1
Share Repurchase Regulation and Corporate Payout Practices in the UK
by
Elisabeth Dedman*
University of Surrey, UK
Thanamas Kungwal
University of Keele, UK
and
Gilad Livne
University of Exeter, UK
Acknowledgements: this paper has benefitted from the advice and colleagues of many
colleagues, including seminar attendants at Bath Management School, Keele Management
School, and Nottingham University Business School. We are also grateful to comments on
previous versions of this work from the audience members at BAFA and EAA. Special
thanks go to Andrew Stark who has offered much advice over several iterations of the paper.
Remaining errors belong to the authors.
*Corresponding author: [email protected]
2
Share Repurchase Regulation and Corporate Payout Practices in the UK
ABSTRACT
We investigate whether reforms to UK regulations regarding the
repurchase of a company’s own shares has had an effect on the distribution
policies of UK listed firms. Originally, any repurchased shares had to be
cancelled but this requirement was relaxed following reforms in late 2003,
after which companies could hold them as treasury shares. In late 2009, a
further liberalising reform took place, with the lifting of a 10% ceiling on
the proportion of issued shares which could be held as treasury shares. Our
study spans these three regulatory time periods to investigate (a) the
determinants of both dividends and share repurchases from 1994-2013,
and (b) whether the payout practices of UK public companies have
changed following the repurchase reforms. As well as reporting the results
of comprehensive models of the determinants of payout choice, we are able
to conclude that: (i) dividends are not disappearing in the UK, though the
number of firms paying them has reduced; (ii) share repurchases gained in
importance following the first reform but reduced in importance during the
financial crisis; (iii) in contrast to what is observed in the US, share
repurchases are not replacing dividends in the UK.
JEL Codes and Keywords: G35, Payout Policy (Dividends and
Repurchases); G38, Regulation;
3
Share Repurchase Regulation and Corporate Payout Practices in the UK
1. Introduction
The US literature documents a strong trend away from dividends over the last forty years,
with a marked increase in the use of stock repurchases as a corporate payout vehicle over a
similar time period. Pure dividend payers, observed to represent only 6.8% of listed
companies in 2005, have been said to be ‘largely extinct’ in this environment (Skinner,
2008), though a reappearance of dividends per se is documented by Julio and Ikenberry
(2004). Whilst there are many similarities between the US and UK capital markets, the
pattern of payouts in the UK has historically differed to that in the US (Renneboog and
Trojanowski, 2011) with repurchases representing a much smaller proportion of shareholder
distributions in the UK, and conveying a much weaker signal than in the US (Andriosopolous
and Lasfer, 2015). Earlier literature argues that the infrequent use of share repurchases in the
UK is due to the onerous regulation of the practice relative to the US (Rau and Vermaelen,
2002). Dhanani and Roberts (2009) provide a summary of these restrictions and show that
the UK regulation was amongst the most stringent globally. However, more recently, the
regulation surrounding share repurchases in the UK has been relaxed in some important
regards. The Finance Act 2003 abolished the previous requirement for repurchased shares to
be cancelled, while a 2009 amendment to the Companies Act 2006 removed the 10% ceiling
on the number of treasury shares allowed to be held by a company. European regulations also
changed late in 2003, providing safe harbour conditions for member state companies wishing
to trade in their own shares.
Prior UK literature on payout practices has mainly focused on the period prior to these
reforms. This study therefore investigates the effects of these regulatory changes on the
payout practices of UK-listed firms, examining a large sample of companies over three
4
regulatory time periods: 1994-2003; 2004-2009; and 2010-2013. Given the tax advantages of
share repurchases, as well as perceived differences in investor expectations of regular
distributions by this method (compared to dividends), reported by US literature, we expect to
observe an increase in the UK use of share repurchases following each liberalising reform. It
will be interesting to investigate whether UK payout practice is now more similar to that of
the US, with the relaxation of UK regulations leading to repurchases replacing dividends.
Our main findings are as follows. First, in line with prior UK literature, we observe a decline
in the propensity of firms to pay out dividends in our earliest time period and up to 2006, the
end of the sample period for prior studies. However, when we extend the period of study
beyond this time, we see the decline slow then largely level off, with an average of 50.3% of
firms distributing regular dividends in our second time period, and 46.5% of firms using this
payout channel in our latest period. Whilst dividends show little sign of disappearing in the
UK after 2004, share repurchase activity increases significantly following the initial reforms.
