Does Ownership Structure Affect Corporations’
Responses to Lower Dividend Tax Rates?
An Analysis of Public and Private Banks
C. Bryan Cloyd Virginia Tech (540) 231-3181 bcloyd@vt.edu
John R. Robinson
The University of Texas at Austin (512) 471-5315
John.Robinson@mccombs.utexas.edu
and
Connie D. Weaver
The University of Texas at Austin (512) 471-5318
Connie.Weaver@mccombs.utexas.edu
August 18, 2005
_______________
The authors gratefully acknowledge research support provided by their respective institutions: Cloyd (John E. Peterson, Jr. Professorship), Robinson (Red McCombs School of Business and C. Aubrey Smith Professorship) and Weaver (Deloitte & Touche Centennial Faculty Fellowship). The authors also
appreciate the constructive comments of two anonymous referees, Jennifer Blouin, Gary Twite and finance workshop participants at the University of Texas at Austin and ATAX/University of New South Wales.
Does Ownership Structure Affect Corporations’ Responses to Lower
Dividend Tax Rates? An Analysis of Public and Private Banks
ABSTRACT
The Jobs and Growth Tax Relief Reconciliation Act of 2003 (JGTRRA) significantly reduces the maximum tax rate imposed on dividend income received in 2003 through 2008 by individual shareholders. Using data from call reports filed with the Federal Reserve Board, we investigate whether federally regulated bank holding companies (BHCs) increased dividend distributions after JGTRRA. By focusing on BHCs, we are able to compare the responses to this important but temporary change in tax policy across publicly-traded and privately-held corporations, while controlling for industry-related factors. Examining private firms’ dividend payouts and reactions to JGTRRA allows us to estimate the effect of differential taxation on dividend policy in a setting that is relatively free of the information asymmetry and agency problems that may dominate the payout decisions of public firms. We find that dividend yield (i.e. dividend distributions deflated by assets) increased after the reduction of the dividend tax rate for both private and public BHCs. However, private BHCs increased dividend payouts to a greater extent than public BHCs. Our comparison of public and private payouts also corroborates evidence that public firms pay dividends to mitigate agency conflicts and this effect may act to dampen the public-firm response to the dividend tax reduction in JGTRRA.
Keywords: Dividend policy; Dividend taxes; Closely-held corporations; Bank holding
companies.
Does Ownership Structure Affect Corporations’ Responses to Lower
Dividend Tax Rates? An Analysis of Public and Private Banks
I. INTRODUCTION
The relation between dividend policy and taxation has puzzled financial economists for decades (Stiglitz 1973; Black 1976). For much of the history of income taxation in the U.S., individuals have faced higher tax rates on dividend income than on long-term capital gains. Hence, a policy of making dividend distributions to individual shareholders is sub-optimal compared to a policy of repurchasing shares or reinvesting profits. Yet, despite the tax disadvantages of paying dividends, many firms make regular dividend distributions in the apparent belief that retail investors prefer dividend paying stocks (Brav et al. 2004). We
investigate the relation between dividend policy and taxation by comparing changes in dividend payouts for a sample of public (publicly-traded) firms to those for a sample of private (privately-held) corporations around the 2003 reduction in dividend taxes. A comparison of the dividend response for private versus public firms around a reduction in dividend taxes provides a unique opportunity to test the extent to which (1) individual taxation discourages corporations from paying dividends and (2) dividend policy is influenced by nontax considerations.
A battered stock market set the stage for proposals to reduce dividend taxation, beginning with President George W. Bush’s economic summit in August 2002 (Esenwein and Gravelle 2003). In late May 2003, Congress passed and President Bush signed the Jobs and Growth Tax Relief Reconciliation Act of 2003 (JGTRRA), which significantly lowered the maximum individual tax rate on dividend income (from 38.1 percent) and equalized it with the maximum
tax rate on capital gains at 15 percent.1 Among the rationales offered by legislators and the Bush administration in support of dividend tax relief were that it would (1) increase the value of the stock market (Davis 2003), (2) mitigate corporate governance problems (Bush 2003)2, and (3) lead to “economic growth and the creation of jobs” by reducing the cost of capital for
corporations (H.R. Rep. No. 108-94, ¶10,110). The first two of these objectives and much of the political rhetoric about lowering the dividend tax rate pertain exclusively to publicly-traded firms.3
Several concurrent studies investigate the effect of JGTRRA’s dividend tax rate cut on dividend initiations and payouts by publicly-traded firms. Using a sample of utility, non-financial firms traded on the three major stock exchanges, Chetty and Saez (2004) find that the probabilities of dividend initiations and dividend increases during the first four quarters
following the tax cut (through 2004 Q2) are greater for firms with low-growth forecasts and for firms whose major shareholders and/or managers also had incentives to increase dividends. Using a sample of publicly-traded firms that declared dividends during the first six months following passage of JGTRRA, Blouin, Ready and Shackelford (2004) find that quarterly dividends per share increased after the tax rate cut. Although they do not find a significant
1 This legislation (P.L. 108-27) decoupled the tax rate for dividends from that on other ordinary income, such as interest and wages, by reducing the tax rate for dividends paid by a domestic or qualified foreign corporation to individuals. The maximum 15 percent rate is temporary, applying to qualified dividends received in 2003 through 2008. For lower-income individuals, a five-percent rate applies to dividend income through 2007, and in 2008 dividend income is tax-free. The Joint Committee on Taxation estimates that this legislation will increase the federal deficit by $350 billion over a ten year period.
2 In his speech given at the signing of JGTRRA, President Bush stated “This will encourage more companies to pay dividends, which, in itself, will not only be good for investors, but will be a corporate reform measure.” His
rationale is similar to that of Prudential Securities economist Ed Yardeni, who Bartlett (2002) paraphrases as arguing that, because “dividends are a better indicator of profitability than easily-manipulated earnings statements,” reducing dividend taxation would solve “a lot of corporate-governance problems” by increasing dividend payments.
3 The third objective of economic growth and job creation pertains less to non-publicly-held ventures because newly formed businesses can organize as limited liability companies or S corporations, thereby avoiding double-taxation entirely.
relation between responses to the tax cut and the extent to which shares are owned by individual investors, they do find some evidence that dividend increases were greater for firms with high levels of insider ownership (i.e. shares owned by officers and directors). Similarly, Brown, Liang and Weisbenner (2005) find that the probability of a dividend increase by publicly-traded firms following enactment of JGTRRA is increasing in direct stock ownership by firms’ top five executives, and decreasing in the number of stock options held by these individuals.4
Collectively, these studies suggest that public firms’ responses to the dividend tax cut depend on the extent to which their top managers’ economic interests are aligned with those of
shareholders. On the other hand, Julio and Ikenberry (2004) suggest that the upward trend in dividend payouts by public firms began at least two years before the tax cut and may have been driven by changing economic conditions and a desire to restore investor confidence in the aftermath of corporate accounting scandals.5
We contribute to this literature by comparing the pre- and post-JGTRRA dividend payouts of publicly-traded versus privately-held bank holding companies (BHCs). To facilitate this comparison, we use data from call reports filed with the Federal Reserve Board to construct a panel of 29 consecutive quarterly observations, from 1998 Q1 to 2005 Q1, for 323 public and 216 private BHCs organized as taxable C corporations.6 Most prior studies of dividend policy exclude BHCs because, unlike other industries, bank holding companies are monitored by bank regulators and must meet minimum capital requirements that may constrain dividend payouts. In
4 Because dividends do not accrue to option holders and they reduce the retained value of firm assets, managers who hold stock options have a disincentive to pay dividends (Lambert, Lanen and Larcker 1989).
