Determinants of the Trends in Aggregate Corporate Payout Policy

Transcription

1 Determinants of the Trends in Aggregate Corporate Payout Policy Jim Hsieh And Qinghai Wang * April 28, 2006 ABSTRACT This study investigates the time-series trends of corporate payout policy in the U.S. industrial firms. In contrast to the common belief that corporate payouts have been declining, we document strong evidence that in the last three decades, the aggregate corporate payout ratio, measured as the sum of dividends and share repurchases divided by earnings or assets, has increased by more than fifty percent. This upward trend is robust to other measures of aggregate payout ratios and is most prominent in top payout firms. We examine three possible sets of explanations for this increasing time trends: changes in firm characteristics, shift in the overall payout environment, and the role of share repurchases in total payouts. Our analysis suggests that the decline of capital expenditures, regulation, firms increasing reliance on external equity financing and the flexibility from repurchase programs help explain the upward trend of corporate payouts. We find scant evidence that catering incentives, institutional holdings, or option-based executive compensations help to drive the overall increase in payouts. * Hsieh is from School of Management, George Mason University, Fairfax, VA Phone: (703) ; jhsieh@gmu.edu. Wang is from School of Business Administration, University of Wisconsin Milwaukee, Milwaukee, WI Phone: (414) ; wangq@uwm.edu. The authors thank Jean Helwege and René Stulz for helpful comments.

2 Determinants of the Trends in Aggregate Corporate Payout Policy 1. Introduction Recent studies on corporate payout policy point out dramatic changes in how firms distribute cash flows to shareholders. Most notably, Fama and French (2001) document a large decline in the proportion of publicly traded firms that pay dividends over the period. DeAngelo, DeAngelo, and Skinner (2004) show that despite the fall in the percentage of dividend payers to total number of U.S. industrial firms, the real dollar value of aggregate dividend payment has increased. As emphasized in DeAngelo et al. (2004), even though the reduction in the number of dividend paying firms and the fall of dividend yields reflect increasing concentration in the supply of dividends, the evidence is commonly interpreted as dividends are disappearing and payouts are declining. 1 This paper investigates the time series trends of aggregate corporate payouts over the past three decades and the determinants of such trends. Our approach differs from previous studies such as Fama and French (2001) and DeAngelo et al. (2004) in two important ways: first, instead of focusing on the number of dividend paying firms and the dollar value of dividends, we examine payout ratios, i.e., the percentage of amount that firms pay out from their earnings. Second, given the increasing importance of share repurchases as a form of payout (see, Grullon and Michaely (2002) and Brav et al. (2005)), we include both cash dividends and share repurchases in measuring total payouts. Our evidence shows that the aggregate payout ratio exhibits a time trend drastically different from the common belief that U.S. industrial firms have reduced their cash distribution to shareholders. In fact, we find that overall, companies have raised their payouts in proportion to 1 See the discussions in DeAngelo et al. (2004) and the references therein. Several recent studies also explain why dividends are disappearing. See, for example, Grullon and Michaely (2002), Baker and Wurgler (2004), Amihud and Li (2006), and Li and Lie (2006). 1

3 earnings during the period. The ratio of aggregate payouts to earnings increased drastically from 46.3% in 1972 to 138.4% in This upward trend is robust to alternative measures such as payouts to assets and payouts to book value of equity (BVE). The ratios of payouts to assets and to BVE have increased from 2.2% to 3.2% and from 5.1% to 8.5%, respectively, over the sample period. We show that the documented upward payout ratios are statistically and economically significant. The increase of aggregate payouts reflects two significant structural changes in corporate payout policies during the sample period. First, as presented in this paper and DeAngelo et al. (2004), corporate earnings are increasingly concentrated in a small number of firms. We further show that these high earners are also high payers, dominating aggregate payouts in both dividend payments and share repurchases. More important, the payout ratio of these high payout firms has experienced the most dramatic increase. Second, the increase of corporate payouts is largely driven by the popularity of share repurchase programs. During the sample period, the aggregate dividend ratio has remained relatively stable, both for the overall sample and for the top payers. 2 In contrast, the aggregate repurchase ratio has risen more than tenfold. The repurchase ratio has surpassed the dividend ratio in the last seven years during our sample period, again for both the full sample and the top payers. Why has corporate payout ratio increased over time? In this paper, we outline and examine three possible sets of explanations related to changes in firm characteristics, structural shifts in overall payout environment, and the new role of repurchases in total payouts. The first set of explanations focuses on whether changes in firm characteristics help to explain the observed trend in total payouts. Existing cross-sectional evidence suggests that some firm characteristics are major determinants of corporate payout decisions. For instance, Lintner (1956) recognizes that firms pay out dividends in accordance with the level of earnings. Fama and 2 The top payers in our sample are selected based on the amount of total payouts (dividends plus repurchases) each year. 2

