking is recognition that this has been the goal all along!

Testing Accounting Rules Once we make explicit the goal of creating an

earnings figure that can be multiplied by a standard price-earnings ratio to give an estimate of value, we have a basis for testing accounting rules: We want that set of accounting rules that makes price-earnings ratios most nearly constant, both across firms and across time.

Since the current set of rules does a pretty good job, we can start with it. Then we can use both theory and empirical work to test a pro- posed set of accounting rules (which would be mostly the same as the old set, but with a few changes) against the old set. If the new set results in more nearly constant price-earnings ratios, then we would want to adopt it.

We are using actual market prices of firms as an important ingredient in deciding on accounting rules, the implied assumption being that markets are largely efficient. It sounds at first as if the procedure is circular: We use market prices to determine accounting rules, and then use these accounting rules to figure earnings, which help determine market prices.

In fact, our procedure is not circular. To develop the accounting rules, we use only the prices of firms that have traded stock, then we apply the rules to firms that don't have traded stock. Also, we test accounting rules on current and past prices, but use the rules in the future; the rules we choose now influence future prices, not current prices.

Finally, a firm's stock price is not used in calculating its earnings. The simplest way to figure earnings for a firm would be to multiply its stock price by 10 per cent. But that procedure would indeed be circular, since application of a price-

FINANCIAL ANALYSTS JOURNAL I NOVEMBER-DECEMBER 1980 O 20



earnings ratio of 10 would guarantee an estimate of value per share equal to the price of the firm's stock.' Thus the firm's stock price is out of bounds.

On the other hand, other firms' stock prices may be used to figure earnings. If, for example, we are appraising the value of an oil reserve, we may look at the prices at which similar oil re- serves have traded. We may look at any market prices outside the firm, or transaction prices in- side the firm, but we can't use the firm's own stock price.

If we make the objective of a constant price- earnings ratio explicit, then earnings figures will become even more useful to investors, managers and economists. Changes in earnings will tie even more closely to changes in stock price, hence to changes in value.

The Accountant and the Analyst I will use "the accountant" to represent the accounting process within the firm that generates figures for the firm's financial statements and "the analyst" to represent the process by which investors and others use these figures in estimating a firm's value. Although the accountant and the analyst seem to be doing very different things, they are really both doing the same thing in slightly different ways. The accountant, in trying to generate an earnings figure that measures value, is trying to estimate value. The analyst, in using that earnings figure, somewhat modified or normalized, is also estimating value.

If the objective of accounting rules is to help in estimating value, then consistency may not necessarily be desirable. It will be desirable if the rules used for one firm will give a good estimate of value for another firm. If the rules need modification to give a good estimate of value for the second firm, then a little inconsistency will be desirable.

If we think of the rules as conditional on the nature of the firm's business and other factors specific to the firm, then consistency seems to lose its meaning. We will be consistent in the sense that, for every firm, we will choose from the same set of accounting rules. We will be inconsistent in the sense that we will use one subset of rules in one situation and another subset in another situation. Perhaps the practical meaning of consistency is that the rules should be simple. The more complex the accounting rules become, the harder they are to use, and the

harder it is to teach them and to change them as conditions change.

Of course, the analyst will understand the figures in a firm's financial statements only if she knows what rules the accountant used to generate the figures. Thus the conditional rules that the accountant uses form a set of "generally ac- cepted accounting principles." But, while the rules they use have the same form, there are also slight differences between his rules and her rules.

For instance, changes in the rules the accountant uses take time; users must agree on the changes, and then the new rules must be made part of the accounting system of each firm. Thus the rules the accountant uses within the firm are not always as up to date as the rules the security analyst uses outside the firm. There are also differences in the data the accountant and the analyst use. They both use data derived from actual transactions, because there is no other kind of objective information. But the accountant uses more data within the firm, while the analyst uses more data outside the firm.

