Embed Size (px)
Citation preview
Dividends Divided:
Australian Dividends and Shareholder Reinvestment
James Murray*
Christchurch Polytechnic Institute of Technology
Michael Skully
Monash University
Abstract
This paper examines dividend size when the firm offers a dividend reinvestment plan (DRP).
Reinvestment affects total dividend size as it is jointly determined by management and
shareholders. Management merely declares the maximum possible dividend but the
effective dividend is determined by shareholder reinvestment. DRPs are often used by listed
Australian companies and on average over forty per cent of their declared cash dividend is
reinvested. Such reinvestment suggests dividend policy research should incorporate these
DRP effects but this is one of the few papers to do so. The empirical results indicate that
management considers the reinvestment expected when determining dividend size and can
impact this reinvestment by modifying plan design. Together this allows them to influence
the effective dividend’s size and manage the agency costs of excess free cash flow.
*Corresponding AuthorDepartment of Business
Christchurch Polytechnic Institute of TechnologyPO Box 540
Christchurch 8140New Zealand
1 Introduction
This paper examines the effect of shareholder reinvestment on dividend size when firms use
a dividend reinvestment plan (DRP). Dividend reinvestment allows management to set the
maximum possible dividend but the actual cash payout will depend on shareholders’
reinvestment decisions. This paper extends on a recent trend to use a broader definition of
dividend which includes other types of distribution.1 Prior reinvestment plan research has
considered the effect of a declared dividend on the level of reinvestment but inter-
relationships, such as whether management considers expected reinvestment levels when
setting the declared dividend, are largely unexamined. To fill this gap in the literature we
explicitly combine the reinvestment rate into an analysis of dividend size.
Firms with DRPs offer shareholders the option of having their cash dividend applied to the
purchase of additional shares. He (2009) classifies DRPs into three main types based on the
source of the shares; open market plans, new issue plans, and combined open market and
new issue plans. In open market plans existing shares are purchased, usually by a third party
trustee or DRP administrator, and then transferred to the reinvesting shareholder’s account.
In new issue plans the firm issues new shares directly to the reinvesting shareholders. In
combined plans the firm can mix the two sources according to their needs. Plan type has
little direct importance for individual shareholders as they receive the same number of
shares. It is significant for the firm as open market plans do not affect the firm’s cash
distribution where as new issue plans provide a regular source of new capital. Larkin, et al.
(2005) found that in the United States the majority of DRP firms used open market plans. In
contrast we find that Australian DRP firms almost exclusively use new issue plans.2
This study uses Australian data because of the prevalence of new issue DRPs and the
relative ease in calculating the reinvestment rate in Australia. Australian data also allows us
1 For example Pattenden, Kerry, and Garry Twite, 2008, Taxes and dividend policy under alternative tax regimes, Journal of Corporate Finance 14, 1-16. measure gross dividend as a combination of cash dividends, script dividends, bonus share plans, special dividends and share repurchases, and Henry, Darren, 2011, Ownership structure and tax-friendly dividends, Journal of Banking and Finance 35, 2747-2760. combines cash dividends with repurchases.2 A recent development has been the use of treasury shares in Australian DRPs, this appears similar to open market plans as they use existing shares purchased in the open market, but unlike open market plans the reinvested funds are retained by the company.
[2]
to generate a large dataset of DRP reinvestment rates that would be difficult to replicate
elsewhere. In the United States, for example, researchers are limited to either surveying
firms (see Pettway and Malone,(1973) and Baker and Johnson (1988)), or searching for
reinvestment rate information in annual reports (Lyroudi (1999). Both approaches have
produced relatively small samples. In contrast Australian firms using new issue plans must
disclose the price and quantity of new share issues. This allows researchers to calculate the
reinvestment rate directly. This examination of dividend policy and reinvestment in Australia
is also motivated by a need to understand the impact of Australia’s imputation tax system
on reinvestment.
With new issue DRPs cash is retained by the firm and reduces the size of the actual payout.
This is important if dividend size affects firm value. However the value relevance of
dividends remains an unresolved issue in finance. At one extreme the Miller and Modigliani
(1961) irrelevance propositions show that in a perfect capital market dividends will not
affect firm value, but as real world capital markets are far from perfect there is an
abundance of research exploring possible reasons dividends are value relevant. Of these
Lease, et al. (2000) identified the three main market imperfections linking dividend policy to
firm value as; information asymmetry, taxation, and agency costs. They also identified as
transaction costs, floatation costs and irrational investor behaviour as minor market
frictions that could make dividends value relevant. If, for any reason, dividends can affect
value then reinvestment should also have an effect.
If dividend size is value relevant then management should adopt a dividend policy that
maximises firm value. Ideally shareholders would recognise this and not reinvest if
reinvestment diminishes firm value. However it is not realistic to expect shareholders to
know the optimal dividend size. Even if they did know, they face a coordination problem in
arranging the optimal level of reinvestment. Alternatively, if reinvestment rates are
predictable then management can declare larger dividends with the expectation that
reinvestment will reduce the effective payout close to the optimal level. If this does not
occur, then new issue reinvestment plans risk undermining managements’ value optimising
dividend decisions.
[3]
The empirical results show that reinvestment firms declare smaller dividends and, with
reinvestment over forty percent, the effective distribution is much smaller. Reinvestment
levels are determined by plan features with a firm’s financial and ownership characteristics
having little impact. Dividend size does not appear to affect shareholders’ reinvestment
decisions, but expected reinvestment is a significant factor in setting the declared dividend.
This means that most relationships between financial characteristics and dividend size are
maintained after taking reinvestment into account.
The rest of this paper is organised as follows. Section 2 reviews the types of reinvestment
plans used by Australian firms and existing literature on dividend size and dividend
reinvestment. Section 3 describes the data and research methodology. Section 4 presents
results and analysis. Section 5 concludes the paper.