5 Blouin et al. (2004) acknowledge that they cannot rule out the possibility that dividend increases reflect improving economic conditions, which occurred during the same period as the tax cut, or a shift in investor focus toward valuing firms based on dividends rather than earnings.
6 The Federal Reserve Board requires both private and public bank holding companies to report financial data and distribution information each calendar quarter, which provides comparable data across firm ownership type while holding industry constant.
contrast to other studies that combine firms across unregulated industries, our research design holds industry effects constant by focusing exclusively on BHCs. It is important to understand the effect of shareholder taxes on BHCs’ payouts because this industry sector accounts for a substantial portion of total dividends paid by public firms. For example, in 2002, the year prior to JGTRRA, the public bank sector (2-digit SIC = 60) paid a total of $61 billion in dividends, more than any other 2-digit SIC group. Moreover, the bank sector represents 12.8 percent of the total market capitalization of all dividend-paying firms in 2002 and accounts for 19 percent of all dividends paid by public firms in that year.7
More importantly, focusing on BHCs allows us to examine whether public and private corporations differ in their responses to the dividend tax rate reduction of JGTRRA. Despite the political rhetoric about lowering the dividend tax rate being centered on public firms, there are several reasons why we expect the reaction to be greater for private firms. First, ownership of private BHCs is more likely to be concentrated among high-tax rate individuals who share tax incentives to avoid dividend payouts prior to JGTRRA. Second, because owners of private BHCs are more likely to be involved in managing the firm, non-tax explanations for paying dividends (e.g. signaling and limited overinvestment by managers) are less applicable to these firms. In sum, private BHCs have more tax reasons to retain earnings and few non-tax reasons to distribute earnings than their public counterparts. Third, private BHCs are less able to use tax-advantaged share repurchases as dividend substitutes because of tax rules that treat share redemptions as dividends (unless the transaction causes a significant reduction in the
shareholder’s proportionate interest in the firm). If maintaining control within a relatively small
7 These statistics are based on an analysis of all dividend-paying firms (n = 2,365) in the Compustat database for 2002. These dividend-paying firms represented 68 separate 2-digit SIC codes with chemicals, utilities and REITs representing the next largest dividend paying sectors.
group of shareholders is important or if shares of private BHCs lack a ready market, then shareholders of these firms are likely to have pent-up demand for liquidity that might be
unleashed by a temporary cut in the dividend tax rate. Consequently, relative to public BHCs, we expect private BHCs to have lower dividend payouts before JGTRRA and to increase dividend payouts to a greater extent as a result of JGTRRA.
Finally, any differential effect of JGTRRA across these two groups is not attributable to changing economic conditions, which are held constant across groups. Nor is a greater response to JGTRRA by private firms attributable to market-driven increases in investors’ preferences for dividends because such market preferences should be largely irrelevant to private firms. Market preferences are not, however, irrelevant to the public firms in our sample. Allen and Michaely (2003) and others conjecture that dividends are paid by public firms to mitigate agency conflicts between managers and shareholders. DeAngelo, DeAngelo, and Stulz (2004) conjecture that agency conflicts become more severe as profits accumulate, and they present evidence that distributions increase in the ratio of retained earnings to total assets. We test their conjecture directly by comparing whether distributions by both public and private firms are positively associated with retained earnings.
Our initial analysis reveals that the magnitude and frequency of dividends over our sample period is seasonal across calendar quarters and generally increasing across years for both public and private firms. To control for these trends as well as financial and regulatory
considerations, we estimate random effects regressions of annual dividend yield on tax regime (an indicator variable identifying periods after the structural shift in tax rates enacted in
and maturity).8 Our regression results indicate that both private and public BHCs significantly increased dividend distributions after the enactment of JGTRRA. We estimate that the average private BHC in our sample increased dividends by 0.096% of total assets during the first four quarters following enactment (i.e. 2003 Q3 through 2004 Q2), while the average public BHC in our sample increased dividends by only 0.026% of total assets. Although both increases
represent a significant tax-induced departure from their pre-JGTRRA payout policies, the average response by private BHCs was significantly greater than the average response by public BHCs. In addition, our regression results indicate that distributions from private BHCs are unrelated to the level of retained earnings. In contrast, we find that distributions from public BHCs are positively related to the level of retained earnings. This finding supports the results reported by DeAngelo, DeAngelo, and Stulz (2004) and suggests that distributions related to agency conflicts may be one explanation for the muted response by public firms to the dividend tax reduction in JGTRRA.
The remainder of this study is organized as follows. The next section summarizes the related literature and develops the hypotheses. The third section presents the research method including a description of the sample. The fourth and fifth sections report our results and supplemental analyses, respectively. The final section concludes.
II. PRIOR LITERATURE AND HYPOTHESES DEVELOPMENT
A substantial body of prior research seeks to identify factors that explain cross-sectional differences in firms’ dividend policies. Allen and Michaely (2003) provide an extensive review of this literature, concluding that the accumulated evidence, based on data from publicly-traded
8 Because our sample includes private BHCs that are not publicly-traded, we calculate dividend yield by deflating distributions by gross assets.
firms, suggests that both dividends and repurchases are paid to reduce potential overinvestment by managers. DeAngelo, DeAngelo, and Stulz (2004) draw a similar conclusion based on their finding that the likelihood of paying dividends is positively related to the ratio of earned equity to total assets. Others argue that managers use dividends to convey private information about firm value to the market (Kalay and Loewenstein 1986; Michaely, Thaler and Womack 1995; Ofer and Siegel 1987; Nissim and Ziv 2001). Fama and French (2001) examine an extended time period (1926-1999) to identify the characteristics of dividend payers. They find that the
likelihood of dividend distributions for industrial firms is related to size, profitability, and growth.
DeAngelo and DeAngelo (2004) argue that firm value requires a genuine prospect of substantial payouts because rational investors will not pay more for shares than the expected net present value of future distributions.9 However, if dividends are taxed more harshly than long-term capital gains, then tax sensitive investors will not be indifferent to the form of payout. Consequently, much of the payout research focuses on the effect of individual taxes on corporate dividend policy, in large part because of the belief that the high tax cost of dividends relative to capital gains serves as an impediment to distributing earnings via dividends versus alternative payout mechanisms. For example, corporations can reduce shareholder taxes on distributions by
9 DeAngelo and DeAngelo’s (2004) primary purpose is to correct misinterpretations of Miller and Modigliani’s (1961) dividend irrelevance theorem. Miller and Modigliani (1961) analytically demonstrate that, in a perfect capital market with no tax rate differential between dividends and capital gains, and with fixed investment, financing, and operating policies, a firm’s dividend policy is irrelevant to firm value. The assumptions of this model are, however, very restrictive. For example, DeAngelo and DeAngelo point out that, by holding investment policy constant, the firm’s choice set is restricted to dividend policies that pay out 100 percent or more of free cash flow in every period, where distributions in excess of free cash flow are financed by fairly priced sales of additional stock. Furthermore, they show that, if the firm reinvests free cash flow on a value-neutral basis, indifference holds only if the present value of expected distributions is equal to that of the full stream of free cash flows.
substituting stock repurchases for dividends.10 Repurchases generally result in lower-taxed capital gains and allow investors to control the timing of income recognition. Although Feldstein and Green (1983) argue that this form of tax avoidance is limited, the extent to which taxes motivate corporations to substitute share repurchases for dividend distributions remains an open question.
