THE INFLUENCE OF GROWTH DURATION ON SHARE PRICES*

CHARLES C .

HoLTf

INTRODUCTION

performance of "growth stocks" in recent years has focused attention on the problem of evaluating the securities of fast-growing companies. Unfortunately, methods for placing valuations on such securities are not yet adequately developed, and investors make their buy-and-sell decisions as best they can. That a company's high rate of "growth" may come to an end is an important, but little-emphasized, investment consideration in the evaluation of growth stocks. To call attention to this point, we present in this paper an exploratory analysis of the relationship between price-earnings ratio, rate of growth, and the duration of growth. In omitting risk from the present analysis, we are explicitly neglecting the fact that investments in growth stocks are often riskier than in non-growth stocks. Consistent with this, the capitalization rates for both kinds of securities are assumed to be the same, and hence any differences in their price-earnings ratios are attributable to differences in their growth of earnings. The obvious investment success of growth stocks has led investors to seek out iJiese securities for purchase, with the result that their prices have been driven up so that growth stocks now generally carry high price-earnings ratios. But just how high it is wise for investors to drive price-earnings ratios is not clear. If a growth stock is evaluated by discounting future growing dividends back to the present, the paradoxical result is obtained that an infinite price-earnings ratio is justified for a stock whose dividends per share are expected to grow at a (per cent per annum) rate that is higher than the discount rate. This clearly untenable result comes from the implicit assumption of an indefinite continuation of exponential growth and may be
T H E SPECTACULAR INVESTMENT Conversations with M. H; MOler and F. Modigliani supplied much of the stimulation for writing this paper. t Professor of Economics, University of Wisconsin.

465

466

The Journal of Finance

avoided by limiting the assumed growth period.^ Another method^ has received considerable attention in investors' literature and seems to have had considerable infiuence. The growth in earnings per share of a company is extrapolated, say five years into the future, at the growth rate indicated from the recent past. The current price of the stock is divided by this forecast of earnings five years hence, to obtain a price-earnings ratio. In this way, more normal, i.e., lower priceearnings ratios are obtained for growth stocks and some useful indication is given on whether the existing price of the security is justified or not. However, this method is rather crude in ignoring any dividends that might be received during the five-year period or growth that might occur after this arbitrarily selected period.^
DURATION OF GROWTH

If investments in the common stocks of growth companies were expected to continue growing indefinitely at a constant exponential rate, then the investor's problem would be largely one of selecting the companies with the highest forecasted growth rates.* But before anyone chose to follow such an investment policy, he would be well advised to question seriously the assumption of indefinitely continued growth. Studies of the past growth in the applications of inventions and in the sales of companies and industries show growth curves in which very high rates of growth are achieved initially, but ultimately the growth rates tend to slow down or stop as maturity is reached. This logistic type of growth curve is rather complicated, and the forecasting of its leveling-off is quite difficult from both the statistical and the forecasting points of view. Although we shall not attempt this degree of refinement, it does seem desirable to assume that the
1. 0. K. Burrell, "A Mathematical Approach to Growth Stock Valuation," Financial Analysts Journal, XVI (May-June, 1960), 69-76; John C. Clendenin and Maurice Van Cleave, "Growth and Common Stock Values," Journal of Finance, IX (1954), 365-76. 2. Julian G. Buckley, "A Method of Evaluating Growth Stocks," Financial Analysts Journal,X.\'I (March-April, 1960), 19-21. 3. The more subtle point that the dividend returns from a growth stock are farther in the future tlian those from non-growth stocks and hence that forecasts of the dividends are riskier has not been adequately treated as yet, nor will it be considered here. See David Durand, "Gro-ivth Stocks and the Petersburg Paradox," Jotirnal of Finance, XII (September, 1957), 348-63; and Henry Allen Latane, "Individual Risk Preference in Portfolio Selection," Journal of Finance, IV (March, 1960), 45-52. 4. For a moi"c refined analysis ol growth see M. H. Miller and F. Modigliani, "Dividend Policy, Growth, and the Valuation of Shares," Journal of Business, XXXIV (October, 1961), 22. Growth in earnings of itself does not nece.ssarily justify a high price-earnings ratio, if dividends are correspondingly lowi.e., the growth requires some form of expanding opportunitie."; for profit. The present paper will not analyze the source of "growth" but only the problem of determining its value.