2 Dividends and Reinvestment Plan LiteratureThe main issue in dividend policy is whether a firm should distribute profits. If a distribution
is made, then there are then two supplementary issues; the amount to be paid and the form of
payment. This paper is primarily concerned with the issue of dividend size when the payment
form includes a reinvestment option. This section first describes the types of reinvestment
plan used in Australia. It then reviews dividend theory and the potential impact of
reinvestment; specifically floatation and transaction costs, investor irrationality, information
asymmetry, taxation, agency costs and reinvestment rates.
2.1 Reinvestment plans in AustraliaAustralian firms use two types of reinvestment plan; the Dividend Reinvestment Plan (DRP)
and the Bonus Share Plan (BSP). Both types allow shareholders to receive shares instead of
cash, but have different tax consequences. DRPs are the most common type. All
shareholders receive taxable dividend income but some allow the firm to retain the cash as
payment for shares. In contrast BSP shareholders elect to receive bonus shares instead of a
dividend. These bonus shares are structured as capital distributions so that participating
shareholders do not receive taxable dividends, but rather pay capital gains tax (CGT) when
the shares are sold.3 As Australia has a full imputation tax system there is no uniform tax
3 Bonus shares are pooled with the original shares from which they are derived, therefore not only are shares owned before the introduction of capital gains tax in 1985 are exempt from CGT but bonus shares derived from pre-CGT shares are also exempt from CGT.
[4]
advantage to DRP or BSP reinvestment. The individual circumstances of some shareholders,
however, can mean that one form of reinvestment is more tax effective than the other.
BSPs were introduced in Australia in 1978. At the time Australia had a classical tax system
with no capital gains tax, creating a significant tax advantage for BSP participants. By
electing to receive shares instead of dividends shareholders avoided income tax and were
not liable for CGT when the shares were sold. Surprisingly despite the obvious tax
advantages BSPs were not widely adopted. Skully (1982) explains this, noting that directors
were uncertain whether the Government would allow the tax advantages to remain and
accounting restrictions meant only firms with a share premium account could issue bonus
shares. The subsequent introduction of CGT, a full imputation tax system, dividend
streaming and capital streaming laws has removed many BSP tax advantages.4 These tax
changes also mean that BSPs and DRPs are now virtually indistinguishable for firms as their
franking account balance is reduced as if a cash dividend was paid, but shareholders do not
receive franking credits if electing BSP reinvestment.
The first Australian DRP was offered in 1981 by General Property Trust. Property trusts are
not subject to income tax when all income is distributed. A DRP allowed General Property
Trust to legally distribute all profits while retaining cash, which would not have occurred if
they had used a BSP. In 1982 Lend Lease Limited became the first Australian company to
offer a DRP.5
Australia’s dividend imputation tax system, introduced in July 1997, has a central role in
corporate dividend policy. Under imputation a resident shareholder’s tax on dividend
income is reduced by a firm’s prior payment of Australian corporate tax. As Marks (1990)
noted, in imputation systems corporate taxes are the prepayment of personal tax. When the
corporate tax rate exceeds a shareholder’s personal tax rate, paying franked dividends
reduces the effective tax rate on corporate income. When the personal tax rate is higher
than the corporate rate, high marginal rate shareholders pay additional tax and should
4 BSP shareholders do not receive franking credits. Before July 1990 the franking credits attached to the portion of dividend not paid due to BSP use were retained by the company. From July 1990 those franking credits were lost but this did not affect unfranked dividends. Capital streaming restrictions introduced in 1998 classify BSP issues as income if attached to a minimally franked dividend. So dividends with a BSP alternative need to be at least 10 percent franked for the BSP shares to be a capital distribution.5 At the time Lend Lease Limited was the manager of General Property Trust.
[5]
prefer investments offering capital returns to dividends. During imputation’s first year of
operation the top marginal rate for individuals was aligned to the corporate rate. Every year
since, however, the top marginal rate has exceeded the corporate rate. The low
concessional tax rates paid by complying superannuation funds have nevertheless created a
strong clientele of institutional investors seeking dividends with franking credits.
Franking credits have no value unless there is an expectation the firm will distribute them
and DRPs make this distribution easier. Bruckner, et al. (1994) raised the possibility that
some firms had accumulated large credit balances which would be difficult to distribute.
This supposition is supported by the Hathaway and Officer (2004) finding that the full value
of credits generated by corporate tax payments was not being passed on to shareholders.
DRPs can help firms manage their franking account by distributing more franking credits
relative to cash. With conventional dividends firms are limited to attaching tax credits worth
D$(tc/1-tc) where D$ is the cash dividend and tc is the corporate tax rate. With DRPs both the
dividend paid in cash (D$) and the dividend paid in shares (Ds) carry franking credits. The
maximum value of credits distributed (D$+Ds)(tc/1-tc) is greater than that of an equivalent
cash only dividend. However, as shareholders determine their level of reinvestment, DRP
firms don’t know how much cash they need to budget for the dividend.6
2.2 Dividend Theory and ReinvestmentWhile the finance literature on normal dividends is far from settled as to why firms pay
dividends and how large those should be, arguments spanning information asymmetry,
taxation, agency costs, floatation costs, transaction costs, and investor irrationality generally
suggest that dividend size is important. Both standard measures of dividend size, yield and
payout ratio, are reduced by reinvestment. Boehm and DeGennaro (2011) distinguish
between the explicit, or declared, dividend and the effective dividend after reinvestment.
They also note that a DRP firm’s declared yield needs to be higher to match the effective
yield of a non-DRP firm.
6 To avoid the issue of uncertain reinvestment levels some firms underwrite their DRPs. Underwriters usually purchase shares at the same discounted price as reinvesting shareholders.
[6]
2.2.1 Floatation Costs, Transaction Costs, and Investor IrrationalityFloatation and transaction cost arguments both suggest firms should select an appropriate
dividend policy that minimises the need for other, costly, transactions. For example if a firm
pays too large a dividend, it will then need to raise funds and incur floatation costs.