An alternative to testing the influence of taxes by analyzing dividends and share repurchases is to examine the effect of changes in the relative tax rates between dividends and capital gains. Although the difference between these tax rates varies over time as the result of frequent tax legislation, the rate change enacted by JGTRRA was both dramatic and relatively uncontaminated by other changes in the tax law (Blouin et al. 2004). There are several reasons, however, that even a dramatic change in the relative level of taxes on dividends might not impact dividend distributions. First, managers should be reluctant to modify dividend distributions in response to a tax rate change because dividends impose immediate taxes on all individual shareholders whereas capital gains taxes are only imposed on shareholders who sell shares. Second, a shift in dividend payout is likely to be viewed by investors as a permanent commitment to making distributions. Hence, an increase in distributions in response to a (presumably) temporary reduction in taxes might be mistaken by investors as a permanent shift in dividend policy. Finally, taxes would seem to be secondary to financial considerations, such as liquidity and profitability, as well as regulatory concerns. 11
10 Grullon and Michaely (2002) examine the substitutability of repurchases and dividends and find that dividend-paying firms use repurchases as a substitute but that the rate of substitution is less than one.
11Likewise, private BHCs might not increase dividend payouts in response to JGTRRA if, as an alternative, they can distribute to owners in ways that are deductible at the corporate level. Owners of private BHCs have alternative mechanisms for receiving their returns (e.g., deductible salaries or rents) to the extent that (1) they actually provide services or use of property in proportion to share ownership and (2) the amounts paid out in this form have not already reached some maximum level of reasonableness. If either constraint is binding, then there are limited
Several concurrent studies (Blouin et al. 2004; Chetty and Saez 2004; Julio and Ikenberry 2004; and Brown, Liang and Weisbenner 2005) investigate the extent to which public
corporations increased dividend distributions after JGTRRA. Taken as a group, the results of these studies suggest that publicly-traded firms responded to the dividend tax cut by initiating or increasing dividends, although Julio and Ikenberry (2004) report that dividend distributions for industrial firms were trending upward well before the dividend tax cut in 2003. These studies also suggest that cross-sectional variation in firms’ responses to the dividend tax cut is at least partially explained by their economic consequences to top executives. Brav et al.’s (2004) survey of financial executives indicates that payout decisions typically originate with the CFO or
Treasurer, and are approved by the CEO before being considered by the Board of Directors. Hence, it is not surprising that managers’ personal outcomes affect payout decisions.
We conjecture that, ceteris paribus, dividend taxes operate analogous to transaction costs, thereby reducing the number and magnitude of distributions during periods of high dividend tax rates. If our expectations hold, we anticipate that the likelihood and amount of dividend
distributions for public and private BHCs will increase after the tax rate reduction enacted in JGTRRA. We expect that nontax factors (e.g., reductions in liquidity, profitability, or capital) may preclude dividend distributions for certain BHCs. Hence, we employ a multivariate regression to control for these nontax considerations.
In addition, we expect that the dividend tax effect will be greater for private BHCs than for BHCs whose stock is publicly-traded on stock exchanges. First, information asymmetry between managers and stockholders and agency problems are likely to be smaller for private BHCs than for public BHCs due to the high concentration of private BHC ownership among
opportunities to pay “deductible dividends”. If neither constraint is binding, then private firms would probably not increase dividend payouts in response to the dividend rate cut.
relatively few shareholders (Beatty and Harris 1999; Niswander and Swanson 2000; Brav, Graham, Harvey and Michaely 2004).12 Our public versus private comparison is analogous to the comparison of public firms across countries with differing levels of legal protection for minority shareholders. La Porta, Lopez-de-Silanes, Shleifer and Vishny (2000) find that dividend policy is related to cross-country differences in shareholder protection. Firms organized in
countries with poor shareholder protection distribute less, presumably because there is little information asymmetry between controlling shareholders and management. 13 Such firms are analogous to the private BHCs in our sample. In contrast, firms organized in countries with good shareholder protection tend to distribute more as dividends, consistent with shareholders using their legal rights to limit professional managers.
Second, as described above, shareholders of private BHCs are very likely to be individuals who are directly affected by the decrease in the dividend tax rate and directly involved in managing the BHC.14 Because managers of private BHCs are likely to be
12 To provide support for this assumption, we examined the most recently filed Annual Report of Bank Holding Companies (FR Y-6) for a sub-sample of 30 private BHCs. Among other items, this report lists “each shareholder of record that directly or indirectly owns, controls, or holds with power to vote 5 percent or more of any class of voting securities” of the BHC. Nineteen of the 30 BHCs in this sub-sample reported ownership information for at least 80 percent of their shareholdings, while the other eleven provided ownership information for, on average, 42 percent of their shareholdings. Of the 19 (11) BHCs providing information for at least (less than) 80 percent of the
shareholdings, 95 (93) percent of the reported shares are owned by individuals. Moreover, the average number of shareholders owning at least a 5 percent interest is 4.6 (3.6) for the BHCs providing information for at least (less than) 80 percent of the shareholdings, respectively. These data suggest that private BHCs have highly concentrated individual ownership.
13 La Porta, Lopez-de-Silanes and Shleifer (1999) find that firms organized in countries with poor legal protection of shareholder rights tend to have controlling shareholders who either manage the firms directly, or are well positioned to monitor hired managers. In contrast, firms in countries with good shareholder protection (e.g., the United States) tend to have widely dispersed ownership and professional management.
14 Although the ownership of private BHCs is concentrated among a small number of individual shareholders, these firms have not elected to be taxed as subchapter S corporations. There are two related explanations for this decision. First, banks were not permitted to elect S status prior to 1997 and, therefore, would have been organized only as C corporations prior to that time. Second, upon conversion to S status, a BHC previously organized as a C corporation would be subject to corporate-level taxes on any net unrealized built-in gains that are realized within the succeeding 10-year period. More importantly, because SFAS No. 109 requires firms to write-off deferred tax assets when their tax status changes, banks electing S status would experience an immediate reduction in their regulatory capital. Hodder et al. (2003) show that the likelihood of banks converting to S corporation status is decreasing in the amount
shareholders, we expect their dividend payout decisions to be much more aligned with
shareholders’ personal tax incentives (Ke 2001). This rationale is supported by prior research on the relation between insider ownership and dividend policy of non-financial, publicly-traded firms. Jensen et al. (1992) show that dividend payouts are negatively related to insider ownership of public firms, which may be explained by lower agency costs or greater attention to individual tax considerations when dividends are tax disadvantaged. Blouin et al. (2004) find that public firms’ responsiveness to JGTRRA is increasing in the level of insider ownership, which again suggests greater attention to personal tax incentives. It is important to note, however, that
regardless of the level of insider ownership, a publicly-traded firm’s share price is determined in public markets. Therefore, status as a private firm represents more than just an extreme case of insider ownership.
III. RESEARCH METHOD
We investigate the dividend tax effect by first examining the trend of quarterly BHC dividend distributions from 1998 Q1 through 2005 Q1 for sample firms classified by ownership structure (private and public BHCs). We then develop a regression model to test the association between dividend yield and the structural shift in dividend tax rates enacted in JGTRRA while controlling for financial and regulatory considerations as well as time and firm invariant effects.