Although only 8% of sample firms repurchased shares in the period when they had to be
cancelled, a quarter of sample firms were buying back their own shares by 2008, following
the first reform, but prior to the second reform which lifted the 10% ceiling on treasury
shares. The regulatory changes relating to share repurchases do seem to have had an effect
on UK payout policy, though practice still differs markedly from what is observed in the US,
with more dividends and fewer repurchases in the UK. A smaller increase in the proportion
of companies repurchasing their own shares follows the 2009 reform, which is arguably the
less radical change. Although fewer firms distribute dividends over time in the UK, the
average amount paid out by dividend payers exhibits a steady positive trend; as in the US
dividends are becoming more concentrated in the UK. The average amount used to
repurchase shares is much more volatile over time, consistent with share repurchases being
used to distribute transitory excess cash.
5
Our second main contribution is our examination of the determinants of payout policy prior
to, and post-reform. We augment models established in prior literature (Fama and French,
2001; Denis and Osobov, 2008), adding variables found to be significantly related to payout
decisions by later authors. We use these more comprehensive models to investigate the
determinants of both dividends and repurchases in a sample of 19,652 firm years over our
three regulatory time periods in the UK.
We find that, consistent with earlier evidence from the US and UK, larger, more mature,
more profitable UK firms remain more likely to distribute dividends. These factors are also
positively associated with the propensity to repurchase shares. Firms with lower market to
book ratios, argued by earlier researchers (Ikenberry, Lakonishok and Vermaelen, 1995) to
indicate undervaluation, are both more likely to issue dividends and to repurchase their own
shares. Consistent with the argument that share repurchases are used to distribute non-
permanent excess cash to shareholders (Dittmar, 2000; Dixon et al., 2008; Lee and Suh,
2011; Oswald and Young, 2008), cash as a proportion of total assets is strongly positively
associated with the likelihood a firm buys back its own shares, but negatively associated with
the likelihood it issues a regular dividend. Consistent with UK research (Dixon et al., 2008)
which finds capital structure rebalancing to be a major motivation for share repurchases, we
find leverage to be significantly associated with the decision to buy back shares. In
examining the determinants of each payout channel independently, we find that the decision
to distribute dividends (repurchase shares) is significantly positively associated with the
adoption of this same payout channel in the prior year, but also with the use of the alternative
payout channel in the current year. We therefore use multinomial regressions to allow for the
payout choices to be simultaneously determined. We find that payout choices have changed
following the share repurchase reforms. Following each reform, the propensity to issue
dividends declines significantly and the propensity to repurchase shares increases
6
significantly, ceteris paribus. We therefore reject our null hypotheses that the reforms have
had no effect on the payout decisions of UK firms.
Finally, our data allow us to comment on whether, as appears to be the case in the US, share
repurchases are replacing dividends in the UK. Here, we find that the proportion of total
payouts (dividends plus repurchases) represented by repurchases follows an increasing trend
between 2003 and 2007, where a peak of 49% of total payout being made via share
repurchases is observed. However, this falls dramatically following the financial crisis, to 9%
in 2009 and 11% in 2010. Whilst some recovery is observed for the latter years of our
sample, we argue it cannot be claimed that share repurchases are replacing dividends in the
UK. Further, following a period of steady reduction in this type of firm from 1994-2005,
since 2006 there has remained a stable core of around 29% of sample firms which may be
classed as ‘pure dividend payers’ (Skinner, 2008).
The rest of this paper is structured as follows. Section 2 discusses the regulatory background
to the study, and the reforms to rules governing UK share repurchases in 2003 and 2009.
Section 3 reviews prior literature and introduces our hypotheses. In Section 4 we explain our
research design and in Section 5 we present our analyses of the data. We summarise our
findings and make our conclusions in Section 6.