Similarly, if a dividend is larger than shareholders want, they can reinvest the surplus
themselves, or if the dividend is too small or non-existent shareholders can sell shares to
create a home-made dividend. Both adjustments involve transaction costs which an
optimally sized dividend will minimise. Without reinvestment there will need to be trade-
offs in setting an optimal dividend. A smaller dividend may reduce floatation costs but
increase shareholders’ transaction costs if they create home-made dividends instead. Even if
transaction costs are minimised, differences between individual shareholders mean home-
made dividends and reinvestment will still occur. A reinvestment plan will lower floatation
costs if DRP issues are less costly than other forms of equity issue. This is likely as there is
less need to prepare prospectuses, engage underwriters or sell shares at a significant
discount to the current market price. Reinvestment plans should minimise transaction costs
by providing large dividends that satisfy shareholders wanting dividend income, eliminating
the need for home-made dividends, while providing reinvestment to those who prefer it.
The investor irrationality arguments summarised by Lease, et al. (2000) indicate that
shareholders still want dividends, even when this imposes other costs. Regular dividend
payments provide income for consumption and help shareholders resist the urge to
withdraw investment capital, or by delegating the dividend decision to management
shareholders are less likely to feel responsible if the decision turns out to be unwise. 7 If
these arguments hold then reinvestment plans undermine a dividend’s benefits, by moving
responsibility for deciding the level of cash received back to the shareholder.
2.2.2 Information Asymmetry and SignallingThe information asymmetry that exists between management and shareholders means that
dividends can signal about a firm’s potential. Lintner (1956) argued that although recent
earnings provide the main basis for shareholders’ expectations of future earnings, dividends
provide valuable supporting evidence. As dividends need to be supported by underlying 7 For example if the share price rises after the dividend it would have been better not to have the dividend and maximize the capital gain, shareholders could feel more regret following a home-made dividend than a normal company dividend.
[7]
earnings, management is unlikely to increase dividends without a similar change to
expected earnings. Such interaction effects, between dividend and earnings announcements
in Australia, have been found by Easton (1991) and How, et al. (1992). Underlying these
signals is the idea that while management can manipulate reported earnings, a declared
dividend must be paid. However, if the firm has a DRP then the effective dividend can be
much smaller than the declared dividend and this will undermine the strength of the signal.
Event studies are commonly used to test information asymmetry and have been extensively
used in DRP research. These use an information asymmetry framework borrowed from
research on share issues, rather than from dividend policy. Generally share issues are seen
as negative signals, as management is more inclined to issue when share prices are
relatively high, but that does not necessarily apply to DRPs. As Scholes and Wolfson (1989)
argue, DRPs commit management to a series of regular small equity issues where timing is
less likely to be of concern. Dubofsky and Bierman (1988) argue that introducing a DRP
provides a positive signal if it shows the firm is using a lower cost source of capital or helping
shareholders avoid transaction costs. The signal could be negative if the DRP moves the firm
away from its optimal capital structure or has high operational costs.
In Australia Chan, et al. (1993) found positive reactions to DRP introductions after the
adoption of the imputation tax system. Similarly Chan, et al. (1996) found the market
responded positively to DRP introduction but their results were mixed when the sample was
divided according to the discount offered . The emphasis of these papers, however, is on the
benefits of DRP use in an imputation environment rather than the reduction of information
asymmetry.
2.2.3 TaxationThe role of tax in dividend policy varies greatly and depends on the design of a country’s tax
system. Classical tax systems discourage dividend payments by imposing an additional layer
of tax when dividends are paid. Integrated tax systems, like Australia’s imputation system,
allow individual tax to be offset by the corporate tax already paid. In Australia’s case the
combination of dividend franking, where tax credits are attached to dividends, and
progressive individual tax rates mean that some shareholders prefer dividends while others
[8]
prefer capital gains. DRP transactions are deemed to occur after dividends, and their
attached franking credits, are paid. Therefore shareholders are taxed identically irrespective
of their preference for cash or shares. However, by reducing the cash outflow DRPs may
allow Australian firms to declare larger dividends and distribute more franking credits than
they could otherwise afford.
Although a classical tax system prevailed in Australia when DRPs were introduced, Bellamy
(1994) and Pattenden and Twite (2008) found a significant increase in the number of firms
offering DRPs following the introduction of dividend imputation. Pattenden and Twite also
show BSP use peaked in 1987 as the introduction of imputation and CGT reduced the tax
advantages BSPs had over DRPs. For example, low tax rate shareholders, including
complying superannuation funds should prefer franked dividends, either as cash or DRP
shares, to capital distributions through a BSP. Both types of plan remain in use today but
very few companies continue to offer BSPs. Almost every BSP firm in our sample
simultaneously offered a DRP.
2.2.4 Agency Costs and Free Cash FlowAgency costs arise when those in control of a firm (agents) do not necessarily act in the best
interests of a firm’s owners (principals), Jensen and Meckling (1976). These costs can take
the form of monitoring costs, bonding costs and residual losses from poor performance
arising from excess free cash flow. Agency costs can arise from the relationship between
shareholders and managers, or between shareholders and debtholders. Dividend policy is
relevant to both relationships, but here we are primarily concerned with the former as both
shareholders and management have a role in determining the effective dividend yield.
Underlying the idea of a free cash flow problem is that managers with excess cash are more
likely to make suboptimal investments, leading to poor performance. Such suboptimal
investment was raised by Jensen and Meckling (1976) and Myers (1977), although they did
not directly address the potential for dividends to reduce the problem. Links between
dividend policy and agency costs have since been well defined. Rozeff (1982) modelled
dividend policy based on the need for finance and monitoring. Firms with adequate
monitoring retain only those earnings required to fund investment while inadequately
[9]
monitored firms could pay larger dividends and raise funds in capital markets, subjecting
them to external scrutiny. Easterbrook (1984) argued that all firms should retain less than
required and use the external capital market to increase monitoring. Jensen (1986) however
regarded dividends as an insufficient solution to the free cash flow problem as management
could too easily reduce dividends. For Jensen interest payments from debt offered a better
solution to the free cash flow problem.
In Australia Henry (2011) found mixed evidence on the use of dividends to reduce agency
costs. For example profitability and retained earnings were positively related to the
likelihood of dividend payment and dividend size, as is expected as these factors indicate
ability to pay. Companies with high current year cash flow, however, were less likely to pay
or paid smaller amounts. Generally tax factors and franking credits showed stronger
connections between ownership and dividend policy.