4 French (2001) and DeAngelo et al. (2004) find that large, profitable firms pay more dividends than those with more investment opportunities. We examine whether cross-sectional firm-level findings extend to aggregate payout ratios. In particular, we study the time series patterns of retained earnings, profitability, investment and financing policies, and their relations with aggregate payouts. Intuitively, if firms increase payouts over time, it could be because they retain lower earnings or cash flows, are more profitable, or use less capital for other purposes such as investments or acquisitions. Complementing the trend of increasing payouts, firms have lower retained earnings over time. The ratios of retained earnings (RE) to BVE and to assets have declined 32% and 16%, respectively, during our sample period of In the mean time, U.S. firms on aggregate are becoming less profitable. Both the overall sample firms and the top payers experience lower returns on assets over the same period. This indicates that it is firms decisions to pay out more earnings, not higher profitability that drives the time trend of higher payouts. Examining the use of funds by firms provides corroborating evidence. While firms have increased their use of fund in financing major acquisitions, the ratio of capital expenditures to total assets has decreased by 2.6% for the full sample. The change of capital expenditures for top payers is even more significant. The decrease of 4.16% for the high payout firms represents a decline of more than 40 percent during our sample period. Existing theories on corporate financing policy suggest that firms can finance their growth internally (retained earnings) or externally (debt or equity issuances). Evidently, corporate financing choices could have a direct impact on payout decisions. It is possible that if capital markets are more accessible, firms could substitute retained earnings with more external financing, thus increasing cash payout to shareholders. We find some evidence supporting this argument. In the last three decades, firms have increased the use of equity and debt financing. In particular, the leverage ratio of U.S. industrial firms has increased from below 20% to above 25% during the sample period. During the same period, firms have also issued equity more frequently. 3

5 The equity issuances are positively correlated with the aggregate payout ratio, especially with the repurchase ratio. Overall, the more popular use of external financing, in combination with the decline in capital expenditures, contributes to the observed increase in corporate payouts. The second set of explanations focuses on changes in the overall payout environment such as regulations, tax policies, and shifts in investor preference for payouts. We consider two major changes in regulations and tax policies related to corporate payouts during the 1980s. 3 First in 1982, the Securities and Exchange Commission (SEC) adopted Rule 10b-18, which ensures repurchasing firms against the anti-manipulative provisions of the Securities Exchange Act of Because the Rule essentially provides a safe harbor for firms to repurchase shares, there was a considerable growth in repurchase activities after We document that the higher total payouts are in part driven by the increasing share repurchase activities after the adoption of the Rule. Second, the Tax Reform Act of 1986 significantly lowered tax rates on earned income, and particularly eliminated special treatments for capital gains. If the 1986 Act results in a shift in corporate payout policy, we should observe a significant increase in aggregate dividend payments. Our findings, however, reveal that dividend ratio remains stable after the Act and the reduction in income tax rates is not correlated with the subsequent changes in cash dividend. We next examine whether shifts in shareholders preference contribute to the increase in corporate payouts. Existing literature provides two possible explanations linking investors preference with the observed changes in corporate payout policy. One is that the time varying investor demand for payouts could explain the time series patterns of payouts. Baker and Wurgler (2004) and Li and Lie (2006) argue that the decision to pay or increase dividends is positively correlated with the dividend premium, a proxy for investors time-varying demand for dividends. We extend this line of argument by incorporating repurchases in corporate payout decisions and study whether the lagged payout premium is correlated with the payout decisions. 3 Another major tax policy change during our sample period is the Jobs and Growth Tax Relief Reconciliation Act of 2003 which reduces tax rates for both dividend and capital gains. Because our sample period ends in 2003, the impact of this Act on the payout trends is negligible. 4

6 Our findings indicate that the payout premium cannot explain the upward trend of payouts. One could argue that if investors demands for dividends and repurchases offset each other, we may fail to uncover the significance of payout premium. To address this issue, we investigate dividend and repurchase premiums separately. The results provide some support for investor demand for dividends, but not for repurchases. The other possibility is that the shift of investor base has brought broad changes in investor preference and such change could affect corporate payout policy. Over the sample period, institutional investors have grown to dominate the U.S. equity market. Institutional investors could exhibit preferences drastically different from individual investors. For example, institutional investors may prefer high payout firms because of their prudence characteristics (Del Guercio (1996)). Allen, Bernardo, and Welch (2001) argue that firms use different levels of dividends to attract various investor clienteles. Because institutional investors are relatively less taxed than individual investors, payouts induce ownership clientele effects. Our time series results reveal that the rise of institutional ownership is not correlated with the payout trend. This complements the findings in Grinstein and Michaely (2005) that institutional holdings have little impact on corporate future payouts. Our final analysis focuses on the role of share repurchases in the overall increase in corporate payouts. The time series patterns illustrate that payouts through share repurchases account for almost entirely the upward trend in the aggregate payout ratio. Recent studies have documented that firms are increasingly substituting share repurchases for dividends (Grullon and Michaely (2002)). Brav et al. (2005), in their interview with corporate executives on payout policies, find that managers now favor share repurchases because repurchase programs provide more flexibility than dividend payments. How did the wide use of repurchases affect the time trend of corporate payouts? If share repurchases simply substitute for cash dividends without providing any additional benefits to corporations and if we assume that firms have the same propensity to pay, we would probably observe a level payout trend no matter whether firms 5