The accountant has access, in principle, to all the transactions within the firm. However, the people within the firm don't want to tell the world about all those transactions, partly because it would be costly and partly because it would give out information the firm might re- gard as proprietary. Thus the accountant has access to some data the analyst will never see.2

On the other hand, the analyst is free to use rules that have not yet been agreed on, and to use data beyond the cutoff date for the firm's financial statements. She has fresher rules and fresher information. She may also be able to use information outside of the firm that the accountant will never be able to use.

Since the analyst wants to modify the accountant's rules, she will use some of the intermediate figures the accountant came up with in the process of calculating earnings; that's one reason for having figures other than earnings figures in the financial statements, and for including text describing recent events that have affected the firm's business. The analyst will combine the accountant's earnings figure, which makes use of some information she doesn't have, with other information in the financial statements and with data on events outside of the firm, to produce a normalized earnings figure. Since she is con- tinually putting in new information, the normalized earnings figure can change from day to 1. Footnotes appear at the end of article




day. Thus the estimate of value based on that normalized earnings figure can change from day to day, just as the firm's stock price changes from day to day.

The accountant's earnings figure seems to refer to a period of time, such as a quarter or a year. Indeed, if the earnings figure measures change in value, it is appropriate to associate it with a period, so that earnings for 1979 are for the calendar year 1979. But if, as I claim, the earnings figure measures value, it is appropriate to associate it with a date, so that earnings for 1979 are for December 31, 1979. It is better to think of earnings as referring to a point in time: A quarterly report contains data on events up to the end of the quarter, so the earnings figure it contains is useful in estimating the firm's value as of the end of the quarter. Thus, expressed as an annual rate, the earnings figure for the fourth quarter of 1979 will be the same as the earnings figure for the year 1979; the rules for computing fourth quarter earnings will be identical to the rules for computing earnings for the year.

If the earnings figure is computed in the best possible way, estimating future growth in earnings will be of no help in estimating the value of the firm: Everything currently known about the future will already be incorporated in the current earnings figure. To discount earnings, the analyst simply multiplies by a price-earnings ratio (or divides by an earnings-price ratio). This is very different from the discounting procedure appropriate for cash flows, which involves pro- jecting a series of future cash flows and applying a series of discount rates.

The price-earnings ratio will not depend on the earnings growth rate. While past growth in earnings may influence current earnings, it should not influence the price-earnings ratio. The objective of accounting is to use a set of rules that makes the price-earnings ratio as constant as possible.

Making the price-earnings ratio as constant as possible means making it independent of any observable characteristics of the firm. After all, any systematic dependence between the price- earnings ratio and some observable can be used to improve the estimate of earnings-i.e., to re- duce variation in the price-earnings ratio. Such a dependence is evidence that we haven't made the best use of observables in fashioning an earnings variable that is the best possible predictor of price.

Past earnings growth is observable, and so is

past earnings variability. Thus both the earnings figure that the accountant produces for the firm's financial statements and the normalized earnings figure that the analyst produces will depend on past earnings growth and past variability of earnings in a way that makes the price-earnings ratio independent of both growth and variability.

While current accounting rules produce price-earnings ratios that sometimes depend on past earnings growth and past earnings variability, Beaver and Morse found little relation between actual price-earnings ratios and future earnings growth or market measures of risk.3 Evidently, current accounting rules are already similar in this respect to the ideal rules I am describing.

Observables The accountant emphasizes information in-

side the firm, while the analyst makes more use of information outside the firm. The accountant uses information that came in before the cutoff date, while the analyst uses more up to date information. Both the accountant and the analyst, however, must use observable data in making their estimates of value. Past transactions provide the raw material.

The description of a past transaction includes the price, which is relatively easy to handle, and a description of the item bought or sold, which is harder to handle. When the transaction is the purchase of a piece of equipment, the description of the equipment is used in setting the right depreciation schedule. Past transactions are observables, and figures derived from them are observables. Averages, growth rates and measures of variability can all be derived from past transactions.