Sample Selection
We construct our sample by identifying all BHCs filing call reports with the Federal Reserve Board in 2003 Q3. We limit sample BHCs to firms that were not subsidiaries or Subchapter S electing corporations. We further restrict our sample to those BHCs that filed call
of their built-in gains on investment securities and lower for banks that would suffer a decline in their regulatory capital category if their deferred tax assets were written off.
reports for each of 29 consecutive quarters beginning with 1998 Q1 and ending with 2005 Q1. We also eliminate BHCs that could not be clearly identified as either privately-held or publicly-traded.15 Finally, we eliminate firms without valid observations for dividends or firm financial attributes (i.e., assets, liquidity, income, retained earnings, risk-adjusted capital, and growth rate) used in our regression analyses. This procedure results in a panel of 216 private BHCs and 323 public BHCs with complete regression data for each quarter from 1998 Q1 through 2005 Q1.16
Descriptive Statistics and Univariate Results
Table 1 provides descriptive statistics for the sample BHCs partitioned by ownership status. Variables are measured on an annual basis and averaged across calendar years within ownership group. Public BHCs are larger (total assets), faster growing (growth in assets), and more profitable (income and return on assets) than their private counterparts.17 In contrast, private BHCs are more liquid (liquidity ratio)18 and better capitalized (measured by either the retained earnings to assets ratio, RE/TA, or the tier 1 risk ratio) 19 than public BHCs. In order to control for size and because our sample consists of both public and private firms, we measure
15 Initially, we classify sample BHCs as private or publicly traded according to the public ownership field obtained from SNL datasource. SNL classifies public firms in two categories, Major Exchange and Other Public. Major Exchange firms are those trading on the NYSE, AMEX or NASDAQ, whereas Other Public firms include any firm for which SNL can obtain a CUSIP. We include firms as public if we can identify both a ticker symbol and a CUSIP number. Private firms are those without CUSIP numbers or ticker symbols.
16 As we add new quarters of data to the sample, we lose firms from the panel that lack data for those quarters. For example, 13 private (one public) firms were dropped due to missing data in 2004 Q3.
17 Tests were conducted using both parametric (t tests) and nonparametric (Wilcoxon tests) statistics (p < 0.05, two-tailed).
18 The liquidity ratio is liquid assets divided by total liabilities at the end of the period. Liquid assets are defined as the sum of cash and balances due, securities, federal funds sold, and trading account assets less pledged securities. 19 The tier 1 risk ratio is the core capital (tier 1 capital) divided by risk-weighted assets. Core capital is total equity less net unrealized gains and unrealized losses less accumulated net gains (losses) on cash flow hedges less nonqualifying perpetual preferred stock and qualifying minority interests in consolidated subsidiaries less disallowed goodwill and other intangible assets and other miscellaneous adjustments. Risk-weighted assets are defined per the regulatory capital requirements, which classifies assets into four risk categories (e.g., category 1 includes cash and equivalents; category 4 includes commercial and consumer loans) and assigns a risk weight to each category (e.g., category 1 assets are assigned a weight of zero; category 4 assets are 100 percent risk-weighted). In general, the tier 1 capital should equal or exceed 4 percent of risk-weighted assets.
relative distributions (dividend yield) as the amount of dividends deflated by various measures of size including total assets (DIV/TA), income (DIV/INC), and tier 1 capital (DIV/T1C). When averaged across the seven full calendar years in our sample, public BHCs pay higher dividends and have a higher dividend yield (regardless of yield measure) than private BHCs.
--- Insert Table 1 here ---
Figure 1 depicts the percentage of firms in each ownership group that make dividend distributions for each quarter in our sample period. Over 90% (50%) of public (private) BHCs pay a dividend in any given quarter, and this rate is considerably higher than the 20% rate for publicly-traded industrial firms reported by Chetty and Saez (2004). Close inspection of Figure 1 reveals no discernible increase in the percentage of public BHCs paying dividends
post-JGTRRA, but this could be attributable to a ceiling effect caused by the already high distribution rate. In fact, only 25 (five) private (public) BHCs paid no dividends during our sample period.
Figure 2 depicts the median and mean dollar amount (in millions) of quarterly
distributions for private (right axis) and public (left axis) BHCs. The patterns of distributions for both private and public BHCs in Figure 1 exhibit seasonal tendencies. Distributions are more common (Figure 1) and larger (Figure 2) in Q2 and Q4, with the largest distributions typically occurring in Q4. There is also a strong upward trend in both the median and mean dollar amount of distributions for public firms throughout the sample period. Although there appears to be a marked increase in distributions in 2003 Q4, particularly for private BHCs, the overall trend makes it difficult to assess how much of this increase is due to the dividend tax cut. Additionally, growth in firm assets over time, particularly for public firms, might explain the upward trend in distributions. To assess the trend in distributions relative to size, Figure 3 presents the quarterly
dividend yields (DIV/TA) over the same time period. Deflating distributions by assets controls for growth over time, but the pattern of seasonality remains and is especially pronounced for the private BHCs.
--- Insert Figures 1, 2, and 3 here
---
Table 2 presents the mean and median dividend distributions, DIV, by quarter across private (Panel A) and public (Panel B) BHCs for 2002, 2003, 2004, and 2005 Q1. For
comparison, Table 2 also presents dividend distributions deflated by various measures of size including assets (DIV/TA), income (DIV/INC), and tier 1 capital (DIV/T1C). Dividend
distributions show a remarkable pattern of increases from 2003 Q3 through 2005 Q1 that is consistent with a tax effect. For example, the mean distribution (DIV) by private BHCs increased 126% in 2003 Q4 relative to the same quarter of 2002 (from $435,000 to $983,000). Although the mean distribution dropped in 2004 Q4 to $639, this amount is still 47% higher than 2002 Q4. Interestingly, the median distribution does not drop off over time as it increased from $212 in 2002 Q4 to $240 in 2003 Q4 and $275 in 2004 Q4. Similar comparisons in Panel B suggest a more steady tax effect for public BHCs. The mean distribution for public BHCs increased 20% from 2002 Q4 to 2003 Q4 (from $16.4 million to $19.6 million) followed by an additional 33% increase from 2003 Q4 to 2004 Q4 (from $19.6 million to $26.1 million). Deflated distribution measures (that is, DIV/TA, DIV/INC, and DIV/T1C) show a similar pattern of increases.
--- Insert Tables 2 and 3 here
---
Table 3 provides descriptive statistics of the financial characteristics and annual distributions for sample firms at year-end 2003 by high and low dividend quartiles for private
(Panel A) and public (Panel B) BHCs. Private BHCs in the high dividend quartile are smaller (TA), more profitable (ROA), and more liquid (LIQ), and better capitalized (T1 Ratio, RE/TA) than private BHCs in the low dividend quartile. Public BHCs in the high dividend quartile are more profitable, more liquid, and better capitalized than the public BHCs in the low quartile. In contrast, public BHCs in the low yield quartile tend to have higher growth rates (Growth) than the firms in the high yield quartile. These differences between high and low dividend quartiles suggest the importance of controlling for firm characteristics in our later tests.
Tests of the Effect of JGTRRA on Dividend Distributions
We test for the effect of the dividend tax reduction after the enactment of JGTRRA by estimating regressions of dividend yield (DIV/TA) on a tax rate cut indicator variable (DRATE) for each ownership group. We employ multiple control variables that represent firm-specific factors previously found to be associated with BHC distributions. Because dividends vary across our panel of firms over time, we control for time and firm effects by estimating the random effects regression in equation (1).20 To allow for the possibility that estimated coefficients on the various explanatory variables might differ by ownership structure, we estimate equation (1) separately for private and public BHCs.
it Kit K t it DRATE X TA DIV / =γ0 +γ1 +γ +ε (1) where: it TA
DIV / = firm i’s dividend yield in period t calculated by dividing total dividend distributions during period t by total assets at the end of period t, t
DRATE = a binary variable that equals one if the top individual tax rate on dividend distributions is 15 percent in period t, and zero otherwise,
20We employ a version of a random effects regression that uses a generalized least squares estimator that produces a matrix-weighted average of the between regression and within regression estimators. In addition to correcting for fixed effects across time periods and between effects across firms, this algorithm also corrects significance levels for first-order autoregressive disturbances. For more information about this methodology, see Wooldridge (2002).