7 choose dividends or repurchases as the means of payout. However, if the use of share repurchases gives firms incentives to increase their total payouts, then we could find a positive correlation between share repurchases and total payouts. There are two reasons why share repurchases, but not dividends, could offer additional incentives for firms to pay out more cash flows to shareholders. As observed by Brav et al., managers favor share repurchases because of the flexibility that share repurchases have in both timing the equity market and deciding the amount of payout in connection with both investment and financing decisions. The cross-sectional evidence in Jagannathan, Stephens, and Weisbach (2000) also links the growing popularity of share repurchases with the flexibility inherent in repurchase programs. They find that firms are more likely to use share repurchases to pay out transitory cash flows. If firms believe that the flexibility of share repurchases lessens the rigidity of dividend policy, they would prefer repurchases as the form of payout and pay out more cash to shareholders. We present strong time-series evidence that links the increase in share repurchase programs with the flexibility embedded in those programs. The popularity of share repurchases could also be related to the increasing use of option based executive and employee compensation packages. Recent studies argue that firms repurchase shares to fund the exercise of executive and employee stock options (Jolls (1998), Weisbenner (1999), Fenn and Liang (2001), and Kahle (2002)). It is worth noting that if share buybacks reduce cash available for other forms of payouts, the practice of option and stock grants should not affect total payout materially. If, however, stock and option grants to corporate executives better align the interests of managers with interests of shareholders, the adoption of stock and options grants could increase overall payouts. We find scant evidence linking management option exercises with the increase of total payouts. Consistent with existing cross-sectional studies, we find that the time varying payout ratios are positively correlated with executive option values during the period Further analysis, however, reveals that the documented relationship might not be economically 6

8 significant. Specifically, we estimate that the exercised option value accounts for merely 1.1% of raw aggregate payouts in It reaches 3.9% in 2000 and then drops to 1.9% in After netting out the option value, the aggregate payouts still exhibit a significant upward trend. More important, the use of stock options does not increase total payouts in cross-sectional regressions, thus inconsistent with the notion that option-based incentive plans increase total payouts. Our study is closely related to, and extends the results in, two recent papers on dividend policy. Fama and French (2001) document a large decline in both the number and the percentage of industrial firms that pay dividends over the period. DeAngelo et al. (2004) find that firms actually have increased their real dollar value of dividend payouts over the last two decades. They show that the increasing concentration in dividends, along with the increasing concentration in earnings, explain why fewer firms are associated with higher dividend amount. In this paper, we provide strong evidence that U.S. corporations have increased the portion of payouts relative to earnings over time. This trend is mainly due to the fact that firms are increasingly distributing their cash flows to shareholders via repurchases while the aggregate dividends remain stable during the same period. As a result, while many firms do not pay out cash to shareholders either because they do not have positive earnings or choose not to pay, the increasing payouts from profitable firms have shifted the overall payout ratio significantly upward. The remainder of the paper is organized as follows. Section 2 describes our sampling procedure and documents the time trends of aggregate payouts. Section 3 studies the relation between changing firm characteristics and the increase in aggregate corporate payouts. Section 4 examines whether changes in government regulations, tax policies and investor preference contribute to the observed payout trend. Section 5 studies the potential causes of the increasing share repurchases, the main component in the increase of corporate payouts. Section 6 concludes. 2. Sampling Procedure and Trends in Total Payouts 2.1 Time Trend of Aggregate Payout Ratio 7

9 The initial sample consists of all U.S. industrial firms listed on Compustat and CRSP with share codes 10 or 11. Utility (SIC codes 4900 to 4949) and financial firms (6000 to 6999) are excluded. Our sample period starts in 1972 and ends in 2003 since CRSP expands to include NASDAQ firms in To remain in the sample, each firm-year observation must have nonmissing values for dividend (DIV), repurchase (REPO), total assets, and earnings (EARN). We define DIV as the total amount of dividend (in millions, Compustat data #21) declared on the common stock. Following Grullon and Michaely (2002), we define REPO as the expenditure (in millions) on the purchase of common and preferred stocks (#115) minus any reduction in redemption value of preferred stock (#56). To avoid ambiguity, the term payout refers to the sum of cash dividend plus share repurchase throughout this paper. Table 1 and Figure 1 detail the number and percentage of firms paying out cash as dividends or repurchases between 1972 and Consistent with existing studies (e.g., Fama and French (2001), DeAngelo et al. (2004), and Dittmar and Dittmar (2004)), the number of dividend payers has declined dramatically from 1630 in 1972 to 801 in It corresponds to 57% and 22% of all U.S. industrial firms covered on the CRSP/Compustat universe, respectively. In contrast, more corporations have started to employ share repurchases as the form of payout. Only 22% of firms conducted share repurchases in 1972; it increased to 36% at the end of the sample period. In fact, starting from 1995, the number of firms that employ share repurchase as a form of payout surpassed the number of firms that pay dividend. However, even when we consider firms with positive payouts via either dividends or repurchases, the proportion of payout firms still exhibits a significant decline of 19% for the sample period. The percentage increase of repurchasing firms still cannot compensate for the loss of dividend firms. During the same period, fewer firms have positive earnings. The fraction of firms with positive earnings decreases from 91% to 60%. Another measure of firm profitability, average past 5-year earnings, also exhibits a similar downward trend. The percentage of firms with positive 5-year average earnings decreases from 91% to 58%. The decreasing number of both 8

