Document Type : Research Paper
Authors
1 Associate Professor, Department of Economics, Sharif University of Technology, Tehran, Iran
2 Master of Economics, Sharif University of Technology, Tehran, Iran
Abstract
The book-to-market ratio is known as an anomaly variable in the financial literature. This variable has a high explanatory power in predicting the returns of companies in different capital markets across world; But understanding why it has the power to explain is still a matter of debate. In this study, we seek a clear understanding of the explanatory power of the ratio of book-to-market ratio in explaining the annual return of cross-sectional data of stocks on the Tehran Stock Exchange. Book value can be divided into two parts: retained earnings and contributed capital, which have different economic meanings for readers of financial statements. Our hypothesis is that the predictive power of the book-to-market ratio arises from a component of book value that could be a good proxy for underlying earnings yield. Using the method of Fama and Macbeth (1973), we regress the annual return of cross-sectional data of companies listed on the Tehran Stock Exchange for the years 2001-2019 on the ratio of book-to-market ratio and its two components. Neither component of book-to-market ratio could eliminate the predictive power of this ratio; however, the ratio of retained Earnings-to-market ratio could show predictive power along with the book-to-market ratio. We contribute to the literature by providing additional evidence from Tehran's Stock Exchange.
1- Introduction
The book-to-market ratio is known as an anomaly variable in the financial literature. It has appeared as a key explanatory variable with high explanatory power in predicting the returns of firms in capital markets across the world, however, understanding the mechanism through which this financial factor functions and its origin of the explanatory power is still a matter of research debates. Empirical researches on the returns and “book to market value” can be divided into two strands. The first group aims to examine the existence of abnormal returns on the ratio of "book to market value" in the stock markets. This stream of works aim to answer the question of whether the "book to market value" is able to predict companies' returns in capital markets or the returns is caused by other sources including random noise. Rosenberg et al. (1985) show, for instance, that in the US capital market, the strategy of the "book to market value" can yield abnormal returns for investors. In terms of this strategy, at the beginning of each month, the shares with a high "book to market value" are bought and the shares that have a low "book to market value" ratio are sold. A relationship between the ratio and average stock returns for the period 1981-1981 in the capital markets of Switzerland, France, Germany and the United Kingdom has also been observed by Coppole, Rollie and Sharp (1992).
The second stream of studies on the "book to market value" seeks to understand the cause of its explanatory power. This issue is an active research area and is still subject of discussions and has been studied from various aspects. One of the most highly cited of them is Fama and French (1993), which attributes high returns in stocks with a higher magnitude of "book to market value", to higher systematic risk. In contrast, Daniel and Titman (1997) introduces the hypothesis of equity characteristics and by providing empirical evidence argues that the returns premia on high book-to-market stocks does not arises because of the co-movements of these stocks with pervasive factors. It is the characteristics of the share rather than the covariance structure of returns that appear to explain the cross-sectional variation in stock returns. So, these are not associated with greater risk tolerance. Ball, Gerakos, Linnaeus, and Nikolaev (2020) examines the "book to market value" through its components (retained earnings and contributed capital) in the US capital market. He argues that the ability of "book to market value" to predict the cross-sectional returns is not because of its intrinsic information contents, but it appears as an appropriate proxy for the actual profitability of the firms, because, the retained earnings component of the book value of equity includes the accumulation and, hence, the averaging of past earnings, instead the contributed capital-to-market has no predictive power.
Hypotheses
We contribute to the literature by providing additional evidence from Tehran's Stock Exchange. Our study aims to provide further evidence to clarify explanatory power of the ratio in predicting the variations of annual returns in cross-sectional data for stocks in the Tehran Stock Exchange. Our hypothesis is that the predictive power of the book-to-market ratio arises from a component of book value that could be an appropriate proxy for underlying earnings yield.
Data and Identification methodology
We use the annual returns and financial statements of all shares traded from the beginning of 2001 to the end of 2020 in Tehran Stock Exchange. Annual returns are calculated from price data recorded and reported in the “tseclient” software and accounting data are downloaded from “codal.ir” website. In this research, financial companies listed in the TSE have not been included in our working sample due to their special nature. Because, by nature of their activities, they have high financial leverage, which is normal for companies active in the financial field. The characteristics might be interpreted as a financially critical situation, whereas, the it is not so for firm that are active in financial fields. The information extracted from the financial statements is matched with the annual return of 1 month after the end of the financial year. The reason for this identification strategy is to make sure that the published financial information affects the share price. For example, if the company's financial year is at the end of March, we will assume that this information was available to the public at the end of April.
Findings
Following the statistical method of Fama and Macbeth (1973), we regress the annual return for cross-sectional data of companies listed on the Tehran Stock Exchange over the years 2001-2019 on the ratio of book-to-market ratio and its two components as well. Neither component of book-to-market ratio could eliminate the predictive power of book-to-market; however, the ratio of retained Earnings-to-market ratio could show predictive power along with the book-to-market ratio. Table (1) reports the Fama and Macbeth (1973) regressions in which, outcome of interest is returns and determinants of the regression are the log of "Book to Market Value", log of " Retained Earnings to the Market Value " and log of "Contributed Capital to Market value". We include a few controlling variables that are identified theoretically as determinants of returns.
