This study empirically examines the determinants o)f the differences in rates of return on the Indian corporate equity in general, and on the role of growth variables in this process, in particular. For this, it traces the relationship between each of the independent variables on the rates of return on equity while holding the other indepedent variables constant in a multiple regression analysis. It concludes that thej'mancial leverage as measured btl debt-eciuity ratio comes out as a more dominant variable vastly influencing the measured rate of return, followed by pay-out ratio.
The focus of the empirical studies generally has been on the reciprocal of these multipliers - P/E ratios. Broadly, these empirical investigations have attempted either to study and explain: (1) the relative value of P/E ratio represented by the ratio of an equity to 'normal" P/E ratio (Holt, 1942; Malkiel, 1963), or (2) the variability of P/E ratios of equities (Benishay, 1961;Whiteback, Kisor, 1963; Wippern, 1966; Bower, Bower, 1969: Malkiel, Gragg, 1970; Chung, 1974). The later studies have investigated the determinants of P/E ratios of equities over time as well as cross-section years. Most of the studies, however, have been based on cross-section analysis. They attempted to study the impact of one or more of the variables on P/E ratios of equities. These variables are: growth, dividend pay-out, risk leverage, size, and firm effects.
The empirical studies by Gupta (1981), and Yalawar (199-) are of special relevance in this context, for, they are pioneering studies on the rates of return and risk of the Indian corporate equities. Another study by Md Obaidulla (1991) indicates that the stock price adjustment to earnings information is biased and inaccurate, and is neither be time-specific nor associated with chance factors. In his study, Puranik (1992) concludes that the prices of the shares seem to have been influenced, potentially by the factors other than the fundamentals. However, these studies have not focused on identifying and studying the statistical significance of the important determinants of the rate of return or risk. No other study seems to have been made to analyse P/E ratios or their reciprocal of different equities of the Indian corporate sector with this focus.
It was around 1985, the Government introduced major industrial policy changes, sometimes described as 'liberalisation" or "deregulation", based on the recommendations of the high-powered committees the Narasimham Committee on shift from case by case discretionary physical controls to fiscal and financial control; the Abid Hussain Committee on export-import policy, and the Patel Committee on reform of stock exchanges - and according to all available indications, their positive impact on the economy was becoming visible in all quarters by 1986. In the light of the emerging and evolving competitive conditions, this study attempts to analyse the determining factors of the differences in the rates of return on corporate equity. Thus, this study encompasses the post-deregulation period - 1986-90.
The multiple regression analysis is the most appropriate statistical technique to assess and delineate the individual and combined effect of a set of independent variables on a dependent variable, and, therefore, has been employed in this study. In fact, this study adopts the model employed by Benishay where he advanced the hypothesis that the measured rate of return of corporate equities as a function of. (1) the trend in earnings, (2) the trend in market value of the equity, (3) the pay-out ratio, (4) the expected stability of the future income stream, (5) the expected stability of the equity value, (6) the size of the firm, and (7) the debt-equity ratio. Among the independent variables, the first three are expected to remove the errors obstructing a valid measurement of the theoretical concept of the rate of return on equity capital, and as such are "correctors", and the remaining shall measure the differential 'risk" or desirability of holding corporate equities and- as such are "explanatives". Notationally, the model can be presented as under:
Benishay tested his model by running cross-sectional linear regressions, in which logarithmic values were used for Y, X3, X4, X5, and X6, and actual values for the other variables. In the proces X3, X4, X5 and X6 are conceived to be having multiplicative effect in non-logarithmic form, and the rest being additive on the dependent variable. Actual values of all variables are considered in this study to avoid this multiplicity of the expected relationship between dependent and independent variables. Whereas, we have computed all variables for the cross-sections, Benishay used 1954 computations of XI, X2, X4, and X5, for 1954, and 1955 regressions, and their 1956 computations were used both in 1956,and 1957 regressions.