10 payout firms and profitable firms seem to suggest that firms are paying out less cash to shareholders simply because firms on aggregate are less profitable. However, DeAngelo et al. (2004) show that firms actually have increased the real amount of dividend over the sample period and the earnings of dividend paying firms have also increased. The increasing concentration in earnings along with the increasing concentration in dividends helps to explain why fewer firms are paying dividend and at the same time the dollar value of dividend has increased. To examine the time trends in corporate payout policy over the past three decades, we focus on payout ratios. Different from Fama and French (2001) and DeAngelo et al. (2004), who study the number of dividend paying firms and dollar value of dividend payouts, we examine how much firms on aggregate pay out from their earnings. We aggregate total payouts for all firms by year based on both cash dividends and share repurchases and obtain aggregate payout ratios on dividends, repurchases, and payouts using average past 5-year earnings (EARN5), total assets (), and book value of equity (BVE) for all firms as denominators. Because of the volatility in corporate earnings, we use the average past 5-year earnings to calculate the ratio of payout relative to earnings. The use of average past 5-year earnings also reflects the idea that corporate payout policy is more likely to be determined by past profitability than current level of earnings (see, e.g. Grullon et al. (2005)). Given the increase in the number of firms with negative earnings, we further use total assets and BVE as denominators to ensure the robustness of our findings. 4 Panel A of Table 2 and Figure 2 show a significant upward trend in all three aggregate payout measures over the sample period. The results demonstrate that firms on aggregate are in fact paying out more cash to shareholders over time. The ratio of payouts to EARN5 has soared threefold from 46% to 138%. Note that the high payout ratio relative to earnings at the end of the 4 To further ensure the validity of the upward trend of payout ratios, we estimate additional payout ratios such as payouts to sales, payouts to annual earnings, and payouts to cash flows. The results using those payout ratios, available upon request from the authors, are qualitatively similar. 9

11 sample period is partly driven by the large losses experienced by many firms during this period. Even with the average past 5-year earnings, we are not able to control for the impact of negative earnings. The payout ratio based on assets and BVE are less affected by the negative earnings. As shown in the table, the two ratios exhibit very similar upward trends. The ratio of payout to assets has increased almost 1% while that of payout to BVE has increased 3.4%. It should be noted that the 1% change in the ratio of payout to assets from the beginning to the end of the sample period represents a 50% increase in the payout ratio. The change of corporate payout ratio relative to BVE also represents a 50% increase. In Panel B, we formally test whether the trends are significant by using a simple linear trend model. Specifically, we regress the aggregate ratios against time t. For instance, the first regression model is (DIV/EARN5) t = α + βt + ε t where t denotes a time trend for years and ε t is a random error with zero mean. The standard errors in all nine regressions are robust to heteroskedasticity and serial correlation up to three lags using the procedure of Newey and West (1987). As shown in Panel B, all trends on payout ratios are statistically significant at the one percent level. They are also economically significant. In a linear time trend model, the coefficient β captures the long-term average change. Thus, the results indicate that the aggregate payout increases 3.13% per year relative to EARN5, 0.04% to assets, and 0.16% to BVE. This evidence is consistent with the time trend we observe in Panel A. To check the robustness of the documented time trends in corporate payouts, we further calculate the three payout ratios only for firms with positive earnings. For each year, we aggregate payouts for firms with positive earnings and calculate the payout ratios relative to EARN5, assets, and BVE for all firms with positive earnings. The newly computed payout ratios from these firms are free from the impact of negative earnings. This subsample also casts special interest of its own since DeAngelo et al. (2004) document strong evidence that firms with positive earnings are more likely to pay dividends. For brevity, we present the results in 7 periods in 10

12 Panel C. The main difference between Panels C and A is that, after eliminating the effect of negative earnings, the upward trends in the payout ratio relative to earnings are less dramatic. Nonetheless, we confirm the significant upward trend in all three payout ratios. The table further illustrates that the increasing trend in corporate payout is mainly driven by the growing use of share repurchases as the form of payout. At the beginning of the sample period, share repurchase is only a fraction of dividend payment (less than 10%). Starting from 1997, the amount of payout through share repurchase has been larger than dividend payment for the next seven years. This reflects a shift in the relative importance of the two forms of payout. Through the sample period, the dividend ratio has fluctuated over time, but overall it has slightly declined. As shown in Panel A, DIV/ and DIV/BVE decline 0.6% and 0.9%, respectively. While DIV/EARN5 increases 19%, the more robust, counterpart ratio in Panel C shows a 5.5% decrease. 2.2 Concentration of Earnings, Payouts and the Time Trend of Payout Ratio We have documented an increasing trend of aggregate payouts in the last three decades. In this subsection, we examine whether the concentrations of corporate earnings and payouts are correlated. DeAngelo et al. (2004) find that corporate earnings and dividend payments are increasingly concentrated in a small number of firms. The two-tier structure in corporate earnings and payouts could generate different patterns in payout policies. We first extend the study of DeAngelo et al. by including repurchases in measuring total payouts and examine the concentration of total payouts. We then examine the time trends of the payout policies for firms with different payout levels. We find similar patterns of increasing payout concentration after including share repurchases. We first document the payout concentration in two years: 1978 and We then 5 To compare with previous studies, we contrast payout concentration in 1978 with that in 2003, the last year in our sample. 11