Kit
X = a vector of K explanatory variables representing the annual financial characteristics of firm i during period t.
Although our random effects regression methodology controls for effects across firms and time, we employ two methods to control for the quarterly seasonality exhibited by our distribution data. First, we form annual periods by aggregating quarterly data across four consecutive calendar quarters beginning with Q3 and ending with Q2 of the subsequent year. This annualized period ending with Q2 of 2004 reflects firm activities for the first four full quarters after the enactment of JGTRRA. We then estimate equation (1) using this annualized data where DRATE is coded as one for the last year in the series (i.e. the four quarters ending 2004 Q2) and zero otherwise. For comparative purposes we also estimate equation (1) using calendar years, where DRATE is coded as one for 2004 and zero for all preceding years. Although this specification allows us to use data for two additional post-JGTRRA quarters (i.e. 2004 Q3 and 2004 Q4), it may also weaken the estimated tax effect because the last two quarters of calendar year 2003 occurred after JGTRRA was enacted.21 Second, as an alternative means of controlling for quarterly seasonality, we estimate equation (1) separately on a quarter-by-quarter basis. That is, we estimate equation (1) using sample firm observations for Q1 exclusively, and then estimate equation (1) using sample data for Q2 exclusively, and so forth. In each of these regressions, DRATE is coded as one for quarters ending after the enactment of JGTRRA.
Consistent with the prior literature, we include variables to control for profitability, liquidity, regulatory constraints based on minimum capital requirements, and agency costs. In his
21 Although JGTRRA was approved by the House by a comfortable margin, approval by the Senate on May 23, 2003 required the tie-breaking vote of Vice President Cheney (51-50). Therefore, ultimate approval by President Bush was assured. The dividend tax provision of JGTRRA was effective retroactively for taxable years beginning after December 31, 2002. However, because the legislation passed the Senate by such a narrow margin, firms were unlikely to change their dividend policy until after May 23, 2003. In sensitivity analysis we find that calendar year regression results estimated with DRATE coded as one for both 2003 and 2004 are largely consistent with the calendar year results reported in the Table 4, despite the fact that 2003 actually includes two pre-JGTRRA quarters.
classic paper examining dividend policy, Lintner (1956) finds a positive relation between dividend payments and firm profitability.22 Specifically, we control for firm profitability by including ROA measured as income for the period divided by total assets at the end of the period.23 For completeness we include a liquidity measure, LIQ, calculated as the ratio of liquid assets to total liabilities at the end of the period. BHCs are subject to regulatory capital
requirements not imposed on industrial firms. Therefore, to control for risk and capital adequacy, we include the Tier 1 risk-adjusted capital ratio, T1 Ratio. Finally, following DeAngelo,
DeAngelo, and Stulz (2004), we include the ratio of retained earnings to total assets, RE/TA, to control for public firms’ incentive to pay dividends to mitigate agency problems that might become more severe as profits accumulate.
IV. REGRESSION RESULTS
Panel A (Panel B) of Table 4 presents the results of estimating equation (1) on an annual basis for private (public) BHCs. In column (1), the regression is estimated using annual periods composed of hypothetical fiscal years ending in Q2 whereas column (2) presents the regression results composed of calendar years. The estimated coefficient for DRATE is significantly
positive across both the regressions for both private and public BHCs. Using a test of equality of coefficients across regressions, we find that the DRATE coefficients for private BHCs are
significantly greater than those of the public BHCs for both the Q2 regression and the calendar
22 Subsequent studies reevaluated the Lintner (1956) model and find the positive relation between profitability and dividends to be robust (Fama and Babiak 1968; Bernartzi, Michaely, and Thaler 1997; Brav, Graham, Harvey and Michaely 2004).
23 In annual regressions, ROA is measured by summing income for four consecutive quarters within period t and dividing by total assets at the end of period t. In quarterly regressions, ROA is income for that quarter divided by total assets at the end of the same quarter.
year regression (χ2 = 3.85; p = 0.03, and χ2 = 6.48; p = 0.01, respectively).24 This is consistent with our expectation that private BHCs respond more readily to their shareholders’ personal tax incentives than do public BHCs. These estimates suggest that from a policy perspective the dividend tax cut was more likely to benefit shareholders of private BHCs than publicly-traded BHCs.
--- Insert Table 4 here ---
The control variables included in the regressions are largely consistent with expectations. Profitability (ROA) exerts a powerful positive influence on distributions for all ownership
structures; whereas, liquidity (LIQ) is not statistically significant in a consistent manner for either private or public BHCs. In contrast, Growth is significantly negative for both public and private firms, which is consistent with growing firms re-investing free cash flow rather than paying dividends. Finally, our results are consistent with the maturity/agency conjecture of DeAngelo, DeAngelo, and Stulz (2004) that higher retained earnings (RE/TA) is positively associated with dividend yield for public BHCs, suggesting that public firms pay dividend to limit
overinvestment by managers. Importantly, RE/TA is not statistically significant for private BHCs, consistent with the lack of agency concerns where ownership is highly concentrated and closely connected to management.
Table 5 presents the results of estimating equation (1) on a quarter-by-quarter basis as an alternative means of controlling for quarterly seasonality. In the Q3 and Q4 models, each firm has five pre-JGTRRA observations and two post-JGTRRA observations. In the Q1 (Q2) model,
24 We use the suest command in STATA to assess the equality of DRATE coefficients across public and private OLS regressions. This method allows testing across equations when error variances are not the same across equations by combining multiple estimation results into simultaneous results with a single stacked parameter vector and a robust variance–covariance matrix (see Clogg et al 1995 for details).
each firm has six pre-JGTRRA observations and two (one) post-JGTRRA observations. A
significantly positive coefficient on DRATE for a particular quarter indicates that dividend yields were higher for quarters after JGTRRA relative to the same quarters in pre-JGTRRA years. Columns (1) through (4) of Table 5 contain the results from estimating quarterly random effects regressions for quarters 3, 4, 1, and 2, respectively, corresponding to the four successive quarters since the enactment of JGTRRA.
--- Insert Table 5 here ---
The results from estimating regressions for each quarter suggest that private BHCs (panel A) significantly increased dividend payouts after enactment of JGTRRA. The coefficients on DRATE are significantly positive for all but the Q2 model. The largest DRATE coefficients, 0.029 and 0.026, occur in the Q4 and Q3 models, respectively. Recall from our earlier analysis of seasonality that private BHCs have historically paid their largest (smallest) distributions in Q4 (Q3). In contrast, the quarterly regressions for the public BHCs provide evidence of a weaker response to the enactment of JGTRRA in Q3 and Q4. The coefficients on DRATE for public BHCs are significantly positive for all but the Q4 model (Table 5, Panel B). However, in comparison to the results for private BHCs, the DRATE coefficients for public BHCs are
significantly smaller in magnitude in the Q3 (χ2=5.44, p=.02) and Q4 (χ2=3.42, p=.03) models.
V. SUPPLEMENTAL ANALYSIS
We assessed the sensitivity of our regression estimates for equation (1) in several ways. First, to mitigate the effect of extremely large banks and money center banks, we verified that our regression results are robust to the elimination of 45 firms with assets in excess of $10.65 billion at the end of 2003 Q3. Second, we analyzed the effect influential observations by
re-estimating the annual regressions after excluding potentially influential observations identified using the procedures in Belsley, Kuh, and Welsch (1980). Exclusion of these observations produces results that are not qualitatively different than those presented in Table 4. Third, we re-estimated the annual regressions using DIV/INC and DIV/T1C as the dependent variable in lieu of using DIV/TA. Again, the estimated tax effects were not qualitatively different than those presented in Table 4. Finally, we also re-estimated equation (1) using alternative fiscal years composed of four consecutive quarters ending on Q1 and Q3. The results using these two alternative samples are also consistent with those presented in Table 4.