Irrespective of whether the dividend is large enough to prevent overinvestment, dividends
do reduce management’s discretionary spending, and the overinvestment problem, but new
issue DRPs undermine this by returning cash to the firm. DRP firms will not necessarily face a
worse free cash flow and overinvestment problem if they declare larger dividends to offset
the effects of reinvestment, or if shareholders consider the firms’ capital requirements
when making reinvestment decisions. Boehm and DeGennaro (2011) found that DRP firms
in the United States declare larger dividends, but did not examine the effective dividend
after reinvestment.
2.2.5 Reinvestment RatesIrrespective of the reasons for paying dividends, the effective cash payments are reduced by
reinvestment. Therefore the determinants of shareholder reinvestment will also affect
dividend size. An early study of reinvestment rates by Pettway and Malone (1973) surveyed
non-financial firms using open market plans in the United States. Their highest reinvestment
ratio was 22 per cent of shareholders. Similarly Fredman and Nichols (1982) reported
reinvestment levels up to 20 per cent of shares outstanding and 33 per cent of shareholders,
while Todd and Domain (1997) reported reinvestment up to 15.1 per cent of shares and
[10]
28.7 per cent shareholders. These results suggest that, on average, smaller shareholders
were more likely to reinvest.
In an early study of the first three Australian firms to offer DRPs Wills (1983) found
reinvestment levels ranging from around 10 per cent to 33 per cent of shareholders.
Similarly Anderson (1986) reported reinvestment ranging from 14 per cent to 29 per cent
for eight Australian firms responding to his survey. Both studies were undertaken when
DRPs were new to Australia. Our more recent data shows reinvestment rates covering
almost the full range of possible values from zero to one-hundred percent. Even after
adjusting for underwritten DRPs, on average forty-one per cent of the dividend is
reinvested.
Hasseldine (1988) suggests four categories of factors affected reinvestment; investor, plan,
company, and economic. As economic factors are likely to affect all DRP firms similarly
researchers have tended to concentrate on the first three to explain differences in
reinvestment levels across firms.
As investor factors are difficult to measure directly, evidence of dividend clienteles is used to
proxy investor factors. Baker and Johnson (1988) found there were different clienteles for
utilities and non-utilities. Utilities had a higher proportion of dividends reinvested but a
smaller percentage of shareholders reinvesting. As utilities tend to have more institutional
investors, this suggests an institutional clientele for utility DRPs, and a small shareholder
clientele for non-utility DRPs. Lyroudi (1999) found this was supported by a negative, though
only weakly significant, relationship between reinvestment and the dividend payout ratio.
Another approach was taken by Chiang, et al. (2006) who surveyed U.S. mutual fund
managers, but found the majority of fund managers were indifferent to the DRP and did not
use them to reinvest.
The main plan design factor is the discount offered on new shares. Baker and Johnson
(1988) found that this discount was linked to the amount reinvested, but not to the number
of reinvesting shareholders. Todd and Domain (1997) and Lyroudi (1998) also found higher
reinvestment levels for discount DRPs. For Australian firms, Wills (1989) found reinvestment
[11]
positively related to the discount offered and the use of later pricing periods. 8 Canil and
Rosser (2008) found that Australian firms offer higher DRP discounts when they have been
issuing equity. This could indicate that management use these discounts to encourage
higher levels of reinvestment.
A firm’s financial characteristics may also affect the reinvestment rate. Pettway and Malone
(1973) found three significant financial variables. The number of shares outstanding
(representing size) and the debt to equity ratio were both positively related to
reinvestment. The P/E ratio’s coefficient was negative which reflected shareholder
reluctance to purchase overpriced shares. Lyroudi (1999) found size had a negative
coefficient and leverage had a positive relationship with reinvestment. This last result was
contrary to her expectation that the new equity issue would decrease leverage but is
consistent with Pettway and Malone (1973).
These reinvestment studies have identified a number of variables connected to the
reinvestment ratio, but the opportunities to link them to the dividend literature are limited.
Where institutional reinvestment clienteles have been identified, we can see the possibility
of improved monitoring. However, few of the financial variables used in the DRP literature
can be directly identified as free cash flow variables. So while reinvestment plans would
seem to undermine the role of dividends in reducing excess free cash flow, this has not been
widely tested. Lyroudi (1999) included a variable to measure excess free cash but it was not
found significant. Also, when examining DRP firm characteristics, Saporoschenko (1998)
found no evidence that DRP firms had excess free cash flow.
In summary, while the issue of dividend relevance has not been settled, extending the
arguments for relevance to incorporate reinvestment plans shows that if dividends affect
value then reinvestment will also affect value. Therefore it is important to consider how
much control over dividend size management has in the presence of a reinvestment plan.
The retention of a suitable level of cash depends on the combination of management and
shareholder reinvestment decisions. Management must rely on shareholders who will be
8 The issue price for DRP and BSP shares is usually calculated by the weighted average price of shares traded over a five day period. Care should be taken with respect to the pricing period as Wills notes that the low discount plans were all using an earlier pricing period.
[12]
less informed about a firm’s needs and prospects. A better understanding of the interactions
between dividend policy and reinvestment requires knowledge of the determinants of
reinvestment and the effective size of dividends after reinvestment.
3 Data and Methodology
The population of interest is all non-financial Australian resident firms listed on the ASX at
any time between 1992 and 2008. While some earlier DRP data is available it is incomplete,
so an earlier start date would increase survivorship bias The decision to end the sample in
2008 minimises the effects of the financial crisis on results. Financial companies are
excluded due to their highly leveraged capital structure and the ownership concentration
limits imposed on Australian banks.
The primary data source for company announcements and accounting data is the
Morningstar DatAnalysis database. Market data for calculating beta, firm size and the
market value of equity is taken from the AGSM-CRIF database. Information on DRP and BSP
use was taken from company announcements and annual reports. Firms are only included in
the final sample if all required data is available for the financial year.