13 calculate the concentration for the entire sample period. There were 2,386 payout firms in 1978; the number dropped to 1,659 in To find top payers, we sort all payout firms in those two years into ten groups of 100 firms each based on total payout amounts. The top 100 payout firms are placed in the first group and the 101 to 200 firms in the second and so on. The eleventh group, also the last, consists of the remaining firms. The first two columns of Table 3 reports the number of payout firms in each group in 1978 and Clearly, most firms used dividends as the form of payout in 1978, but repurchases become an important choice in This change is evident in the top group. In 1978, only 34 of the top 100 payers repurchased shares and only one firm made the top 100 payer list through share repurchase alone. In contrast, in 2003, 82 of the top 100 payers repurchased shares. In the top 100 payers, 19 made the list through share repurchase alone and 18 made the list through dividend payments alone. The other top groups show similar patterns. The next six columns of the table show rising concentrations of both dividends and repurchases, and consequently total payouts. The top 100 payers distributed 65% of all dividends in 1978 and almost 80% in Those top payers distributed 76% of all repurchases in 2003, up from 55% in In total, the top 100 payers contribute about 64% of total payout in 1978 and 77% in If we consider the top 500 payers, these firms account for 90% of total payout in The number went up to 97% in These figures reveal that, similar to the concentration of dividend payouts, corporate payouts through share repurchases also exhibit high and increasing concentration. Thus the pattern of increasing concentration of total payouts parallels that of dividend payments documented in DeAngelo et al. (2004). We further verify that the difference in the payout concentration in years 1978 and 2003 also reflects the overall time trend of payout concentration. For the sample period of , we use the Herfindahl index to capture the change in payout concentration. The Herfindahl index of dividends is calculated by first aggregating dividends within each group i where i = 1, 11. We then compute the index as H(DIV)=Σ i (D i *D i )/(Σ i D i ) 2 where D i is the aggregate dividends in each 12

14 group. The Herfindahl indices of repurchases and total payouts, H(REPO) and H(Payout), are calculated similarly. Figure 3 shows upward trends of payout concentrations over time. H(DIV) increases from 49% to 64% during the sample period. The change of H(REPO), from 33% to 59%, is even more dramatic. H(Payout) also shows an upward movement from 47% to 61% although it has a surge during mid-1980s. The concentration of payouts is accompanied by the concentration of earnings. The last two columns of Table 3 display the percentage and cumulative percentage of total earnings from top payout firms. The top 100 payout firms generated 56% of the earnings from all payout firms in 1978; the figure increased to 73% in As mentioned above, the top 500 payers accounted for 90% of total payouts in 1978 and 97% in They also cumulatively generated 86% of total earnings in 1978 and 94% in The findings demonstrate that payouts and earnings are concentrated in top payers in 1978, and they are more so in Given the significant growth of share repurchase as the form of payout over the sample period, the results also indicate that the top payers are increasingly dispensing earnings through share repurchases. We present, in more detail, the time trends of payouts for three different tiers of payout firms in Table 4: Top 100, 101 to 500, and 501 to 1000 payers. Payout firms are ranked annually by the total amount of dividends plus repurchases as in Table 3. Panel A reports aggregate payout ratios for top 100 payers, Panel B for top 101 to 500 payers, and Panel C for 501 to 1000 payers. Here, we focus on the payout policies of the top 500 payers, given their dominance in aggregate payouts. Again, the results are presented in seven time periods. Overall, the increase of payouts is mostly driven by the top 500 payers. In particular, the payout ratios of the top 100 payers show the most significant increase. The payout to EARN5 ratio almost doubled over the time period for the top payers. It changes from 52% to 103%. Notice that because the aggregate earnings of the top payers are less affected by negative earnings, the trend of the payout ratio based on earnings is quite similar to those based on assets and BVE. In addition, the increase of the payout ratio for the top 100 payers is much higher than 13