2. Regulatory Background
Prior to 1981, UK legislation prevented companies from repurchasing their own shares, other
than redeemable preference shares. The prohibition was designed to prevent firms from: (a)
reducing their capital to the detriment of creditors; (b) privileging certain shareholders in off-
market transactions; (c) rigging the market for their shares; and (d) engaging in ‘greenmail’
transactions to ward off takeover threats. Changes made in the Companies Act of 1981,
7
however, allowed share repurchases as an alternative distribution channel to UK firms but
with some fairly stringent restrictions. These included: requirements for prior permissions, to
be granted with a limited life; quantitative restrictions with payments to be made from
distributable profits; public disclosure of any repurchase amount and price paid by noon the
following day, with summary figures disclosed in the annual reports; restrictions relating to
timing (to avoid insider trading); and the cancellation of all repurchased shares (Bank of
England, 1988). The Finance Bill 2003 subsequently amended the regulations to allow
companies to retain repurchased shares as treasury stock (Dixon et al., 2008). The
Companies Act 2006, Chapter 6 codifies the changed regulation, which allowed publicly
listed companies to repurchase their own, fully paid up, qualifying1 shares out of distributable
profits and to hold them as treasury shares. As a company is not allowed to own itself,
treasury shares receive no dividends, carry no voting rights, and are not included in the
calculation of weighted average number of shares for earnings per share (EPS) calculations.
The maximum aggregate nominal value of treasury shares allowable to be held was 10% of
the nominal value of the issued share capital of the company at that time (s.725 CA2006).
The 2003 rule change, which came into force in November of that year, introduced a greater
degree of flexibility for firms in the management of their capital. For example, allowing
companies to retain repurchased shares made them available for use in future transactions,
such as acquisitions (Andriosopolous and Lasfer, 2015).
Just a month later, in December 2003, the European Commission passed a regulation which
provided safe harbour to EU member firms in relation to share repurchases. The Market
Abuse Directive (MAD), introduced in January 2003 for adoption by member states by
October 2004, had introduced strict rules on market manipulation, which could have deterred
1 Here ‘qualifying’ refers to a stock being listed on at least one of: (1) the Official List; (2) AIM; (3) an official
European Economic Area market; or (4) any other regulated market.
8
firms from repurchasing their own shares. The new EC Regulation provided a set of safe
harbour conditions in relation to share repurchases meeting certain criteria, which enabled
firms to protect themselves from potential penalties arising under MAD (Siems and De
Cesari, 2012). The introduction of safe harbour provisions has been found to coincide with a
significant increase in share repurchase activity in the US (Grullon and Michaely, 2002).
In October 2009, section 725 CA2006 was deleted, removing the 10% ceiling on holding
treasury shares. Now there is no legal maximum to the holding of treasury shares, subject to
the existence of at least one non-treasury share. Listed companies remain bound by rules of
their Stock Exchange, however and, in the case of share repurchases, the London Stock
Exchange imposes restrictions. For firms with a premium listing on the Main Market, the
intention to repurchase over 15% of issued share capital of a particular class triggers the
requirement for a tender offer to all holders of that class on a pro rata basis (Scott, 2014).
This rule does not apply to AIM companies.
This study examines UK payout practices over three regulatory time periods, each with
differing, progressively liberal, rules relating to share repurchases: (1) 1994-2003; (2) 2004-
2009; and (3) 2010-2013. The next section discusses prior literature relevant to our
investigation.
3. Prior Literature
Fama and French (2001) document a steep decline in the proportion of US firms paying
dividends between 1978, when 66.5% of listed firms paid dividends, and 1999, by which time
only 20.8% of firms made such distributions. Further investigation establishes that three
main factors largely determined whether or not a dividend would be paid, with distributions
positively related to firm size and profitability, and negatively associated with measures of
9
growth potential (market-to-book ratio and asset growth). DeAngelo, DeAngelo and Skinner
(2004) confirm that the number of US industrial firms which pay dividends continues to
decline slightly beyond the end of the sample used by Fama and French (2001), reporting a
drop of 58.8% in the number of dividend payers from 1978-2000. However, the amount paid
in dividends increased in both real and nominal terms over the same time period, indicating
an increase in the concentration of dividends over time. In the 1990s, there had been a surge
in the number of US firms returning cash to shareholders via share repurchase programmes
(Julio and Ikenberry, 2004). Grullon and Michaely (2002) therefore examine the association
between share repurchases and dividends over a similar time period to Fama and French
(2001). They report that aggregate expenditure on share repurchase programmes increased
from 4.8% of aggregate earnings in 1980 to 41.8% in 2000. The number of firms
repurchasing shares also increased, from 31% in 1972 to 80% in 2000. They argue that total
cash distributions (i.e. repurchases plus dividends) remain fairly constant over their 20-year
time period but that the mix has changed, with firms favouring repurchases over time.
Further evidence of a substitutive relationship between dividends and share repurchases
emerges when they adapt the model developed in Lintner (1956)2 to predict firm dividend
payments3 and find that firms which pay lower dividends than expected spend more on share
repurchases. The market also provides supportive evidence, responding less negatively to
announcements of dividend cuts in firms which spend more on share repurchases.