Like prior reinvestment research, this paper examines the determinants of the reinvestment
rate, but also considers the inter-relationships between reinvestment and dividend size.
Three primary measures of dividend size are used; yield, payout ratio, and the dividend to
total assets ratio. For reinvestment firms both the declared dividend and effective dividend
after reinvestment are examined. Declared dividend measures the potential cash dividend if
no reinvestment occurs. Effective dividend measures the actual cash distribution after
shareholder reinvestment through DRP or BSP.
To examine the determinants of reinvestment rates and dividend size, cross sectional panel
models were estimated. Most models are estimated using ordinary least squares (OLS), but
as the reinvestment rate is an explanatory variable in the yield model and yield is an
explanatory variable in the reinvestment rate model we also estimate two-stage least
squares (2SLS) models. In these we treat declared dividend size and the reinvestment rate
as simultaneously determined endogenous variables. With this approach the expected
[13]
reinvestment rates, or dividend size is estimated in the first stage. These expected values
are subsequently used as explanatory variables in the second stage. Panel ordinary least
squares models are also estimated for reduced form versions of the two-stage models.
The regression models’ independent variables include accounting ratios, growth
opportunities, ownership structure, franking credits, reinvestment plan design, risk, and
size. The three accounting ratios are return on equity (ROE), the current ratio (CURRENT),
and the debt to assets ratio (LEVERAGE). Growth opportunities are proxied by the book
value of assets to the market value of assets. Ownership structure is measured by the
natural log of the total number of shareholders (SHER). Tax is represented by the
percentage each dividend is franked (FRANKING). The main reinvestment plan variable is the
discount to market price at which the new shares are issued (DISC). As Wills (1989) found
the timing of the pricing period was relevant we use a dummy variable coded zero for
pricing periods before, and one for pricing after, the close of books (PP). We also use
dummy variables to signify firms with a bonus share plan (BSPDUM), and when the
reinvestment plan is underwritten (UNDDUM). The two control variables are firm size and
risk. Following Fama and French (2001), size is the percentage of listed firms smaller than
the subject firm (SIZE). Beta is included as a measure of each firm’s systematic risk (BETA).
Betas were calculated for each firm using monthly returns data for three years prior to the
balance date. This requirement of three years’ market data to calculate beta excludes newly
listed companies from the sample. Beta is the standard risk measure in dividend and DRP
models.
4 Results and Analysis
This section summarises and analyses the results of empirical testing the reinvestment rates
and dividend size for listed Australian firms. Table 1 reports descriptive statistics for the
three dividend size measures. Table 2 presents four models of the reinvestment rate,
updating and extending prior research on DRP participation in Australia and the United
States. Tables 3 to 5 report the estimated regression coefficients for the dividend size
models. Table 3 reports on yield models Table 4 reports on payout ratio model and table 5
reports on dividend to total assets ratio models. Tables 3 to 5 include dividend size models
incorporating reinvestment, which is the main innovation in this paper. The payout ratio
[14]
models have smaller samples as observations with negative earnings are excluded and
contain fewer significant variables. Accordingly the discussion here concentrates on the
yield and dividend to asset models.
The regression models for reinvestment rate, dividend yield, dividend payout ratio and
dividend to total assets ratio all have satisfactory R2 values and highly significant F-statistics.
Durban Watson statistics indicate autocorrelation is not an issue for these models.
Likelihood tests support the use of fixed effect panel models.
The descriptive statistics for dividend size presented in Table 1 show that the average
reinvestment rate of forty-one percent has a considerable effect on dividend size. This can
be seen in differences between declared and effective dividends. This table also shows the
differences between dividends from cash only firms and DRP firms. Curiously the latter
group has smaller dividends even before the effects of reinvestment are taken into account.
F-tests and t-tests, not reported here, confirmed that the reinvestment firm dividends were
significantly smaller and less variable than dividends from non-DRP firms. It is also
interesting to note that with the DRP firm sample there is more variation in declared
dividends than effective dividends after reinvestment.
Table 2 reports estimated regression coefficients for reinvestment rate models. Financial
characteristics and dividend size have little impact on reinvestment. Instead plan design
features dominate the models. As expected offering shares at a discount or including a BSP
option increases reinvestment. Underwritten plans also receive more reinvestment from
existing shareholders. The negative coefficient on the pricing period variable is unexpected
as it reverses Wills (1989) finding higher reinvestment in firms using a later pricing period. It
could be due to pricing periods before the close of books provides shareholders with a
better opportunity to estimate the issue price, reducing uncertainty and thereby
encouraging reinvestment.
If firms use DRPs to distribute more franking credits, and want shareholders to support this
tactic, then both dividend size and franking should have a positive and significant
relationship with the reinvestment rate. However, these variables proved not significant.
[15]
The financial characteristics were more likely to be significant if institutional investors had a
larger role in determining the reinvestment rate. The lack of significance for those variables
raises the risk that reinvestment undermines the potential for dividends to help control
agency costs.
The OLS models for DRP firm yield, table 3, and the dividend to asset ratio, table 5, show
plan design features are statistically significant. Pricing period’s significant and positive
coefficient in the effective dividend models is consistent with its negative coefficient in the
reinvestment rate models. Firms with later pricing periods have less reinvestment and
therefore larger effective dividends. The two-stage least squares yield and dividend to assets
models also show that the expected reinvestment has a positive and significant coefficient.
This suggests firms declare larger dividends expecting that cash flow required will be
reduced by higher levels of reinvestment.
Given the effect of reinvestment on dividend size, and the finding that financial variables
have almost no role in determining reinvestment levels, there is a risk that DRPs undermine
the use of dividends to reduce free cash flow. Profitability is expected to be positively
related to dividend size but for DRP firms this relationship only holds for the declared
dividends, not effective ones. The more profitable DRP firms are able to pay larger
dividends, and declaring them accordingly. However, after reinvestment they are not
paying larger dividends than less profitable firms, indicating potential free cash flow
problems. Jensen (1986) argued that debt is better than dividends for reducing free cash
flow, and the results show DRP firms have an inverse relationship between leverage and
dividend size, when dividends are measured by yield or dividend to assets. This relationship
is found for both declared and effective dividends. So debt and dividends are substitute cash
flow controls even in the presence of reinvestment. Results are mixed regarding growth.