15 the increase in the full sample (see Table 2). Such an increase is particularly remarkable given that the top payers already have a much higher payout ratio at the beginning of the period. The payout ratios of the remaining 400 top payers exhibit patterns that are similar to, but slightly weaker than those of the top 100 payers. Interestingly, the other 500 payers do not exhibit strong increasing trends in their payouts. Both payout ratios based on assets and BVE in fact show a slight decline because of the considerable drop in the last period. Table 4 further suggests that the surge of payout in top 500 payers is mainly driven by the increasing popularity of share repurchase programs. Based on the payout ratios, the amount of total payout through repurchases became larger than the amount of dividend payments at the end of the period. For example, the dividend ratio, DIV/, is 2.33%, and the ratio of REPO/ is 2.98% for the top 100 payers. The corresponding two ratios for the 101 to 500 payers are 1.01% and 1.39%, respectively. The upward trend in payout ratios does not extend to dividend ratios, even for the top payers. Nevertheless, the dividend ratios of the top payers seem quite stable except for the last period, Time Trends in Total Payouts and Changes in Firm Characteristics 3.1 Firm Characteristics We now investigate whether firms preference toward higher payouts is related to changing firm characteristics such as profitability, investment opportunities, and financing policies. We focus on whether the increasing level of payouts (outflow of funds) can be explained by how firms manage their funds including retention, usage, and replenishment. Table 5 presents summary statistics of those variables. Panel A presents results from the full sample while Panel B is from the subsamples based on payout rankings. For each variable, we first calculate the aggregate value in each year, and then normalize the value by aggregated total assets or book value of equity. Again, for brevity, we present the figures in seven periods. 14

16 We first examine whether the observed trends in payout policies can be explained by how corporations retain their capital. Columns (1) and (2) show that firms have retained less earnings relative to book value of equity or total assets, a result complement to our finding that corporate payout ratio has increased over time. The RE/BVE (RE/) ratio was 67% (28%) during , peaked at 73% (31%) in , and then declined to 30% (11%) in the last period, The change from 1972 to 2003 is a dramatic -32%. Corporate cash level in column (3) is somewhat inconclusive with a positive, yet small, change of 1% from 1972 to Corporate cash holdings initially declined significantly, remained relatively stable from the mid 80 s to 2000, and increased in the last period. Overall, the pattern in cash holdings seems to suggest that the time trend of payout policies is not correlated with cash retention policies. Next, we examine the possibility that changes in corporate profitability can explain the observed trend in aggregate payout ratio. Several papers find a positive relation between profitability and dividend policy (see, e.g., Fama and French (2001) and DeAngelo, DeAngelo, and Stulz (2006)). They attribute the disappearing dividends to lower profitability especially for newly listed firms. We use EBITDA and free cash flows 6 to measure the level of corporate profitability. Column 9 indicates that corporate profitability has diminished from 14% to 12%. The ratio of free cash flows to assets in column (10) also depicts a similar picture, with the highest level (11%) in the second period and decreasing to 6% in the last period. Because we observe a significant increase in aggregate payouts, it is evident that change in corporate profitability cannot explain the upward trend of aggregate payouts. Columns 4 to 6 present time series patterns of corporate use of funds in investment opportunities and major acquisitions. Presumably, firms with more investments are less likely to distribute cash flows to equity holders. Here, we investigate whether the increase of payouts can be attributed to declining growth opportunities. The results indicate that on aggregate firms have 6 Free cash flows are net income to common plus depreciation and amortization. We normalize free cash flows with total assets. 15

17 reduced capital expenditures over time. Column 4 shows that firms have reduced the ratio of capital expenditures to total assets from 7.8% to 5.3% during our sample period. The time trend of aggregate corporate capital expenditures is noteworthy in the magnitude of the decline. The negative change of 2.6% represents about 33% decrease from the original level. The decline of R&D expenses, however, is not present. Column 5 shows that the ratio of R&D expenses to total assets has increased almost twofold from 1.2% to 2.2%. Overall the R&D expenses have remained stable from the later 80s onward. We further examine another form of use of funds: cash outflows in cash-financed mergers and acquisitions. Bagwell and Shoven (1989) argue that even though cash-financed mergers and acquisitions do not go to a firm s existing shareholders, they represent one major form of corporate cash outflows. We treat cash payout through acquisitions as a form of fund use because such distributions are similar to capital expenditures and related to firm s growth opportunities. Column 6 shows that cash distributions through mergers and acquisitions have generally increased over time. The increase of this channel of cash outflow is not stable, however. Such time trend could possibly be highly correlated with various merger waves. For example, the amount used in acquisitions was much higher in the later 80s and later 90s than in other periods. Clearly, the cash outlay through acquisitions can not explain the observed upward trend in corporate payouts. In fact, if we follow Bagwell and Shoven (1989) and classify cash distributions in acquisitions as a form of payouts, our results would have indicated an even more dramatic increase in total payouts. So far we have documented several stylized facts that firms have increased aggregate payouts and acquisitions, but retained less earnings or cash flows. In particular, the increase of corporate payouts is not financed by rising corporate profits. To the contrary, both corporate profitability and cash flows available for payout have actually declined over the sample period. As the results on the use of funds indicate, falling capital expenditures is the only consistent explanation. 16