Table(1): Contributed Capital and Retained Earnings in the Fama and Macbeth Regression
(1)
(2)
(3)
(4)
(5)
(6)
Variables
-0.129**
-0.128**
-0.116**
-0.0901**
-0.126**
-0.103**
Log(Market Value)
(-2.680)
(-2.762)
(-2.492)
(-2.474)
(-2.257)
(-2.228)
0.498**
0.210**
0.508**
Log( Book-to-Maket)
(2.744)
(2.471)
(2.342)
10.53***
8.557**
9.914**
Log(Retained Earnings to market Value)
(2.992)
(2.426)
(2.890)
0.371***
0.00406
0.255***
Log(Contributed Capital)
(3.446)
(0.0438)
(3.343)
0.619***
0.560***
0.415**
Binary if profit>0
(3.382)
(3.058)
(2.256)
2.973**
-19.64***
-15.47**
2.429**
2.959**
-18.34**
Constant
(2.731)
(-2.907)
(-2.272)
(2.825)
(2.534)
(-2.806)
3,794
3,794
3,794
3,794
3,794
3,794
#OBS
0.121
0.144
0.188
0.099
0.135
0.189
R-Square
21
21
21
21
21
21
# Groups
*** p<0.01, ** p<0.05, * p<0.1, t-stats in parenthesis
Note: the firms fixed effect regression over 2001- 2021 across 181 firms are reported in the columns. Contributed capital includes all of the book value accounts except retained earnings.
Column (1) shows the regression of annual stock returns on the logarithm of "book to market value" in the presence of a control variable, logarithm of market value. The estimated coefficient for "logarithm of book to market value" equals to 0.498 with t-statistic t = 2.74, which is statistically significant at 5 percent critical region. The result is in the same direction with those in previous studies on the "book to market value". In column (2), "logarithm of retained earnings to market value" has been replaced for "logarithm book to market value". The coefficient of " logarithm of retained earnings to market value" is equal to 10.53 and is statistically different from zero at the 1 percent significance level with the estimated t = 2.99. In column (3), two variables "logarithm of book to market value" and "logarithm of retained earnings to market value" are included in the model. The coefficients of "logarithm of retained earnings on market value" and the "logarithm of book to market value" are significant at the conventional significance level. It suggesting that, "logarithm of book to market value" and "logarithm of retained earnings market value" are not able to fully represent the information contained in their competitors, as determinants of the firms' annual returns.
The columns (4) and (5), report similar regressions by substituting "logarithm of contributed capital to market value" in place of "logarithm of retained earnings to market value". Once, we include this determinant alone, it significantly impacts (coefficient 0.371 with t = 3.446) annual returns, but if we add "logarithm of book to market value", to the specification "logarithm of contributed capital on market value" loses its significance and its t statistic drops to 0.0438. Meanwhile, the "logarithm of book to market value" remains significant at the 5 percent level. In the column (6), in addition to the "Book to market Ratio "we keep both "logarithm of retained earnings to market value" and "logarithm of contributed capital to market value" in the specification. The coefficient of "logarithm of retained earnings to market value" remains almost with no tangible change 9.914 with and significant, and the coefficient of "logarithm of contributed capital on market value" is appears significant as well.
The inability of "logarithm of retained earnings to market value" to absorb the effect of "logarithm of book to market value" can be due to the weakness of this financial account in representing the companies' profitability information. This might originates in the fact that the retained earnings account is not an appropriate representative of the company's profitability. More specifically, this account is the balance of profits that have not been distributed among investors, it is not representative of all the company's acquired profits, and in each period that: (1) the company distributes profits among investors or (2) transfers an amount from this account to another account in equity, a part of the information in the accumulated profit will also be removed from this account. Consequently, this account cannot contain all the profitability information of the company. When the company distributes profits to shareholders, the company's profitability information is removed away from both the retained earnings balance and the book value. For this reason, we simply return the amounts transferred from the retained earnings account to other equity accounts to the retained earnings account and define the adjusted retained earnings account and the adjusted contributed capital as follows:
Adjusted retained earnings = retained earnings + legal reserve + plan and development reserve + other reserves + total capital increase from retained earnings until the end of the reported year + total other transfers from retained earnings until the end of the reported year
Adjusted Contributed Capital = Equity - Adjusted Retained Earnings
Adjusted retained earnings is the balance of all profits earned by the company during its life and not withdrawn from the company. The adjusted contributed capital is equal to the book value minus the adjusted retained earnings. To test our hypothesis, we separated "book to market value" into two parts (1) "adjusted retained earnings on market value" and (2) "adjusted contributed capital on market value". The significance level of the coefficient of "book to market value" decreases when it is included in the model beside to "adjusted retained earnings to market value", in contrast to the specification that includes the "retained earnings to market value", however, the coefficient of "book to market value" is still significant at the 5 percent significance level. The significance of the coefficient of "adjusted retained earnings to market value" also improves, in comparison to all similar regressions in which unadjusted "retained earnings to market value" are used as determinant. All in all, this evidence shows that a part of the information in "book value to market value" is caused by a variable that is related to the company's profitability, but not all the information in "book to market value" is caused by the company's profitability.
Keywords
Main Subjects
)