We use several sets of proxies to measure the degree of information asym- metry. We first consider proxies for firm maturity: firm size and age. As larger and older firms are typically better known, we expect them to face lower in- formation asymmetry (Frank and Goyal 2003; Lemmon and Zender 2010).
We also use two measures of the nature of a firm’s assets: asset tangibility and the market-to-book ratio. Tangible assets are likely to be easier for outside investors to value than either intangible assets or growth opportunities (Harris and Raviv 1991). The third set of proxies is related to measures of volatility: We include the volatility of both earnings and stock returns as measures of risk and information asymmetry (e.g., Brennan and Subrahmanyam 1996; O’Hara 2003). The fourth set covers attributes of analyst forecasts. We include the dispersion in analysts’ forecast as well as the accuracy of analysts’ fore- cast as proxies for the information gap between insiders and outside investors (Lang and Lundholm 1996). While the previous measures capture information asymmetry between managers and investors, some dividend theories focus on information asymmetry between informed and uninformed outside investors (e.g., Brennan and Thakor 1990). In addition to the above, we also use the number of analysts following each firm as a proxy for this type of information asymmetry (e.g., Babenko, Tserlukevich, and Vedrashko 2009).
Proxies for Market Frictions
This report states the dataset for the cross-sectional analysis. This dataset consists the databases from both the CRSP and the Compustat databases which include all the firms apart from financial firms (SIC codes 6000–6999) and firms which are related to mergers and acquisitions. To correspond with the analysis of in-situational holdings data which is mentioned in the Thompson 13 F database, the selection of sample period is done. Going back to the late 1925, the extract from the Moody’s Industrial Manual is taken for analysis of the time series information which is later mentioned in section 4.
For the analysis to be made regarding the dividend behaviour, it is necessary to divide the firms. This enables us to collect the adequate data in order to compute the smoothing behaviour of the dividend. For achieving these two goals, only those firms are taken that has a track record of payment of dividend in at least 10 years throughout the sample period for calculation. (See Appendix B for a full description of our sample-selection procedure). The number of sample firms is reduced from 13,872 to 3,877 by not including the firms which do not pay dividend. Further the sample firms are also reduced to 1,574 with the constraint of at least 10 years between the first and last dividend paid by the firm. All the above mentioned limitations also limit our sample from excluding the several small firms stated in the compustat database. For example, this stage of our sample in compustat consists of only 18% firms in the year 2015. Further the market capitalization of these firms represented 59% of all the compustat firms. In the subsequent sections, there is a requirement of substitute values of at least five years for market frictions and control variables. For final calculation 1,335 firms are taken and 21,400 firms are taken for each year observations (an average of 16 years of data per firm).
From the above it can be concluded that the firms taken from the universe of compustat firms is not a random sample. This is because the objective of the report is majorly to compute the dividend smoothing and make decisions by limiting the scope of our study. In addition to this, the regression analysis presented herein below depicted that there was no control over any possible prejudice from the sample selection, and also had no major impact over the results.
Summary Statistics
In this study, it is shown that in the market frictions how smoothing behaviour differs with different proxies in each model. This is done in order to assess the empirical relevance of several motivations for smoothing the dividend as mentioned in section 1. In section 1 it is discussed that the proxy that is used in this study is similar to that which is used in earlier payout policy and structuring of capital studies.
Various sets of proxies are used for measuring the degree of asymmetry information. Firstly the proxies for firm maturity then firm size and last the firm age is considered. Large and old firms shows lower information asymmetry as these are popularly known firms (Frank and Goyal 2003; Lemmon and Zender 2010).
Asset tangibility and market-to-book ratio is used as the two measures for determining the nature of firm’s assets. Instead of intangible assets, tangible assets are easy to value by outside investors (Harris and Raviv 1991). Measures of volatility being the third proxy set include earnings and stock returns volatility to measure risk and information asymmetry (e.g., Brennan and Subrahmanyam 1996; O’Hara 2003). The fourth set of proxy is related to the dispersion and accuracy in analyst forecasts (Lang and Lundholm 1996) as there are proxies between insiders and outside investors’ information gap. While the Information asymmetry between managers and investors are provided in the earlier methods, so the information asymmetry between informed and uninformed outside investors is depicted by various other dividend theories (e.g., Brennan and Thakor 1990). In addition, various analytical views are issued for this kind of information asymmetry where each firm acts as single proxy (e.g., Babenko, Tserlukevich, and Vedrashko 2009).
Coming on the agency cost proxies, Firms which have low market-to-book ratios generate surplus cash in place of investment opportunities generating profits (Jensen, 1986; Fama and French 2002). According to Lang and Litzenberger (1989), the severe problem of free cash flow is substituted by market to book ratio in order to help with overinvestment. Additionally, cash cows firms are profitable firms which show high credit ratings along with lower P/E ratio. And these cash cows firms are more sensitive with the problem of agency (Brav et al. 2005). Further, to capture the conflicts of principal and agent, the strength of governance measures is used (the Gompers, Ishii, and Metrick 2003 index).