Injluence oj Growth Duration on Share Prices

467

growth opportunities of a "growth" company are likely at some point in time to slow down to the rate that is normally achieved by companies generally. Presumably, this more modest growth rate can be maintained indefinitely. As the period of high growth passes, the price-earnings ratio of a company will drop back to the normal level characteristic of "non-growth stocks." This perhaps distant, but almost inevitable, decline in price-earnings ratio constitutes one of the important risks of investing in a growth stock, especially since the termination of rapid growth is so difficult to forecast. To simplify matters, we shall make the assumption that the growth in earnings per share (adjusted for stock dividends and splits) of a company will continue at a high constant exponential rate until some point in time when the rate drops abruptly to the average rate for non-growth companies. Under this assumption, the duration oj growth for a company becomes a simple concept, i.e., the time duration of the high growth rate. Clearly, companies with long durations of growth should be valued more highly than those with short durations of growth, other things being equal. Also companies with high growth rates of earnings should be valued higher than companies with low growth rates, other things being equal. Both the duration and the rate of growth need to be taken into account in valuing a growth stock. One way to do this is to consider the following question. How long, at a minimum, will the present high rate of earnings growth of a company have to continue in order to attain the same level of earnings that can be achieved by an alternative investment in non-growth stocks of comparable risk? Assume that beyond Ithis time the high growth rate drops to the normal rate, the low dividend pay-out rises, and the high price-earnings ratio falls so that the two investments become virtually equivalent. In saying this, we have, of course, roughed over the uncertainty problem bj^ assuming comparable risk for both investments. This time is the minimum required growth duration for the growth stock to justify its high price-earnings ratio. Of course, in both cases we need to take into Eiccount the dividend jdelds. If we can formulate an analysis for determining the duration of growth estimate that is implicit in the market price of a growth stock, this may be useful to investors in making judgments as to whether the high price-earnings ratio of the growth stock is justified or not. We can obtain the market's estimate of duration of growth as follows: If we let E'(t) be the earnings per share (adjusted for stock splits and stock dividends) of a common stock in the year i (measured from

468

The Journal oj Finance

the present, when / = 0) and let ^E be the per cent per annum growth rate of earnings per share, then an estimate of future earnings per share as long as this growth rate continues is given by the following expression: E'(O='(O)(1-|-A)'. (1) It is convenient for analysis to assume the reinvestment of dividend income to obtain additional "growth" so that it can be combined simply with the above expression.' This is done by pretending that the dividends are used to buy more (perhaps fractional) shares of the same stock. This assumption is purely for analytical purposes, to put all securities on a common "no dividend payment" basis. Thus if D is the constant per cent per annum dividend yield (i.e., ratio of dividends to market price), the number of shares N{t) at the end of year t is {l + D ) ' , (2) assuming that one share was bought originally when ^ = 0. The total earnings E{t) at the end of year t on the original and purchased shares combined are E{t)=E'{t)N{t) ='(0) [(1-FAE)(1+ ! > ) ] ' . (3) Since D and AE are "small" and for the one original share E'{0) = E(0), we obtain
)' .
(4)

This growth measurement of investment return is equally applicable to growth and non-growth stocks. We apply it to both and introduce the subscript g to indicate a growth stock and the subscript a to indicate an alternative non-growth stock of comparable risk. After the duration oj growth, which we designate as T, we have assumed that the growth stock has the same general characteristics as the non-growth stock. Hence in the year r their market values will be in direct proportion to their earnings of that year.'' Since uncertainty has been left out of our analysis and since no dividends are withdrawn, we would expect that the market would tend to value the shares of the two stocks for current purchase in direct proportion to their value in year r and hence in direct proportion to the forecasted earnings in the year r. If this proportionality condition were not
5. There is little meaningful distinction between dividend income and capital gains aside from factors that we are not considering now, namely, risk, taxes, brokerage commissions for odd-lot transactions, and administrative convenience. 6. (f) is given by substituting r for t in expression (4).

Influence oj Growth Duration on Share Prices

469

satisfied, investors would tend to buy the relatively underpriced stock., in order to be in a better position in the year T, thereby driving up the low price. Thus the market will tend to satisfy this relation between the current share prices of the growth and the non-growth stocks, Pa{0) and P<i{Q), respectively: Pai.0) ,..

or, equivalently. P^(O)/E,(O)\ _ / l + A E . - H ' . Y P<,(O)/a(O)/ U + A a + I a / '

,g.