The yield models show more mature firms have higher declared and effective dividends, as
expected. Dividend to assets models instead show mature firms with lower declared and
effective dividends.
The other way dividend policy can help control agency problems is through attracting a
monitoring clientele. The number of shareholders is the only variable significant across all
[16]
dividend size models, but is only significant in the two-stage least squares reinvestment rate
model. Coefficient signs are inconsistent in the non-DRP firm models, but are consistently
positive in DRP firm ones. DRP firms with more shareholders declare larger dividends. As the
number of shareholders rarely affects reinvestment rates, this carries through with a
positive relationship between the number of shareholders and effective dividend size. If
dividends substitute for institutional monitoring, then reinvestment is not changing the
relationship between ownership and dividend size.
5 Conclusions
Under dividend imputation Australian firms need to pay dividends in order to distribute
valuable franking credits to their shareholders. DRPs allow firms to distribute more franking
credits, relative to cash distributed, than they could with an equivalent cash only dividend.
Despite this need, the data indicates that firms are not using DRPs to declare large
dividends. Instead, firms offering reinvestment plans typically have lower declared dividend
yields than non-reinvestment firms. With strong reinvestment levels the effective yields paid
become even lower, raising the risk of excess free cash flow retention and the associated
agency costs. Overall the evidence only provides limited support for the use of dividends as
an agency cost control mechanism when firms use reinvestment plans. However, if
Australian firms are using dividends to control free cash flow, then dividend reinvestment
does not appear to change fundamentally the relationship between financial or ownership
variables and dividend size. These results indicate that management retains control over
distribution size even after shareholder reinvestment. Reinvestment rates reflect plan
design factors, which are determined by management. As long as managers are aware how
plan design affects reinvestment they can modify plans to influence reinvestment levels.
ReferencesAnderson, Ray H., 1986, Dividend reinvestment and Australian companies, Professional
Administrator 38, 46-47.Baker, H. Kent, and Martha C. Johnson, 1988, Dividend reinvestment plans after two
decades, Financial Management 17, 7-8.
[17]
Bellamy, David E., 1994, Evidence of imputation clienteles in the Australian equity market, Asia Pacific Journal of Management 11, 275-287.
Boehm, Thomas P., and Ramon P. DeGennaro, 2011, A discrete choice model of dividend reinvestment plans: Classification and prediction, Managerial and Decision Economics 32, 215-229.
Bruckner, Kris, Nigel Dews, and David White, 1994. Capturing Value From Dividend Imputation (McKinsey & Company, Sydney).
Canil, Jean M., and Bruce A. Rosser, 2008, Dividend reinvestment plans and the reinvestment discount, JASSA 26-30.
Chan, Keith K. W., Damien W. McColough, and Michael T. Skully, 1993, Australian tax changes and dividend reinvestment announcement effects: A pre- and post-imputation study, Australian Journal of Management 18, 41-62.
Chan, Keith K. W., Damien W. McColough, and Michael T. Skully, 1996, Australian dividend reinvestment plans: An event study on discount rates, Applied Financial Economics 6, 551-561.
Chiang, Kevin, George M. Frankfurter, Arman Kosedag, and Bob G. Wood Jr, 2006, The perceptions of dividends by professional investors, Managerial Finance 32, 60-81.
Dubofsky, David A., and Leonard Bierman, 1988, The effect of discount dividend reinvestment plan announcements on equity value, Akron Business and Economic Review 19, 58-68.
Easterbrook, Frank H., 1984, Two agency-cost explanations of dividends, American Economic Review 74, 650-659.
Easton, Stephen, 1991, Earnings and dividends: Is there an interaction effect?, Journal of Business Finance and Accounting 18, 255-266.
Fama, Eugene F., and Kenneth R. French, 2001, Disappearing dividends: Changing firm characteristics or lower propensity to pay?, Journal of Financial Economics 60, 3-43.
Fredman, Albert J., and John R. Nichols, 1982, Sizing up new capital dividend reinvestment plans, California Management Review 24, 77-84.
Hasseldine, D. J., 1988, Dividend reinvestment schemes in New Zealand, Asian-Pacific Tax and Investment Bulletin 6, 515-525.
Hathaway, Neville, and Bob Officer, 2004, The Value of Imputation Tax Credits, (Capital Research Pty Ltd, Melbourne).
He, Wei, 2009, Dividend reinvestment plans, in H. Kent Baker, ed.: Dividends and Dividend Policy (John Wiley & Sons, Hoboken, NJ).
Henry, Darren, 2011, Ownership structure and tax-friendly dividends, Journal of Banking and Finance 35, 2747-2760.
How, J. C. Y., C. S. Teo, and H. Y. Izan, 1992, The interaction effect of earnings and dividend announcements on share price: Australian evidence, Managerial Finance 18, 22-33.
Jensen, Michael C., 1986, Agency costs of free cash flow, corporate finance, and takeovers, American Economic Review 76, 323-329.
Jensen, Michael C., and William H. Meckling, 1976, Theory of the firm: Management behavior, agency costs and ownership structure, Journal of Financial Economics 3, 305-360.
Larkin, Patrick, Baeyong Lee, and Abdoul Wane, 2005, Repurchase dividend reinvestment plans (Repurchase DRIPS), Journal of Business and Economics Research 3, 55-62.
Lease, Ronald C., Kose John, Avner Kalay, Uri Loewenstein, and Oded H. Sarig, 2000. Dividend Policy: Its Impact on Firm Value (Harvard Business School Press, Boston).
Lintner, John, 1956, Distribution of incomes of corporations among dividends, retained earnings and taxes, American Economic Review 46, 97-113.
[18]
Lyroudi, Katerina, 1998, Dividend reinvestment plans with a price discount: theoretical and empirical evidence, Corporate Finance Review 3, 6-20.