The current earnings figure, then, can depend on various characteristics of past transactions, including averages, growth rates and measures of volatility. At times, price-earnings ratios may depend on growth rates and measures of volatility, but it is convenient to change accounting rules when that happens so that (as already noted) price-earnings ratios are made largely independent of past growth and variability.

Economists sometimes claim that an earnings figure can be based on estimated future cash flows-that it is intended to be some sort of long-range average of future cash flows. The problem with this notion is that future cash flows are not observable; neither are growth rates of future cash flows. Thus there is no way to test




whether a set of accounting rules gives an earnings figure that is a good estimate of the average level of future cash flows. Cash flows are just too variable.

We might try to test an earnings figure against a smoothed estimate of the average level of past cash flows. But how would we do the smoothing? How would we test alternate smoothing methods? Indeed, the earnings figure itself is a smoothed function of past cash flows, so this would be like testing an earnings figure against itself.

We can't even rely on judgment, because there is no sound way for the people who would be making the judgments to learn to estimate the average level of future cash flows. A firm's stock price is observable, but its average future cash flow is not. Thus we can learn to compute earnings figures that will be a guide to value, while we can't learn to compute earnings figures that will be a guide to cash flows in the near future.

In fact, the firm's value gives, I believe, the best summary of the firm's future cash flows. There's a lot of choice in the timing of future cash flows, but the total is limited by the firm's value. The firm's value is the discounted present value of any possible arrangement of future cash flows.

A final reason for using accounting rules based on observables is that the procedures used can be repeated to verify the results. The auditor can check the accountant's work, and the bodies that develop new accounting rules can check each other's work. Earnings figures are thus verifi- able.

Why Do Price-Earnings Ratios Vary? The remarkable thing about efficient markets is that stock prices reflect so much information about internal and external transactions that affect the firm. With a good accounting system and good methods of security analysis, prices will reflect information on important developments quickly and accurately. But stock prices will give good estimates of value even when accountants and analysts are doing a poor job.

In my view, the objective of a set of rules for estimating earnings or normalized earnings is to give a figure that can be multiplied by a constant (say, 10) to give an estimate of the value of a firm's stock. When price-earnings ratios differ from this constant, the usual reason is that the earnings figure is distorted. It's very rare for an unusual price-earnings ratio to mean an under- priced or overpriced stock, because that would

imply an obvious profit opportunity that investors are overlooking.

When the price-earnings ratio is out of line, one should normally assume that the price is correct and the earnings figure is incorrect (as a guide to value). Prices are far more reliable as guides to firm values than earnings figures.

As we have noted, the accountant's earnings figure (the one that appears in the firm's financial statements) makes use of rules that have been agreed on by professional bodies and applied to information that came in before the cutoff date for the financial statement; both the rules and the information will always be somewhat out of date. Moreover, the accountant ignores much information external to the firm that nevertheless affects the firm's stock price. Thus it's not surprising that price-earnings ratios based on the accountant's earnings figure vary. The analyst can incorporate external information in creating a normalized earnings figure, but has only some of the relevant information; she is missing some of the internal information the accountant works with. So price-earnings ratios based on the analyst's earnings figure will vary a great deal, too.

Finally, price-earnings ratios vary because the goal of having a constant price-earnings ratio has never been made explicit. Some changes in accounting rules take us away from the goal, rather than toward it. For example, if the goal were made explicit, accounting rules would be changed so that earnings always turn out to be positive. Losses would be nonexistent: Since stock prices are always positive, we must start with a positive earnings figure if we are to multiply it by a standard price-earnings ratio to give an estimate of value.

The Magic in Earnings The magic in earnings is that price-earnings

ratios are remarkably constant now. Constant? How can I say that price-earnings ratios are constant when some are enormously large and others can't be computed because the firm has negative earnings?