18 We now examine the time trend of corporate financing activities and whether those activities help to explain the time trend in corporate payout policies. It is possible that the financing flexibility and easiness might have increased over time such that firms are not required to maintain a high level of internal funds for future investment opportunities. Here, we use the aggregate equity and debt issues as indications for easiness of financing. If it is easier for firms to raise external capital, they are more likely to pay out cash flows and reserve less for future investments. 7 Columns 7 and 8 in Table 5 suggest that both the two external financing measures exhibit upward trends. The ratio of aggregate seasoned equity issuance relative to assets (SEO/assets) increases from 0.56% to 1.29% during the sample period. Firms have constantly raised more capital over time. More remarkably, the aggregate debt ratio of industrial firms (Debt/) has risen from 20% to 26%. The increase of corporate leverage has been steady over the last three decades. 3.2 Changing Characteristics of Top Payers We now examine the time trends of firm characteristics of the top payers. Fama and French (2001), DeAngelo et al. (2004), and others have shown that payout firms exhibit distinctive characteristics such as profitability, growth opportunity than those of non-payers. Here, we are more concerned about the time trend of those firm characteristics, and particularly whether the time series patterns we observed in the full sample apply to the top payers. Given the concentration of corporate earnings and payouts, it is possible that the top payers exhibit both different characteristics and time trends in the changes of those characteristics. We report the results based on the three levels of payers in Panel B. As it turns out, the patterns of changing characteristics of top payers are similar to those of the full sample. For 7 A significant relation between debt and equity issuances and the increase of payouts also supports the agency-cost explanations of payouts as in Easterbrook (1984). Easterbrook argues that firms raise new capital and pay out cash flows via dividends or repurchases almost simultaneously to reduce the monitoring problem between management and shareholders. 17

19 brevity, we focus our discussions on the variables that have time series patterns different from those in the full sample. The top 100 payers have very high ratios of retained earnings relative to equity or assets, consistent with the life cycle argument of DeAngelo et al. (2005). The time trend of the retained earnings ratios, however, is less conclusive. The ratio of retained earnings to assets on average declined, but the ratio of RE/BVE has generally increased. As expected, the top 100 payers are more profitable than the remaining payers even though their profitability has declined. The most striking results from the top 100 payers are the decline in capital expenditures. The CapEx/ ratio in these highly profitable firms decreases by more than 4%, much steeper than the 2.5% in the full sample. The financing activities of the top payers show a less dramatic trend than those of the full sample. Equity issues exhibit an insignificant negative change of 0.09%. The remaining top 400 payers and the other 500 payers exhibit similar characteristics except for the profitability. The RE ratios of both groups have declined significantly. Both groups of payers disperse large amount of cash through acquisitions, but their overall capital expenditures do not show significant declines. The most striking trend from these two groups of payers is the increase of debt ratios. The equity issues exhibit a higher positive change of 0.49% in the top 101 to 500 payers than the full sample. 3.3 Relation between Corporate Payout and Changing Firm Characteristics We apply time series analysis to test whether the increase of aggregate payouts is systematically related to firm characteristics and other variables. We run separate regressions for our variables of interest along with two control variables: time trend and past market returns. Previous articles have shown that corporate dividend and repurchase policies are correlated with past market returns (Ikenberry, Lakonishok, and Vermaelen (1995), Dittmar and Dittmar (2004), and Baker and Wurgler (2004)). In addition, to avoid potential causality issues, all explanatory variables, except for regulation and tax, are from previous year. We report p-values associated with t-statistics which are adjusted for heteroskedasticity and serial correlation (up to three lags) 18

20 using the Newey-West (1987) procedure. Table 6 presents results for total payouts (Panel A), dividends (Panel B), and repurchases (Panel C). 8 Model 1 relates the aggregate payout ratio to cash outflows in capital expenditures, R&D, and acquisitions while model 2 studies the relation between payouts and profitability. We find that lagged capital expenditures are negatively associated with total payouts, confirming the univariate result. In addition, models 6 and 11 indicate that the negative correlation is largely driven through share repurchases. The positive relations of R&D and acquisitions with payout ratios also confirm the earlier findings that the two variables do not explain the observed increasing time trend in corporate payout. Finally, firm profitability (as shown in models 2, 7, and 12) is not related to corporate payouts during the sample period. Notice that some firm characteristics may be more important at the firm level than at the aggregate level such that crosssectional results documented in previous studies do not hold in our time-series settings. Specification 3 examines the time series relation between corporate financial and payout policies. Overall, there is little statistical relation between the debt ratio and total payouts, inconsistent with the earlier observation of negative relation (Table 5). In contrast, results on equity issuances confirm our earlier evidence. Model 13 further shows a strong time series association between aggregate repurchase ratios and equity issuances. This indicates that the flexibility of corporate financing is positively correlated with firms propensity to payout cash flows to shareholders via repurchases. 9 Furthermore, Table 6 reveals that the dividend and repurchase ratios exhibit quite different relations with firm characteristics. Although both forms of payout serve the same function of distributing excess cash flows to shareholders, they are different in terms of flexibility. Dividends are known to be sticky : Lintner (1956) presented evidence that firms are 8 For brevity, we only report our findings using Payout/EARN5, DIV/EARN5, and REPO/EARN5 in Table 6. Results using different ratios, available from the authors upon request, are qualitatively similar. 9 The positive relation between payouts and equity issues uncovered here also support the agency-cost explanations of payouts in Easterbrook (1984). 19