Lyroudi, Katerina, 1999, Significance and implications of participation in dividend reinvestment plans: Empirical research findings, Journal of Financial Management and Analysis 12, 37-53.
Marks, Bernard, 1990. Corporate Taxation in Australia: Distributions and Imputation (CCH Australia, Sydney).
Miller, Merton H., and Franco Modigliani, 1961, Dividend policy, growth, and the valuation of shares, Journal of Business 34, 411-433.
Myers, Stewart C., 1977, Determinants of corporate borrowing, Journal of Financial Economics 5, 147-175.
Pattenden, Kerry, and Garry Twite, 2008, Taxes and dividend policy under alternative tax regimes, Journal of Corporate Finance 14, 1-16.
Pettway, Richard H., and R. Phil Malone, 1973, Automatic dividend reinvestment plans of nonfinancial corporations, Financial Management 2, 11-18.
Rozeff, Michael S., 1982, Growth, beta and agency costs as determinants of dividend payout ratios, Journal of Financial Research 5, 249-259.
Saporoschenko, Andrew, 1998, Do dividend reinvestment plans contribute to industrial firm value and efficiency?, Financial Services Review 7, 273-289.
Scholes, Myron S., and Mark A. Wolfson, 1989, Decentralized investment banking: The case of discount dividend-reinvestment and stock-purchase plans, Journal of Financial Economics 24, 7-35.
Skully, Michael T., 1982. Dividend Reinvestment Plans: Their Development and Operations in Australia and the United States (Committee for the Economic Development of Australia, Melbourne).
Todd, Janet M., and Dale L. Domain, 1997, Participation rates of dividend reinvestment plans: Differences between utility and nonutility firms, Review of Financial Economics 6, 121-135.
Wills, Ann, 1983, Dividend reinvestment plans: A new source of equity capital, Australian Accountant 53, 611-14.
Wills, Ann, 1989, A decade of dividend reinvestment, Australian Accountant 59, 65-73.
[19]
Table 1 Final Dividend Size
Panel A: YieldsNon Reinvesting Firms Reinvesting Firms
Declared EffectiveMean 3.1832% 2.7474% 1.6360%Maximum 53.5714% 11.0507% 9.0083%Minimum 0.1124% 0.1140% 0.0017%Std. Dev. 3.0258% 1.5313% 1.1549%Observations 1988 630 630Panel B: Payout Ratios
Non Reinvesting Firms Reinvesting FirmsDeclared Effective
Mean 0.5202 0.4276 0.2466Maximum 36.3636 4.5000 2.0078Minimum 0.0190 0.0300 0.0002Std. Dev. 1.4518 0.3389 0.1939Observations 1942 615 615Panel C: Dividend to Assets Ratios
Non Reinvesting Firms Reinvesting FirmsDeclared Effective
Mean 0.0309 0.0222 0.0133Maximum 0.4647 0.1382 0.1127Minimum 0.0001 0.0003 0.0000Std. Dev. 0.0331 0.0151 0.0113Observations 1988 630 630
Descriptive statistics for dividend size.
[20]
Table 2 Determinants of the Reinvestment Rate
Model Type:
Ordinary Least Squares
Two-Stage Least Squares
Two-Stage Least Squares
Two-Stage Least Squares
Variable
Coefficient
Probability
Coefficient
Probability
Coefficient
Probability
Coefficient
Probability
Constant0.508
5 0.0190.446
3 0.0340.726
1 0.0000.442
8 0.014
YIELD
-0.000
8 0.972
PAYOUT
-0.920
0 0.144
DIVTA1.476
1 0.391
ROE0.067
6 0.117
CURRENT
-0.006
2 0.020 *
LEVERAGE
-0.004
0 0.950
BVAMVA
-0.035
9 0.556
LOG(SHERS)
-0.025
0 0.063
-0.028
8 0.0720.017
2 0.629
-0.029
4 0.020 *
FRANKING0.010
4 0.5430.011
3 0.539
-0.013
4 0.7310.011
3 0.527
DISC0.881
5 0.005**
0.8869 0.006
**
0.7223 0.038 *
0.8739 0.006
**
PP
-0.126
5 0.011 *
-0.116
9 0.021 *
-0.121
0 0.016 *
-0.115
1 0.018 *
BSPDUM0.119
9 0.003**
0.1207 0.001
**
0.0777 0.216
0.1120 0.009
**
UNDDUM0.051
0 0.012 *0.044
1 0.010**
0.0133 0.664
0.0431 0.030 *
BETA0.018
6 0.0970.019
0 0.092
-0.009
2 0.7030.017
2 0.147
SIZE0.108
6 0.6050.184
1 0.483
-0.306
4 0.4050.156
2 0.353Adjusted 0.540 0.539 0.339 0.532
[21]
R-squared 7 2 3 3
F-statistic4.349
9 0.000**
4.3730 0.000
**
4.4698 0.000
**
4.3891 0.000
**
Durbin-Watson
1.9318
1.9158
1.9162
1.9242
Observations 630 630 615 630*,** Significant at the 5% and 1% levels respectively.Fixed effect cross sectional panel models of the reinvestment rate. Instruments used in the two-stage least squares model are: Constant, ROE, CURRENT, LEVERAGE, BVAMVA, LOG(SHERS), FRANKING, DISC ,PP, BSPDUM, UNDDUM, BETA, SIZE. All variables are as described in the text.