It may perhaps be useful to deduce the a priori relationships between the independent and the dependent variables at this stage. The dependent variable, measured rate of return 'Y, is a nonnalised rate of return where the numerator is a weighted average of EAT for the cross-section year, and the eight preceding years, and the denominator is the arithmetic mean of the high and low values of equity outstanding in the cross-section year.
The growth in earnings, "Xi", is expected to have a negative correlation with the measured rate of return. As the numerator in "Y" is not adjusted for trend in past income, it may diverge downwards from the expected income when the past income growth has been high, and upwards when it has been low. A high positive correlation in growth of EAT between two past adjacent periods can be a good predictor of the future growth rate in earnings, and as such may be related negatively to the measured rate of return.
A recent revaluation of the expected income in the measured rate of return can be corrected through incorporating the past trend in the value of the equity, (X2). The measure of earning in the numerator of "Y" may fail to reflect the change in the expected income, while the market value of the equity in denominator will reflect immediately. Consequently, the measured rate may be smaller than the true rate when the expected income has risen and vice versa. The larger this percentage change is, per unit of time, the less likely is the empirical representation of earnings to keep up with the expected income and the greater will be the negative correlation between growth in equity value and the measured rate of return.
Theoretically, dividend pay-out ratio is expected to be negatively influencing the
measured rate of return, especially in a market of "short-sighted" (Chandra, 199-) nature. Because the investors prefer distribution
to retention of earnings, the payout ration (X3), and the rate of return are
negatively correlated. The explanation for such relationship may be provided by an
examination of the effects of errors in the measurement of earnings. Two types of
errors are possible: (1) a difference between the expectation of the population of annual
book earnings and expected income, due either to a consistent accounting bias in the
measurement of earnings or to the fact that measured earnings however unbiased
"ex-post", may lag behind the expected income when the latter has changed
abruptly; (2) the difference between the earnings measured on the basis of a limited
sample and containing therefore a transitory element, and their population mean. The
introduction of X3 thus intended to correct for these errors.
The larger the variance of the distiibution of the expected earnings, (X4), the greater will be the probability of failure of the firm, and the cost or inconvenience incurred by the investor in maintaining his level of expenditure. Thus, a high stability of income is a desirable property and will tend to produce a low price-earnings ratio.
The speculative rather than the precautionary motive is dominant and, therefore, an equity variability is sought. The stock-holders are more encouraged by a likelihood of a high price than discouraged by an equal likelihood of a lower one. If the equity-holders have a preference for the stability of equity value, (X5), then it will have a negative coefficient.
Size, (X6), of the firm measured in terms of market value is expected to be negatively associated with the rate of return. Of the risk variables D/E ratio, (X7), a measure of financial leverage or risk, is expected to have a positive coefficient.
The results emerging out of Benishay's study affirmed the above a prior! reasoning excepting for the financial leverage. His observation with regard to this element was that it is "difficult to interpret" as the coefficients of leverage had negative sign. The conclusions of the other mentioned studies are more or less close to those of Benishay's.
A look at the table reveals the following:
Table 1: Regression Coefficients and their 't' Ratios
This may further be tested by increasing the sample size, probably redefining certain
variables, and following exactly the methodology of Benishay lock
, stock, and barrel. This is desired for two reasons: (1) the sample size, although
statistically considered to be large, is not large enough to be representative, and (2)
the sample scrips might have induced some bias on the results, for, they are mostly
speculative, overtly or covertly, in nature. However, the result should help in
designing the methodology for future research.
----,(1991): The Volume of the Nature of Speculation on Indian Stock Exchanges: Regulatory Implications. Expert Study of Trading in Shares in Stock Exchanges. New Delhi, Society for Capital Market Research and Development, Aug., pp 1-21.
-----,Gragg J G (1970): Expectations and the structure of share prices. American Economic Review. Sept., 467- 94.
Patel G S (1984): High Powered Committee on Stock Exchange Reforms. New Delhi, Govt. of India. p.66 (About 90 per cent of the trading activity on most of the stock exchanges was of a purely speculative nature).
|| | | | ||