In all of the preceding analyses, we control for quarterly seasonality by either pooling quarters within years (i.e., annualizing) or estimating the model on a quarter-by-quarter basis. As an alternative, we also estimate equation (1) using all quarterly observations in a model that includes both annual and quarter fixed effects, as well as firm between-effects and a correction for autocorrelation. This model is potentially more powerful because it uses all available information, including observations for 6,264 (9,367) firm-quarters for private (public) BHCs. Untabulated results indicate that these regressions are statistically significant (χ2 tests p < .001) but have less explanatory power than the annual regressions reported in Table 4 (R2 is 2.7 for private and 8.1 for public BHCs). Nevertheless, the estimated coefficient on DRATE is
significantly positive for both private and public BHCs, consistent with our earlier analyses. The estimated coefficients on the control variables are also consistent with those reported previously. Because dividend payouts are naturally censored at zero, we also estimated annual tobit
regressions to control for firm-quarter observations where the firm did not make a dividend distribution. Untabulated results of these tobit regressions are qualitatively similar to those
reported in Table 4, which suggests that including firm-quarter observations without dividend distributions does not bias our results.
The regression results presented in Tables 4 and 5 are consistent with the contention that the tax rate reduction enacted by JGTRRA caused BHCs to increase dividend distributions. A related issue is whether this legislation also made dividend distributions more likely for firms that had not previously paid dividends. Figure 1 suggests a very high proportion of BHCs pay regular dividends. Indeed, only 25 (5) private (public) firms did not make at least one quarterly dividend payment sometime during the period 1998 through 2005 Q1. To further examine the tendency of BHCs to make distributions, Table 6 presents the number of firms that did not make any dividend distribution during each calendar year. Excluding 2004, the proportion of private firms forgoing a dividend in any calendar year from 1998 through 2004 varies from 14.4 percent (31 firms) to 17.6 percent (38 firms) of our sample of 216 private BHCs. In contrast, the
proportion (and number) of public firms not making at least one dividend is smaller (from 3.7 percent to 4.6 percent of our sample).
--- Insert Table 6 here ---
If a dividend tax rate reduction stimulates the non-paying private BHCs to initiate
dividend distributions, this effect might be evidenced by a higher proportion of non-paying firms initiating dividend distributions after JGTRRA.25 Table 6 also presents the number and
25 It is also possible that firms substituted dividend distributions for stock repurchases after JGTRRA. Because stock repurchases for closely-held companies are typically taxed as dividend distributions, we expect that any substitution between dividends and stock repurchases due to the dividend tax cut would be limited to public firms. The use of stock repurchases by private firms is likely to be driven by nontax considerations (e.g. buying out dissenting or retiring shareholders). Using a sample of firms with calendar year information for stock repurchases, we re-estimated our random effects regression with stock repurchases (the amount of stock repurchases deflated by total assets) as the dependent measure. Consistent with our expectation, the regression results (untabulated) indicate that the level of stock repurchases declined for public firms after JGTRRA.
proportion of non-paying BHCs that make a dividend payment in the calendar year following a calendar year without a dividend. While these numbers are small, there is a slight increase in the proportion of non-paying firms that initiate dividends after enactment of JGTRRA in 2003 (15.8 percent for private firms and 28.6 percent of public firms). Because of the high proportion of public BHCs that regularly pay dividends, it seems unlikely that the reduction in the dividend tax rate would have a discernable effect on the few public BHCs that do not pay dividends. Hence, we explore the potential effect of JGTRRA on the likelihood of dividend payments by BHCs by focusing our attention of private BHCs.
To formally test whether private firms are more likely to pay a dividend after JGTRRA, we estimate a self-selection model that integrates the decision to pay a dividend into dividend yield estimation. We are unaware of previous work to model the dividend payout decision for private firms; therefore, we view this analysis as exploratory. The self-selection model is specified as follows: it Kit K t it DRATE X u PAY =δ0 +δ1 +δ + 1 (2) it Kit K t it DRATE X u TA DIV/ =
γ
0+γ
1 +γ
+ 2 (3) where:PAY it = one if firm pays a dividend during the year, and zero otherwise,
it
TA
DIV / = firm i’s dividend yield in period t calculated by dividing total dividend distributions during period t by total assets at the end of period t, t
DRATE = a binary variable that equals one if the top individual tax rate on dividend distributions is 15 percent in period t, and zero otherwise,
Kit
X = a vector of K explanatory variables used in equation (1) representing the annual financial characteristics of firm i during period t.
Table 7 reports the results of the distribution choice (Panel A) and distribution amount (Panel B) equations.26 Overall, the results from the distribution choice model show no evidence that the likelihood of paying dividends changes after the enactment of JGTRRA. ROA is
significantly positively related to the likelihood of a private BHC making a dividend distribution indicating that more profitable firms are more likely to pay dividends. Growth is significantly negatively related to the likelihood of paying dividends suggesting that higher growth firms are less likely to make dividend distributions. None of the other control variables are significant.
--- Insert Table 7 here ---
The results from the dividend amount model are consistent with our previous results for private BHCs: dividend yield increased after the enactment of JGTRRA. Interestingly, only ROA is significant in the distribution amount equation once the self-selection issue is addressed. Growth is not significant suggesting that although growth is an important factor in the decision to pay a dividend, once the decision is made to pay a dividend, growth does not influence the amount of the distribution. Similar to our results in Table 4, LIQ, T1 Risk, and RE/TA are not statistically significant for private firms. The χ2 tests of independent equations are not significant in any of the models indicating that the distribution choice and the distribution amount are not correlated for this sample. In essence, this means that we may estimate the dividend amount equation using OLS without concern of any selection bias.
26 Table 7 does not present the results from estimating the Quarter 2 fiscal year because the Heckman estimation
VI. CONCLUSION
We analyze the dividend policy for private and public bank holding companies with the expectation that private firms are more likely to be influenced by individual tax considerations while public firms are more likely to consider nontax factors. We compare the level of dividend distributions before and after enactment of JGTRRA for both public and private BHCs to assess the extent to which differential taxation of dividend income versus capital gains for individual shareholders influences corporate dividend payouts.
So far as we are aware, this is the first study to compare the dividend policies of publicly-traded firms to those of privately-held firms. Relative to publicly-publicly-traded BHCs, privately-held BHCs have more tax reasons to avoid dividends because their shareholders are more likely to be individuals and have fewer non-tax reasons to pay dividends (e.g. agency costs) because their shareholders are likely to be more directly involved in managing the firm. Therefore, comparing payouts of private versus public firms, in combination with the dividend tax cut that was a central part of JGTRRA, provides a very powerful test of the hypotheses that the tax
disadvantages of dividends for individual shareholders discourage corporations from paying dividends while agency costs have the opposite effect. Our analysis of distributions by bank holding companies also fills a void in the financial literature, most of which features tests based exclusively on industrial firms.
We find that BHCs’ distributions are seasonal across quarters and trending upward over time throughout our sample period. Unlike industrial firms, the vast majority of BHCs paid some dividends prior to JGTRRA regardless of ownership structure. Consequently, we do not find a significant increase in the percentage of BHCs paying dividends after JGTRRA. However, we do find that the amount of dividends paid by BHCs (as a percentage of assets) is sensitive to the
differential tax rates on dividend income versus capital gains for individuals, particularly for privately-held BHCs. In the periods prior to JGTRRA, when the maximum tax rate on dividend income exceeded the maximum tax rate on capital gains for individuals, we find that privately-held BHCs have significantly lower dividend yields than the publicly-traded counterparts. This result suggests that private BHCs retain earnings consistent with tax incentives for individual shareholders and/or that public BHCs distribute earnings to mitigate potential agency problems. In the quarters after enactment of the dividend tax cut, we find that both private and public BHCs increased dividend payouts, but private BHCs increased dividend yields to a significantly greater extent than publicly-traded BHCs. This result is consistent with the notion that reducing the tax disadvantage of dividends relative to capital gains enabled private firms to satisfy pent-up liquidity demands of their individual shareholders at minimal tax costs.