[22]
Table 3 Determinants of Final Dividend Yield
Sample: Non DRP Firms DRP Firms DRP Firms DRP FirmsDependent Variable: Yield Declared Yield Declared Yield Effective YieldModel Type: OLS OLS 2SLS OLS
Variable
Coefficient
Probability
Coefficient
Probability
Coefficient
Probability
Coefficient
Probability
Constant3.207
2 0.0043.930
7 0.0002.810
2 0.0491.989
0 0.087
ROE0.434
2 0.8090.578
9 0.000**
0.4584 0.008
**
0.0409 0.803
CURRENT
-0.005
5 0.0870.009
0 0.7050.024
4 0.3660.031
8 0.144
LEVERAGE1.148
8 0.014 *
-1.277
1 0.004**
-0.896
4 0.033 *
-0.790
6 0.023 *
BVAMVA1.932
6 0.002**
2.1731 0.000
**
2.2326 0.000
**
1.4806 0.000
**
LOG(SHERS)
0.1201 0.012 *
0.1689 0.005
**
0.2126 0.006
**
0.1553 0.007
**
FRANKING
-0.084
5 0.776
-0.171
3 0.254
-0.211
0 0.218
-0.159
4 0.247REINVESTMENT
2.3783 0.009
**
DISC3.156
4 0.047 *0.710
6 0.616
PP0.246
4 0.2380.456
5 0.002**
BSPDUM0.406
5 0.001**
-0.058
8 0.712
UNDDUM0.309
0 0.015 *0.030
0 0.815
BETA
-0.045
9 0.5070.002
2 0.978
-0.045
7 0.630
-0.031
3 0.631
SIZE
-4.227
2 0.000**
-5.107
2 0.000**
-5.202
1 0.000**
-3.216
4 0.000**
Adjusted R-squared
0.4702
0.7124
0.6553
0.6257
F-statistic4.143
9 0.000**
8.0509 0.000
**
8.0664 0.000
**
5.7584 0.000
**
Durbin-Watson
2.0370
2.3041
2.2536
2.2024
Observatio 1988 630 630 630
[23]
ns*,** Significant at the 5% and 1% levels respectively.Fixed effect panel models of the dividend yield. All variables are as described in the text.
[24]
Table 4 Determinants of Final Dividend Payout Ratio
Sample: Non DRP Firms DRP Firms DRP Firms DRP FirmsDependent Variable: Payout Ratio
Declared Payout Ratio
Declared Payout Ratio
Effective Payout Ratio
Model Type: OLS OLS 2SLS OLS
Variable
Coefficient
Probability
Coefficient
Probability
Coefficient
Probability
Coefficient
Probability
Constant1.302
3 0.0080.318
6 0.3240.339
0 0.3000.252
6 0.203
ROE
-0.927
7 0.002**
-0.053
4 0.307
-0.059
4 0.289
-0.046
6 0.122
CURRENT
-0.000
5 0.7080.002
2 0.7820.002
1 0.7540.003
1 0.506
LEVERAGE0.023
3 0.852
-0.135
4 0.552
-0.181
3 0.340
-0.087
2 0.456
BVAMVA
-0.160
8 0.3260.047
7 0.4190.058
7 0.3110.026
9 0.470
LOG(SHERS)
-0.042
4 0.048 *0.071
7 0.001**
0.0723 0.005
**
0.0514 0.001
**
FRANKING0.041
9 0.484
-0.022
8 0.679
-0.020
7 0.696
-0.016
1 0.579REINVESTMENT
0.0750 0.872
DISC0.650
9 0.280
-0.086
1 0.783
PP
-0.044
3 0.1110.021
6 0.388
BSPDUM
-0.030
6 0.755
-0.052
8 0.172
UNDDUM
-0.032
3 0.507
-0.033
5 0.163
BETA
-0.026
1 0.157
-0.029
7 0.156
-0.030
9 0.158
-0.024
7 0.046 *
SIZE
-0.285
1 0.299
-0.570
1 0.007**
-0.607
3 0.007**
-0.483
2 0.001**
Adjusted 0.288 0.268 0.273 0.352
[25]
R-squared 1 3 7 7
F-statistic2.420
5 0.000**
2.0423 0.000
**
2.0748 0.000
**
2.5492 0.000
**
Durbin-Watson
2.1184
2.1863
2.1918
2.1805
Observations 1942 615 615 615*,** Significant at the 5% and 1% levels respectively.Fixed effect panel models of the dividend payout ratio. All variables are as described in the text.
[26]
Table 5 Determinants of Final Dividend to Total Assets Ratio
Sample: Non DRP Firms DRP Firms DRP Firms DRP FirmsDependent Variable:
Dividend to Assets Ratio
Declared Dividend
Declared Dividend
Effective Dividend
Model Type: OLS OLS 2SLS OLS
Variable
Coefficient
Probability
Coefficient
Probability
Coefficient
Probability
Coefficient
Probability
Constant0.072
4 0.0000.061
8 0.0000.050
0 0.0010.031
6 0.001
ROE0.033
3 0.000**
0.0041 0.010
**
0.0030 0.194
0.0000 0.986
CURRENT
-0.000
1 0.0900.000
4 0.1890.000
6 0.1250.000
6 0.008**
LEVERAGE
-0.024
6 0.000**
-0.041
2 0.000**
-0.037
2 0.000**
-0.023
3 0.000**
BVAMVA
-0.027
6 0.000**
-0.023
9 0.000**
-0.023
6 0.000**
-0.015
0 0.000**
LOG(SHERS)
0.0001 0.906
0.0024 0.000
**
0.0029 0.001
**
0.0021 0.000
**
FRANKING
-0.001
4 0.494
-0.000
9 0.570
-0.001
3 0.414
-0.000
4 0.686REINVESTMENT
0.0219 0.036 *
DISC0.007
6 0.674
-0.018
3 0.153
PP0.002
4 0.0710.003
9 0.006**
BSPDUM0.005
3 0.000**
0.0008 0.492
UNDDUM0.002
9 0.000**
0.0004 0.517
BETA
-0.001
7 0.019 *0.001
1 0.1650.000
6 0.4840.000
7 0.292
SIZE
-0.015
3 0.070
-0.031
4 0.000**
-0.032
0 0.000**
-0.019
1 0.008**
Adjusted R-squared
0.6945
0.7757
0.7206
0.7508
F-statistic9.052
1 0.00010.84
01 0.00010.77
63 0.000**
9.5736 0.000
**
Durbin-Watson
2.4532
2.2211
2.1398
2.1216
[27]
Observations 1988 630 630 630*,** Significant at the 5% and 1% levels respectively.Fixed effect panel models of the Dividend to Assets Ratio. All variables are as described in the text.
[28]