Here we see that the ratio between the price-earnings ratios of the two stocks is equal to the ratio of their composite growth rates raised to the rth power. The compounded growth offsets the high price-earnings ratio of the growth stock. We may solve for r by taking logarithms. For simplicity, we have dropped the time parentheses indicating current prices and earnings but without changing the meaning:

Since this equation is linear in the logarithms, we may graph it on semilog paper to obtain the simple relation shown in Figure 1. Fortunately, the log scales on the graph avoid any necessity of dealing with logarithms. The intersection of a horizontal line representing the relative price-earnings ratio with a sloping line representing the relative growth rate determines a point. Dropping vertically, we can read the market estimate of growth duration r. In this way, by plotting the point of intersection, we can obtain r for a security. An example will clarify the results of the analysis. Suppose that the current priceearnings ratio for a growth stock is 45 and for a non-growth stock of comparable risk is 15. Then the relative price-earnings ratio is (Pg/Eg) -^- (Pa/Ea) = 45/15 =: 3, as shown by the heavy horizontal line on Figure 1. Suppose, further, that the growth in earnings per share of the growth stock is expected to continue at the rate of 30 per cent per year but the dividend yield is only 1 per cent per year. The non-growth stock, on the other hand, is expected to have only a 5 per cent growth in earnings but has a 5 per cent dividend yield. Thus the relative growth rate is (1 + AEg -f Dg) -f- (1 -f AEa -f Da) = (1 + 0.30 + 0.01) --- (1 + 0.05 -f 0.05) = 1.31/1.10 = 1.2, as

470

The Journal of Finance

shown by the heavy sloping line through the origin of Figure 1. The intersection of these two lines determines a point labeled A. Dropping vertically, we can read from the horizontal scale that the market estimate of duration of growth is evidently 6 years. That is to say, the market is valuing the growth stock as if its present high rate of growth would continue for 6 years and then decline sharply to the

6.0

5.0

4.0

.2"

3.0

P:

2-0

1.0

1 i
/
4 2 3 5

/ 1 1/ 1 1 '1 / / /
1.40 / / 6

Relative Growth Rate

/ /

A /
/

/ / /

1.10

1.05

10

11

12

Duration of Growth T in Years

Injluence oj Growth Duration on Share Prices

471

normal level. The graph has made it easy to find that T == 6. This value, of course, satisfies equation (6) for this exa.mple:

-30 + 0.01Y Y
15'^Vl-f-0.05-f 0.05 7 Another way to interpret the six-year growth duration is in terms of total growth potential. A 30 per cent growth rate is, say, 25 per cent above the normal growth rate, taking the non-growth stock as the standard. If this rate continues for 6 years, we would have a total growth potential of (1.25)*' = 3.82 or 382 per cent. This amounts to forecasting that the ratio of earnings per share of tlie growth stock to the non-growth stock will ultimately improve by almost a factor of 4.
A N ILLUSTRATION USING MARKET DATA

The use of this analysis may be illustrated by selecting the DowJones index as a representative alternative non-growth investment, and selecting the following growth common stocks: Ampex, International Business Machines, Litton, Polaroid, and Texas Instruments. The dividend yields and price-earnings ratios for May, 1960, were used. For the Dow-Jones index, the price earnings ratio Pa/Ea was 18, Da was 3 per cent per year, and AEa was 5 per cent per year. These forecasts were extrapolations of the previous 5 years' history. We have obtained AEg for each of the growth companies by plotting on semilog paper the earnings per share, adjusted for stock splits and stock dividends, for 1956 through 1959 and estimated the slope. Before plotting points for each growth company. Figure 1 was modified by changing its scales to incorporate the Dow-Jones index as the standard non-growth alternative^ (see Fig. 2). Plotting these data on Figure 2 indicates the growth duration periods shown in the second column of Table 1. In interpreting these results, one is tempted, at first blush, to say that Litton is the better buy because it will justify its high price-earnings ratio relative to a Dow-Jones investment in a shorter period of time than the other companies. However, valid conclusions require a comparison for each company between the market estimate of duration of growth and the investor's own estimate. The investor can compare his judgments with those of the market and act accordingly where he feels the market is in error. Clearty, if growth duration actually proves to be longer than the market's estimate, the stock will have proved to be a bargain
7. The growth-rate scale on Fig. 2 plus 1 equals the relative growth-rate scale of Fig. 1 multiplied by (1 + 0.30 -f- 0.01).

472

The Journal oj Finance

purchase. A stock with growth duration shorter than the market estimate will prove to have been a poor investment. As always, the investor must bet that his forecast is better than that of the market. If he can forecast growth durations successfully, growth stocks may offer important investment opportunities. This analysis may be used in another way. An estimate of the Growth Rate {^E, -f
50% / 40%

100 90 80 70 60 50

1
J

I I 1

%?_
/

30% / /
/ / /

25%

40

30

20

i
^

/ / / /1/ / / ., ^~
1 /
/

1
/ /

</TI _/

IBM /
/ /

20%

/ /

/
15%

- ^
. .