Much of the political rhetoric preceding JGTRRA focused on the intended benefits to shareholders of public firms (e.g. increasing the value of the stock market and mitigating corporate governance problems). Indeed, we find that dividend increases by public BHCs were both statistically and economically significant. However, despite the focus on these stated objectives on public firms, our analyses indicate that, at least for our sample of bank holding companies, JGTRRA bestows even greater benefits on the shareholders of privately-held corporations.
REFERENCES
Allen, F., and R. Michaely, 2003, “Payout Policy,” in Constantinides, G., M. Harris and R. Stulz (eds.) Handbook of the Economics of Finance (Amsterdam: North-Holland).
Barlett, B. 2002. My conference call. National Review Online (August 21).
Beatty, A., and D. Harris. 1999. The effects of taxes, agency costs and information asymmetry on earnings management: A comparison of public and private firms. Review of
Accounting Studies 3: 299-326.
Belsley, David A., Edwin Q. Kuh, and Roy E. Welsch, 1980,Regression Diagnostics:
Identifying Influential Data and Sources of Collinearity (John Wiley & Sons, New York). Bernartzi, R. Michaely, and Thaler, 1997, Do changes in dividends signal the future or the past?
Journal of Finance 52 (3), 1007-1034.
Black, F. 1976. The dividend puzzle. Journal of Portfolio Management 2: 5-8.
Blouin, J. L., J. S. Ready, and D. A. Shackelford, 2004, Did dividends increase after the 2003 reduction in dividend tax rates? Working paper, University of North Carolina.
Brav, A., J. R. Graham, C. R. Harvey, and R. Michaely, 2004, Payout policy in the 21st century. NBER working paper 9657.
Brown, J.R., N. Liang, and S. Weisbenner. 2005. Executive financial incentives and payout policy: Firm responses to the 2003 dividend tax cut. Working paper, University of Illinois.
Bush, G.W. 2003. Remarks by the President at signing of Jobs Growth Tax Relief Reconciliation Act of 2003. Available at: http://www.whitehouse.gov/news/releases/2003/05/20030528-9.html.
Chetty, R. and E. Saez, 2004, Do dividend payments respond to taxes? Preliminary evidence from the 2003 dividend tax cut. NBER working paper 10572.
Clogg, C., E. Petkova, and A. Haritou. 1995. Statistical methods for comparing regression coefficients between models. American Journal of Sociology 100 (5): 1261-1293. Davis, B. 2003. Treasury chief struggles to sell tax plan. The Wall Street Journal (May 5): A.4. DeAngelo, H., and L. DeAngelo, 2004, Payout policy irrelevance and the dividend puzzle.
Working paper, University of Southern California.
DeAngelo, H., L. DeAngelo, and R. M. Stulz, 2004, Dividend policy, agency costs, and earned equity. Working paper, University of Southern California.
Esenwein, G.A., and J.G. Gravelle. 2003. The taxation of dividend income: An overview and economic analysis of the issues. Congressional Research Service report, updated May 15, 2003.
Fama, E. and H. Babiak, 1968, Dividend policy: an empirical analysis. Journal of the American Statistical Association, 63 (324), 1132-1161.
Fama, E. and K. French, 2001, Disappearing dividends: Changing firm characteristics or lower propensity to pay? Journal of Financial Economics 60, 3-43.
Feldstein, M. and J. Green, 1983, Why do companies pay dividends? American Economic Review 73, 17-30.
Grullon, G. and R. Michaely, 2002, Dividends, share repurchases and the substitution hypothesis. Journal of Finance 57 (4), 1649-1684.
Grullon, G., R. Michaely, and B, Swaminathan, 2002, Are dividend changes a sign of firm maturity? Journal of Business 75 (3), 387-424.
Hodder, L., M. McAnally, C. Weaver. 2003. The Influence of Tax and Non-tax Factors on Banks’ Choice of Organizational Form. The Accounting Review 78 (1): 297-325. Julio, B. and D. L. Ikenberry, 2004, Reappearing dividends. Journal of Applied Corporate
Finance 16 (4, Fall): 89-100.
Kalay, A. and U. Loewenstein, 1986, The information content of the timing of dividend announcements. Journal of Financial Economics 16, 373-388.
Ke, B. 2001. Taxes as a determinant of managerial compensation in privately held insurance companies. The Accounting Review 76 (4, October): 655-674.
Lambert, R., W. Lanen, and D. Larcker. 1989. Executive stock option plans and corporate dividend policy. Journal of Financial and Quantitative Analysis 24 (4): 409-425.
La Porta, R., F. Lopez-de-Silanes, and A. Shleifer. 1999. Corporate ownership around the world. The Journal of Finance 54 (2, April): 471-517.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny. 2000. Agency problems and dividend policies around the world. The Journal of Finance 55 (1, February): 1-33. Lintner, J. 1956, Distribution of incomes of corporations among dividends, retained earnings,
and taxes. American Economic Review 46 (2), 97-113.
Michaely, R., R. Thaler, and K. Womack, 1995, Price reactions to dividend initiations and omissions: overreaction or drift? Journal of Finance 50 (2), 573-608.
Miller, M. and F. Modiglianni. 1961. Dividend policy, growth, and the valuation of shares. Journal of Business 34, 411-433.
Nissim, D. and A. Ziv, 2001, Dividend changes and future profitability. Journal of Finance 56 (6), 2111-2133.
Niswander, F. and E. Swanson, 2000, Loan, security, and dividend choices by individual
(unconsolidated) public and private commerce BHCs. Journal of Accounting and Public Policy 19, 201-235.
Ofer, A. R. and D. R. Siegel, 1987, Corporate financial policy, information, and market expectations: An empirical investigation of dividends. Journal of Finance 42 (4), 889-911.
Stiglitz, J. 1973. Taxation, corporate financial policy, and the cost of capital. Journal of Public Economics 2, 1-34.
Wooldridge, J.M. 2002. Chapter 10. Econometric Analysis of Cross Section and Panel Data. Cambridge, MA: The MIT Press.