Addressing one of the unanswered questions from Fama and French (2001) – why did firms
not substitute dividends earlier (given the tax advantages of share repurchases)? – Grullon
2 In the standard model below, the coefficient on lagged payout is the (negative of the) speed of adjustment
coefficient while the coefficient on earnings is the product of the target payout ratio and the speed of adjustment
coefficient (Skinner, 2008) ∆𝐷𝐼𝑉𝑡 = 𝛼0 + 𝛼1𝐸𝑡 + 𝛼2𝐷𝑡−1 + 𝜇𝑡 3 𝐸𝑅𝑅𝑂𝑅𝑡,𝑖 = [∆𝐷𝐼𝑉𝑡 = (𝛽1,𝑖 + 𝛽2,𝑖𝐸𝐴𝑅𝑁𝑡,𝑖 + 𝛽3,𝑖𝐷𝐼𝑉𝑡−1,𝑖)]/𝑀𝑉𝑡−1,𝑖
10
and Michaely (2002) provide a regulatory justification. Prior to 1982, large scale repurchase
programmes were susceptible to being viewed by the SEC as attempts at price manipulations.
In 1982, however, the SEC’s adoption of Rule 10b-18 provided safe harbour for repurchasing
firms against charges of price manipulation; the removal of this barrier corresponded with a
surge in repurchase activity and coincidental decline in dividend payments. Grullon and
Ikenberry (2000), using a sample from 1980-1999, document a declining dividend payout
ratio but a stable total payout ratio, which suggests US firms were substituting repurchases
for dividends over time.
Skinner (2008) reports evidence to support Grullon and Michaely (2002) and Grullon and
Ikenberry (2000). Examining payout policies in US firms over 26 years to 2005, he finds that
firms are increasingly using share repurchases as a vehicle to distribute cash to shareholders,
often in tandem with regular dividends. Between 1970 and 2005, the proportion of firms
paying only dividends declines from 13.2% to 6.8%, leading Skinner (2008) to assert that
‘Pure dividend payers are disappearing’ [p.587]. Using logit regressions to estimate the
probability a firm repurchases stock in any given year reveals that repurchases are
significantly positively related to profitability (ROA) and negatively related to recent (past 3
years) stock price performance. Managers buy their own stock when accounting performance
is good and the price is low. This is consistent with other US work which finds evidence to
support undervaluation as a primary motivation for share repurchases (Ikenberry et al., 1995).
This study finds repurchases by ‘value’ stock firms, i.e. those with high book-to-market
ratios, are followed by an average 45% abnormal buy and hold return over the next four
years. It seems firms recognise when they are undervalued but the market takes some time to
catch up.
11
The evidence from the US is strong and consistent. Dividends are disappearing, and
repurchases are taking their place. Several researchers, recognising that it would be
premature to assume this change is being replicated elsewhere, have investigated trends in
payout policy in other environments, including the UK.
Denis and Osobov (2008) report results from a study of the dividend practices of firms in six
developed countries: US, Canada, UK, Germany, France and Japan. Covering the time
period from 1989-2002, they look at the characteristics of dividend payers, relative to non-
payers, and whether these have changed over time, as well as investigating whether there has
been a non-US decline in the propensity to pay. They find that some of the determinants of
dividends are common across sample countries, where firms are more likely to pay dividends
if they are larger and more profitable, with a higher proportion of earned equity on their
balance sheet. The presence of growth opportunities is mixed in its effect across countries
however, with a positive association with dividends in Germany, France and Japan, but a
mixed or negative relation with dividends in the US, Canada and the UK. They then develop
a model of predicted dividends by estimating an out-of-sample logit model of the
determinants of dividends and applying the coefficients obtained to observations from 1994
onwards. Consistent with earlier US findings from Fama and French (2001), Denis and
Osobov (2008) report that fewer US firms are distributing dividends over time, and that the
difference between actual and expected grows monotonically from 1994 (3.1%) to 2002
(13.6%). Although for all other sample countries bar the UK, fewer dividends are paid than
expected, the size differences between actual and expected dividends outside the US are
scattered through time. In the UK, however, the difference between expected and actual
dividends is negative in four of the nine years tested; around half of the time, UK companies
pay more dividends than predicted by the model. Denis and Osobov (2008) conclude that
there is little evidence of a decline in the propensity to pay dividends in the UK.