10%

.--

10

12

Duration of Growth T in Years

Influence of Growth Duration on Share Prices

473

growth duration for a company may be made by the investor and the corresponding line drawn vertically on Figure 1 or Figure 2 until it intersects the sloping growth-rate line and then move left horizontally to read the price-earnings ratio. This is the price-earnings ratio that would be warranted by this duration of growth. This could be compared with the price-earnings ratio existing in the market and purchase-sell decisions made accordingly.
ADJUSTMENT FOR TAXES

An extension of the analysis might well take into account the fact that long;-term capital gains are taxed at an advantageously low rate, so that income through price appreciation is more desirable than
TABLE 1
Market Estimate of Duration of Growth (Years)

Company

Adjusted for Taxes (Years)

Ampex. . . International Business Machines... Litton Polaroid Texas Instruments

4.1 7.5 3 4 6 3
5.3

4.2 7.5 3 4 6 8 5.7

dividend income. An exact adjustment for taxes would greatly complicate the analysis, but, by ignoring the fact that capital gains taxes are postponable, we may make an approximate adjustment. By making the simplifying assumption that capital gains are taxed in the year in which they occur, we can modify formula (7) to refiect growth after taxes. The growth from dividend reinvestment is reduced by taxes to (1 KD)D, where KD is the marginal tax rate applicable to dividend income. Similarly, the growth of earnings is reduced to (1 Kc)^E, where Ko is the marginal tax rate applicable to capital gains income. A person whose marginal income tax rate was 30 per cent would be taxed at half that rate on capital gains, i.e., KB = 0.3 and Kc = 0.15. The maximum Ko is, of course, 0.25. Recalling that the price-earnings ratio of the growth stock falls as the termination of the high growth rate approaches, we need to recognize that all the gain in earnings is not reflected in gain in market price of the shares. Since we have applied the capital gains tax to the full earnings appreciation, we need to make an offsetting adjustment

474

The Journal of Finance

by applying the capital gains tax to the decline in price-earnings ratio. If we rewrite the left parentheses of formula (7), PJE, Pa/Ea+{Pa/Eo-PJEa)

we can interpret the right parentheses in formula (8) as the capital "loss" component for the growth stock as the result of the reduction in price-earnings ratio to that of the non-growth stock. We apply the capital gains tax to this "loss" adjustment by multiplying the "loss" term by (1 Kc): PjEa+ {PJE,-PJE,) ji-Kc) , y/Pg/Eo

Incorporating the above adjustments in formula (7), we obtain

n + (i-Kc)AE+ (l-KD)D,l + (1 - Kc)AEa+ {1 The inclusion of the tax adjustment does not complicate the graphs, but it does complicate somewhat the computation step before entering the graphs. The effect of the adjustment will be to reflect the preference of the high-income person for companies that do not pay out their growth in the form oi dividends. Applying this adjustment to the companies that we have already considered and assuming an investor whose tax rates are KD = O.S and Kc = 0.2S, we obtain the overtake periods shown in the third column of Table 1. In this case, the overtake periods were not greatly affected by the adjustment for taxes.
CONCLUSION

Four limitations of this analysis need to be noted. Firstand most importantthe uncertainty of forecasting earnings in the distant future is a much more important consideration for growth stocks than for non-growth stocks. Hence, ignoring the risk inherent in probabilistic forecast errors is an important omission which will tend to make growth stocks appear more attractive than they really are. Second, the analysis rests on forecasts of constant earnings-per-share growth rates and constant dividend-per-share jdelds. Such forecasts on a per share basis need to be made critically because past data may refiect purely financial transactions, such as increasing leverage, etc..

Influence of Growth Duration on Share Prices

47 S

which may reflect increasing risk rather than growth. Third, the relative tax advantage of capital gains income, which is the principal return from growth stocks, is somewhat understated by neglecting the deferred collection of the capital gains tax, so that growth stocks appear less attractive than they really areespecially for highincome investors. Fourth, we hardly expect the growth of earnings to terminate sharply but, rather, would expect the growth rate to decline gradually as the special advantages enjoyed by the growtli company are whittled away by increasing competition, expiration of patents, appearance of substitute products, etc. Thus the high growth rate would tend to last longer than the above analysis would indicate, but the rate of growth would be reduced. The need for a really adequate theory of investment under certainty is emphasized by the gross approximations that were used in this exploratory analysis. Granting these limitations, the analysis does offer a systematic framework for evaluating growth stocks that includes many of the relevant variables with a minimum of complexity and computation. Hopefully, the analysis is simple enougheven using the adjustment for taxesto be useful to investors.*
8. An independent, but equivalent, analysis by iRobert Ferguson has recently yielded a nomograph which reduces the computations to a simple mechanical procedure ("A Nomograph for Valuing Growth Stocks," Financial Analysts Journal, XVII [MayJune, 1961], 29-34). Unfortunately, the underlying assumptions of his analysis are not fully explicit.