For the most important three frictions which include taxes, asymmetric information and agency, the proxy which is used here is the level of institutional holdings. This proxy is also used for investor’s tax clientele (Hotchkiss and Lawrence 2007; Ferreira, Massa, and Matos 2009). Further it is also argued that, to reduce the amount of agency conflicts, institutes work as better monitors (Allen, Bernardo, and Welch (2000). In addition to this, to gather information and lower the information asymmetry between insiders and outsiders, institutions are the best (e.g., Barclay and Smith 1988; Brennan and Thakor 1990).
In order to correlate the study from the above article, it is important to discuss the relation between the levels of dividend and the firm characteristics. Table 1 represents the outcomes of the pay-out ratio (dividends divided by earnings).
Dividend yield portrait the same as dividend level. The complete sample results is presented by panel A (including non-dividend-paying firms), and firms that have a track record for at least 10 years throughout the sample period is presented by panel B.
Documents of table 1 shows that the firms which are large in size are less risky compared with firms paying higher dividends have lower beta and earnings volatility. Apart from this the firms which have low stock prices and shows growth pay lesser dividends as compared with firms which have lower leverage. Considering the previous studies, the institutional investors choose that firms which have a tendency to pay less dividends. Panel C include zero payers and panel D includes positive payers of dividends have alike assumptions like above. But one of the differences is that the firms which repurchase their share back are preferred most by the institutions. This sample study is done in the same manner like all other previous studies and the results propose that same firms are used in the samples which have been used in pay-out policies studies.
Initially the brief figures taken out as mean and median are 0.14 and 0.11 respectively from the sample. So it can be concluded that the median firms in the sample has a gap of nearly six years to cover the gap between actual and target dividends. According to Fama and babiak (1968), the sample median less than mean (median) SOA of 0.37 (0.30). Further the observation of assessed smoothing distribution measures is plotted by figure 4 of panel A. as the adjustments speeds are relatively slow, so in the distribution there is considerable right tail. It is also concluded that the cross sectional distribution in relation to relative volatility is more disperse to that of SOA
The sample is segregated into two group’s namely high and low earnings volatility. This is done to safeguard that smoothing of dividend is not the only function of earning volatility. The observation of each subsample related to distribution of dividend smoothing is depicted by figure 4 of panel. Generally only those firms will smooth less which has high earning volatility. So further in can be concluded that same earning volatility of various firms may not smooth equally.
By observing and proper evaluation it is verified that the relativeness between dividend levels and smoothing is acute. This means that firms paying higher dividends smooth as compared to firms which pay less or no dividends. The sample is again separated into two in panel C of figure 4 as high and low dividend yield respectively. And the empirical
Conclusion
In the United States, the publicly traded firms have a wide spectrum of smoothing behaviour of dividend and total pay-out. The significance of variation of time series in smoothing dividend is documented. From the last 80 years, a distinctive set of data shows that the dividend smoothing has been gradually increasing. This was done in the mid of 1980, when the firms did not repurchased shares. In 1930s the adjustment average speed went down from about 0.5 and in the 1940s it went down to about 0.2 in the past two eras.
It is also indicated by the results that to smooth dividends in the cross sectional variation, it showed a wide tendency. When agency costs are higher, smoothing becomes more prevalent. And when agency costs are lower, there is more smoothing of the firms that is large in size, is older, have low market to book ratio and which have weak governance. Firms suffer with a problem of asymmetry information and smooth less if these firms are young, have high growth potential, have less tangible assets, more volatile, few investors, less following from the analyst and have more scattered forecast from the analyst. At last the strong relationship between institutional holding levels and smoothing shows that the smoothing decisions are not affected majorly by the tax clientele. The cross sectional analysis does not explain the Earnings smoothness and predictability even though smoothing of dividend is related to these properties. In addition to this, it is also found that dividend smoothing is more asymmetric with respect to changing in earnings. Further, with positive change in earnings the dividends adjust quickly than to negative changing in earnings. So this indicates that SOAs from below target level are higher than from the above level.
These observations and facts together to explain are a challenge for theories which are based on information asymmetry, taxes, or financing constraints. But the results are stable with various assumptions models which are based on free cash flow agency costs (Easterbrook 1984; Allen et al. 2002; DeAngelo and DeAngelo 2007; Lambrecht and Myers 2010). This is the most significance peace of literature for further development in this area.
Even though the firms smooth dividends more by the use of repurchases, there is no evidence that states that more repurchasing means more smooth dividends. Further it can be said that dividend smoothing is independent of repurchase policy and varies with firm characteristics.
From the above study it is found that proxies for financial frictions (agency, asymmetric in- formation, and taxes) express a lower portion when talked about the cross-sectional variation in total payout smoothing. It is also found that the firms actively smoothens the dividend than total payout. From the last past decades, the repurchases of shares and cash dividends have helped the firm equally to achieve the firm’s payout. But the time series properties need more attention in the total payout.
According to researchers and professionals, the firms are smoothing the dividend from last five eras. The new evidence from this article gives a closer understanding about the types of firms which smooth dividends and the reason to do so.