21 reluctant to cut dividends once they begin. Repurchases, in contrast, provide more flexibility. Jagannathan, Stephens, and Weisbach (2000) link the growing popularity of share repurchases with the flexibility inherent in repurchase programs. Thus, if firms believe that financing opportunities vary over time, they would prefer repurchases as the choice of payout. This could also explain why equity issues are positively correlated with repurchases but not with dividends. 4. Time Trends in Total Payouts and Changing Payout Environment In this section, we study the relation between changes in the overall payout environment and the trend of corporate payouts. Changes in government regulation and tax policies as well as the overall shifts in investor preference could have significant impact on corporate payout polices that are not captured by firm characteristics. 4.1 Changes in Government Regulation and Tax Policies We consider two major changes in government regulation and tax polices that could affect the aggregate corporate payouts. First, the Tax Reform Act of 1986 is one of the most important tax legislations during the sample period. The 1986 Act significantly lowered tax rate on personal income, and particularly eliminated special treatment for capital gains. Figure 4 plots the maximum marginal tax rates for personal income and capital gains. Personal income tax rate decreased significantly and the difference of tax rates on dividends and capital gains dropped dramatically after These changes should have more dramatic impact on dividend payments since dividends are less disadvantaged than repurchases. Our univariate findings in Table 2, however, indicate that the reduction in income tax rate may not lead to increases in dividends. Dividend ratios instead have remained stable even after the passage of the Tax Reform Act. In contrast to dividend ratios, cash payout through share repurchase has increased since the 1980s even though the relative tax advantage of capital gains vs. dividend income has become smaller. The higher share repurchase activity is more likely as a result of regulation changes on share repurchases. In 1982, the Securities and Exchange Commission (SEC) adopted Rule 10b- 20

22 18, which ensures repurchasing firms against the anti-manipulative provisions of the Securities Exchange Act of Because Rule 10b-18 provides a safe harbor for firms to repurchase shares, we expect to observe a significant increase in repurchase activities after We formally test the effect of the two tax changes on the time trend of payout. In the time series models, we add two variables to examine whether the shifts in government tax policies affect corporate payout policies. One is the difference of the maximum marginal tax rates on dividends and capital gains ( Tax Difference ) and the other is a Rule 10b-18 dummy which equals one for years 1983 and after. As shown in Table 6, the increase in aggregate payouts is not related to the favorable tax treatment in dividends after the 1986 Tax Reform Act. The coefficient of tax variables is insignificant. Interestingly, model 10 shows that the favorable treatment in dividends is associated with dividend decreases, an opposite relation predicted by the hypothesis. Thus, it is evident that the 1986 Act does not provide sufficient incentives for firms to increase payouts. The Rule 10b-18, in contrast, is responsible for the increase in aggregate payout ratios. On average, the ratio of payouts to EARN5 increases 18.9% after the adoption of the Rule (model 5). An undocumented result indicates that the ratios of payouts to assets and to book value of equity also increase 1.0% and 2.7%, respectively. Interestingly, models 10 and 15 further reveal that Rule 10b-18 has a positive impact on share repurchases, but it does not cause firms to lower dividends. 4.2 Changes in Investor Preference We next examine whether changes in the preference of shareholders contribute to the increase of corporate payouts. Existing literature provides two possible explanations related to investor preferences. One is that the shift of investor base has caused broad changes in investor preference and such change could affect corporate payout policy while the other is that the time varying investor demand for payouts could explain the increase of payouts. We investigate the first reason first. 21

23 Institutional investors have grown to dominate the U.S. equity market over the last three decades. Institutional investors could have preferences drastically different from individual investors. For example, institutional investors may prefer high payout firms because of their prudence characteristics (Del Guercio (1996)). Also, institutional investors have long been considered better informed than individual investors and have more incentives to monitor management in a firm. Jensen s (1986) free cash flow hypothesis implies that larger institutional holdings will have a positive effect on payouts. Several recent papers also suggest a significant interaction between institutional ownership and corporate dividend policy (see, e.g., Brennan and Thakor (1990) and Allen, Bernardo, and Welch (2000)). Despite the theoretical implications, in a panel study, Grinstein and Michaely (2005) find that larger institutional ownership does not lead to higher payouts. Here we study whether the increasing trend of aggregate payouts coincides with the time series trend of institutional holdings. As shown in model 4 of Table 6, the lagged institutional ownership is not significantly correlated with the aggregate payout ratios, a result consistent with the findings from Grinstein and Michaely. Higher institutional holdings do not cause firms to increase payouts. Models 9 and 14 further demonstrate that institutional holdings are not related to dividend or repurchase patterns. We next investigate the time varying investor demand for payouts. Baker and Wurgler (2004) and Li and Lie (2006) argue that the decision to pay or increase dividends is positively correlated with the dividend premium, a proxy for investors time-varying demand for dividends. They find that firms cater to investor demand: initiate or increase dividends when the stock price of payers is high and omit dividends when the price is low. We extend this line of argument by incorporating repurchases in corporate payout decisions. According to the catering theory, when investors demand for payouts increases, firms are more likely to increase payouts (via either dividends or repurchases). The observed increase in aggregate payout could be a result of shifting investor preference towards corporate payouts. 22