Price-earnings ratios are constant in compari- son to a major alternative-price to book value ratios. Accountants, economists and financial analysts all agree that the book value of a firm's equity is a number that roughly measures market value or perhaps what the market value of the equity should be. They do not all agree that earnings should measure market value. In fact, however, empirical work by Stephen Stickells indi-

FINANCIAL ANALYSTS JOURNAL 1 NOVEMBER-DECEMBER 1980 U 23



cates that price-earnings ratios are more constant than price to book value ratios.4

There are, of course, several common measures of earnings and several common measures of book value. Stickells looked at earnings both including and excluding extraordinary items, and he looked at book value including and excluding intangibles. Per share figures for both earnings and book values were obtained by dividing the totals by the number of common shares outstanding plus common stock equiva- lents such as those associated with convertibles and warrants.

Because both earnings and book value figures may be small or negative, Stickells inverted the ratios. He studied the stability of earnings to price and book value to price ratios. This elimi- nated the problem of a firm switching from an enormous positive price-earnings ratio to an enormous negative price-earnings ratio as the earnings switched from a small positive number to a small negative number.

While the ratios figured with the two measures of earnings were highly correlated, the earnings figure excluding extraordinary items turned out, as one might expect, to be more closely related to price than the earnings figure including extraordinary items. The book value figure including goodwill was more closely related to price than the book value figure excluding goodwill. Though he could have used either definition for each variable without changing the results sub- stantially, Stickells used the earnings figure excluding extraordinary items and the book value figure including goodwill.

Using 1967-78 data from the 1979 Expanded Annual Industrials Compustat tape, Stickells computed these ratios for a sample of 329 firms representing industries of at least five firms and having fiscal years matching calendar years. He divided the figures for each ratio and each calendar year into quartiles. He calls the respective dividing lines between quartiles the upper hinge, the median and the lower hinge. His measure of variability is the "standardized hinge spread," or the ratio of the upper hinge minus the lower hinge to the median. Table I shows the standardized hinge spreads for the two ratios. In every single year, the variability of the book value to price ratio exceeds the variability of the earnings to price ratio.

Moreover, the variability of the book value to price ratios over time exceeds the variability of the earnings to price ratios over time. Stickells

Table I Standardized Hinge Spreads

Book Value Earnings to Price to Price

1967 0.98 0.64 1968 0.90 0.64 1969 0.96 0.73 1970 0.87 0.72 1971 0.94 0.76 1972 0.84 0.75 1973 1.10 0.77 1974 0.92 0.84 1975 0.85 0.77 1976 0.80 0.62 1977 0.81 0.55 1978 0.74 0.57

figured the median for each ratio for each year, and then divided the yearly medians into quartiles. Again he measured variability by figuring a standardized hinge spread, only this time he had only 12 data points for each ratio. The standardized hinge spreads for the median book value to price and earnings to price ratios were 9.9 and 3.5.

Thus the price-earnings ratios seem more constant than the price to book value ratios both across firms and across time. The earnings figure seems a better measure of value than the book value figure. Even though accountants have not formally recognized the goal of having an earnings figure that measures value, they have done a remarkably good job of achieving this goal. That's the magic in earnings. U

Footnotes

1. While the firm's stock price shouldn't be used in figuring earnings for its financial statement, taking 10 per cent of the value of a firm's stock is a good way to figure the contribution of the firm's eamings to national income.

2. We are not dealing here with the issue of what information firms should generate, or how much they should disclose. For example, the debate over oil and gas accounting revolves in part around whether firms should appraise the values of their holdings as of the ddite of a financial statement. Clearly one can get a better idea of how much an oil and gas firm is worth when the appraisals are available. But disclosure that benefits analysts, or soci- ety, may be costly to the firm. Whether such disclo- sures are desirable or not is immaterial to the ques- tion of how whatever infonnation is available should be manipulated to give the best possible estimate of the firm's value through the earnings figure.

3. William Beaver and Dale Morse, "What Determines Price-Earnings Ratios?" Financial Analysts Journal, July/August 1978.

4. Stephen C. Stickells, "A Comparative Analysis of the Relationships of Price to Earnings and Price to Book Values" (Master's Thesis, MIT, May 1980).




the magic in earnings: economic earnings versus accounting earnings

Documents