Figure 1
Percentage of BHCs making distributions by ownership structure
0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0% 1998 q1 q2 q3 q4 1999 q1 q2 q3 q4 2000 q1 q2 q3 q4 2001 q1 q2 q3 q4 2002 q1 q2 q3 q4 2003 q1 q2 q3 q4 2004 q1 q2 q3 q4 2005 q1 Quarter P e rcen tag e private public Figure 2 Quarterly distributions ($ in millions) -100 4900 9900 14900 19900 24900 29900 1998 q1 q2 q3 q4 1999 q1 q2 q3 q4 2000 q1 q2 q3 q4 2001 q1 q2 q3 q4 2002 q1 q2 q3 q4 2003 q1 q2 q3 q4 2004 q1 q2 q3 q4 2005 q1 Quarter publ ic dis tr ibutions 0 200 400 600 800 1000 1200 p rivat e d istrib u ti o ns
Figure 3
Quarterly dividend yield (DIV/TA)
0.000 0.020 0.040 0.060 0.080 0.100 0.120 0.140 1998 q1 q2 q3 q4 1999 q1 q2 q3 q4 2000 q1 q2 q3 q4 2001 q1 q2 q3 q4 2002 q1 q2 q3 q4 2003 q1 q2 q3 q4 2004 q1 q2 q3 q4 2005 q1 Quarter yi eld ( p er cen t)
median private median public mean private mean public
TABLE 1
Financial characteristics of pooled panel of bank holding companies (BHCs) for calendar years 1998-2004 by ownership structure
Variable Ownership Mean Std dev Quartile 1 Median Quartile 3 Assets (TA) $ millions Private 636 838 277 379 621
Public 12121 64855 486 971 3091
Income $ millions Private 6.74 10.64 2.61 3.85 6.45 Public 142.91 703.32 4.86 10.06 35.31 ROA % Private 1.041 0.465 0.809 1.028 1.245
Public 1.119 0.462 0.895 1.108 1.365 Liquidity % Private 26.72 14.48 15.92 24.30 35.00 Public 20.35 11.58 12.12 17.90 25.59 Tier 1 Capital % Private 56.13 67.45 25.39 34.38 55.16
Public 825.51 3911.93 42.46 81.15 259.13 Tier 1 Ratio % Private 15.56 9.16 10.68 13.27 18.00
Public 12.68 4.28 10.21 11.82 13.98
RE/TA % Private 7.80 3.60 5.52 7.69 9.57
Public 5.20 2.88 3.00 5.11 7.10
Growth % Private 8.14 10.27 2.53 6.62 11.52 Public 12.02 16.12 3.77 8.68 15.47 Dividends (DIV) $ millions Private 1.39 3.62 0.25 0.77 1.46
Public 62.25 322.99 1.58 3.58 11.16 DIV/TA % Private 0.257 0.299 0.055 0.177 0.366 Public 0.432 0.252 0.273 0.418 0.567 DIV/INC % Private 24.99 34.04 6.21 18.83 34.72 Public 42.20 58.83 27.64 37.80 47.06 DIV/T1C % Private 2.71 3.02 0.65 2.07 3.88 Public 5.06 2.93 3.30 4.93 6.52
Notes: The sample consists of a panel of 553 bank holding companies (216 private firms and 323 public firms) that filed consecutive call reports with the Federal Reserve Board for each quarter from 1998 through the first quarter of 2005 (29 quarters). Sample BHCs are classified as Private if the firm’s stock ownership is not traded on a public exchange or Public if the firm’s shares are on traded on a major exchange. Assets represent total reported assets at the end of the period; Income represents net income for the calendar year; ROA is the ratio of net income to total assets; LIQ is the ratio of liquid assets (cash and balances due plus securities plus federal funds sold plus trading account assets less pledged securities) to total liabilities; Tier 1 Capital is core capital; Tier 1 Ratio is the ratio of tier 1 capital to risk-weighted assets; RE/TA is the ratio of retained earnings to total assets; Growth is annual growth in assets from the prior year (excluding 1998); Dividends are total dividend distributions made during the calendar year; DIVTA is the ratio of total dividend distributions to total assets; DIV/INC is the ratio of total dividend distributions to net income (missing if negative net income); and DIV/T1C is the ratio of total dividend distributions to tier 1 capital. All ratios are expressed as a percentage in whole numbers.
TABLE 2
Quarterly dividend distributions and yield for 2002-2005
Panel A: Quarterly distributions and dividend yield for 216 Private firms
Dividends (DIV) DIV/TA DIV/INC DIV/T1C
Year Quarter Mean Median Mean Median Mean Median Mean Median 2002 1 297 124 0.051 0.027 20.23 10.08 0.550 0.324 2 309 110 0.057 0.024 19.23 9.45 0.600 0.257 3 254 12 0.040 0.002 14.16 2.45 0.443 0.026 4 435 212 0.088 0.046 40.91 20.36 0.903 0.522 2003 1 324 131 0.051 0.031 20.16 11.75 0.553 0.361 2 357 118 0.064 0.026 20.62 10.55 0.634 0.288 3 377 99 0.069 0.018 30.27 6.74 0.700 0.227 4 983 240 0.122 0.045 51.69 23.36 1.197 0.550 2004 1 384 151 0.070 0.029 32.48 12.38 0.741 0.339 2 394 154 0.067 0.030 23.65 13.37 0.668 0.368 3 376 106 0.062 0.020 22.45 8.18 0.654 0.233 4 639 275 0.113 0.066 82.65 25.12 1.120 0.649 2005 1 442 162 0.063 0.034 24.19 14.21 0.636 0.377
Panel B: Quarterly distributions and dividend yield for 323 Public firms
Dividends (DIV) DIV/TA DIV/INC DIV/T1C
Year Quarter Mean Median Mean Median Mean Median Mean Median 2002 1 15344 907 0.108 0.105 37.29 35.98 1.258 1.252 2 16022 982 0.113 0.107 45.40 35.73 1.319 1.274 3 16106 965 0.103 0.102 41.21 33.10 1.232 1.243 4 16407 1041 0.116 0.110 40.18 36.76 1.358 1.310 2003 1 16734 1042 0.104 0.106 36.72 37.01 1.256 1.252 2 17293 1082 0.112 0.108 42.74 36.91 1.343 1.296 3 19171 1107 0.106 0.107 48.42 36.19 1.264 1.254 4 19632 1266 0.126 0.114 53.34 40.30 1.449 1.341 2004 1 20155 1192 0.114 0.109 40.93 39.33 1.339 1.284 2 21947 1231 0.121 0.112 50.69 40.76 1.427 1.309 3 24387 1194 0.109 0.109 38.04 37.41 1.296 1.277 4 26076 1326 0.121 0.113 50.45 40.41 1.416 1.345 2005 1 25964 1345 0.116 0.109 43.23 40.00 1.364 1.300
Notes: Dividends are total dividend distributions (undeflated dollars in thousands) made during the calendar quarter. Table 1 provides a detailed description of the dividend yield measures.
TABLE 3
Financial Characteristics for BHC panel at year-end 2003 by ownership structure and dividend yield
Panel A: Private firms by high and low dividend (DIV/TA) yield
Variable Yield Mean Std dev Quartile 1 Median Quartile 3
Assets (TA) * Low yield 994 1245 339 565 1013
High yield 629 986 282 341 494
Income (INC ) Low yield 9.19 12.19 1.82 4.69 10.08
High yield 8.08 15.43 2.95 4.07 7.08
ROA * Low yield 0.876 0.641 0.518 0.827 1.036
High yield 1.157 0.509 0.961 1.114 1.405
Liquidity Low yield 26.20 13.38 14.46 24.55 33.20
High yield 29.41 15.29 18.04 26.69 35.87
Tier 1 Capital * Low yield 85.91 93.82 31.84 49.10 94.33
High yield 61.89 80.37 29.33 37.32 48.53
Tier 1 Ratio * Low yield 14.29 5.86 10.58 12.33 16.13
High yield 17.55 7.59 12.87 15.65 19.79
RE/TA * Low yield 7.57 4.54 5.02 7.82 9.54
High yield 9.12 3.67 7.34 8.71 10.39
Growth Low yield 10.74 20.17 1.06 6.74 12.24
High yield 4.50 4.69 1.93 4.60 7.85
Dividends * Low yield 0.19 0.44 0.00 0.00 0.21
High yield 5.09 13.81 1.34 2.24 3.31
DIV/TA * Low yield 0.028 0.053 0.000 0.000 0.035
High yield 0.686 0.590 0.430 0.542 0.700
DIV/INC * Low yield 6.60 27.87 0.00 0.00 4.55
High yield 59.99 47.48 36.62 46.85 58.02
DIV/T1C * Low yield 0.32 0.64 0.00 0.00 0.37