12
Share repurchases are not included in the study conducted by Denis and Osobov (2008) but
are examined in a European study by von Eije and Megginson (2008). This study uses as its
sample listed companies from the 15 countries which made up the pre-2004 member states of
the European Union (EU) from 1989-2005. They report a large decline in the percentage of
UK listed firms which pay a cash dividend over their time period. They also observe a
coincident (though non-monotonic) increase in the value of share repurchases in the UK, with
values almost doubling around the time of the first repurchase regulatory reform we discuss
in this paper. The average value of total share repurchases for the two years prior to the
reform (2002 and 2003) was €23,131m; for the two years following the reform (2004 and
2005), the average was €44,827.5m. Though increases in repurchase activity were reported
for the other EU member states, the UK was by far the most aggressive purchasing regime,
with around half of all sample repurchases by value being made by UK firms.
In another international study which covers 33 countries, Fatemi and Bildik (2012) report a
general reduction in the percentage of firms distributing dividends each year. For the UK,
there was a monotonic decline from 99% in 1985 to 43% in 2006, which was not atypical of
the sample countries overall. The determinants of a cash dividend payment are relatively
stable over time, with annual logit regressions on the full sample showing that the likelihood
a firm distributes a cash dividend is positively related to firm size and earnings, and
negatively related to asset growth. In common with the US findings of Fama and French
(2001), Fatemi and Bildik (2012) find that the steep decline in propensity to pay dividends is
only partially explained by a change in the characteristics of listed firms over time. In
contrast with Denis and Osobov (2008), Fatemi and Bildik (2012) conclude that ‘the
phenomenon of disappearing dividends… … is global’. [p.677]
13
Focusing on UK firms, Renneboog and Trojanowski (2011) examine dividend and share
repurchase behaviour for a sample of firms from 1992-2004. An interesting feature of their
sample period is that it includes an important change to the UK tax system, which occurred in
1997 and which negated a strong preference for many large institutional investors (pension
funds and charities) for companies to pay out earnings as dividends rather than retaining them
in the firm. The data indicate a significant decline in the propensity of UK firms to distribute
cash dividends, with a fall from 84% of firms in 1992 to only 78% in 2004. Earlier empirical
evidence of pension funds shifting from a preference for dividends over capital gains to
indifference between the two is provided by Bell and Jenkinson (2002). The later sample of
Renneboog and Trojanowski (2011) provides contrasting results, however, with the
association between tax-exempt institutional investors and the likelihood a firm pays
dividends only significant in the post-1997 time period. A rigorous set of robustness checks
fails to find evidence of a tax-clientele effect and so a tax explanation for the declining
propensity to issue dividends is rejected for the UK, in contrast to the evidence from the US
(Julio and Ikenberry, 2004). Renneboog and Trojanowski (2011) also observe a marked
increase in the proportion of firms repurchasing shares (from 5% in 1992 to 16% in 2004)
and an increase in the proportion of companies making no distribution to shareholders (from
16% in 1992 to 22% in 2004).
Dittmar (2000) examines the determinants of share repurchases in the US. As Ikenberry et
al. (1995) show that the market-to-book (MTB) ratio provides a measure of undervaluation,
with low MTB firms earning abnormal returns in subsequent periods, Dittmar (2000)
includes MTB as an explanatory variable in his model. He also tests whether firms are using
share repurchases to move towards a desired capital structure by including leverage as an
independent variable, as well as adding cash measures to examine whether repurchases are a
way to distribute cash to shareholders. Results of annual regressions from 1977-1996
14
consistently show that the propensity to repurchase shares is positively related to firm size,
cash and cash flows, and leverage, and negatively associated with MTB. Dittmar (2000)
concludes that US firms repurchase stock to take advantage of undervaluation, to distribute
excess cash and to alter their capital structures. Having included dividend payout ratios in his
regressions, with no regular, significant association observed between these and share
repurchases, he is also able to assert that ‘repurchases do not replace dividends’ [p.354].
To investigate the motivations for share repurchases in the UK, Dixon et al. (2008) survey
the finance directors of large listed companies, including both repurchase and non-repurchase
firms. At the time of the survey, repurchased shares had to be cancelled, which is reflected in
the finding that the most popular given reason for undertaking a repurchase programme was
to attain a preferred, more highly levered, capital structure. The second most popular reason
was simply to return excess cash to shareholders. Very limited support was found for
information signalling or undervaluation by UK firms as a reason for repurchasing their own
shares. Consistent with this result, Andriosopolous and Lasfer (2015) document significantly
lower market reactions to the announcement of share repurchases by UK firms when
compared to their US counterparts. For example, the UK study finds average short run
excess returns of 1.68% compared to 3.54% reported by Ikenberry et al. (1995) in the US.
Andriosopolous and Lasfer (2015) report one of the few studies of UK payout issues with a
sample that straddles our first two regulatory time periods. Interestingly, the share price
response to UK announcements is much weaker following the Finance Act 2003 and
MAD/EC Regulations of the same year. The authors attribute this to share repurchase
announcements containing a weaker signal of firm quality post-reforms. Dhanani and
Roberts (2009) conduct a survey which is administered following the first set of repurchase
reforms in the UK and are therefore able to augment the findings of Dixon et al., (2008).
Dhanani and Roberts (2009) report that the primary motivation for share repurchases, once
15
repurchased shares could be retained, was to return excess cash to shareholders, though
optimising gearing levels remained an important consideration, as documented by Dixon et
al., (2008). Other key factors in the repurchase decision were to signal undervaluation
(though managers did not believe that any share price ‘correction’ would be immediate) and
to manage earnings per share (EPS).
We examine the following hypotheses, presented in null form:
Hypothesis 1: The 2003 change in regulations relating to share repurchases had, ceteris
paribus, no impact on the distribution policies of listed firms.
Hypothesis 2: The change in regulations in 2009, had, ceteris paribus no incremental
impact on the distribution policies of listed firms.
We also address the question of whether share repurchases are replacing dividends in the UK.
We use the literature discussed above to develop our models in order to satisfy our ceteris
paribus requirement. The next section provides details of this, our sample selection, and
other aspects of our methodology.
4. Research Design
4.1 Binomial models of the determinants of dividends and share repurchases
Both Fama and French (2001) and Denis and Osobov (2008) model the determinants of a
dividend payment as a linear function of size, market-to-book, asset growth, and profitability:
𝐷𝐼𝑉𝐷𝑈𝑀𝑡 =∝ +𝛽1𝑆𝐼𝑍𝐸𝑡 + 𝛽2𝑉𝑡
𝑇𝐴𝑡⁄ + 𝛽3
𝑑𝑇𝐴𝑡𝑇𝐴𝑡
⁄ + 𝛽4𝐸𝑡
𝑇𝐴𝑡⁄
(1)
16
Where: DIVDUM = 1 if a firm pays a dividend, 0 otherwise; SIZE is the percentage of other
sample firms with a smaller market capitalisation of the firm in that year; V/TA is the market
value of total capital divided by total assets and is a measure of growth opportunities; dTA/TA
is the percentage change in total assets over the year; E/TA is earnings before interest but
after tax, deflated by total assets. This is the starting point for our model, to which we add
other variables found to be relevant to the dividend decision.
We also propose an adaptation of their model, with an alternative left-hand side variable to
capture share repurchases: SPDUM = 1 if a firm repurchases shares, 0 otherwise, again as a
starting point for a more comprehensive model.
𝑆𝑃𝐷𝑈𝑀𝑡 =∝ +𝛽1𝑆𝐼𝑍𝐸𝑡 + 𝛽2𝑉𝑡
𝑇𝐴𝑡⁄ + 𝛽3
𝑑𝑇𝐴𝑡𝑇𝐴𝑡
⁄ + 𝛽4𝐸𝑡
𝑇𝐴𝑡⁄
(2)
We then add to equations (1) and (2) further independent variables found to be associated
with dividend distributions by other scholars. Von Eije and Megginson (2008) find an
association between firm age and dividend distributions in Europe so we include firm age
(AGE) as a measure of maturity. As both dividends and share repurchases must be funded
from distributable profits, we include retained earnings (RE) in our models. Prior research
(Dittmar, 2000; Dixon et al., 2008; Lee and Suh, 2011; Oswald and Young, 2008) further
establishes that excess cash is positively associated with share repurchase activity, and
Dhanani and Roberts (2009) find that 81% of repurchases are paid for from existing cash
balances, so we also include a measure of cash and cash equivalents (CASH), which we
deflate by total assets (CASH/TA). We add a lagged dependent variable to the models in
order to measure the persistence of dividends and share repurchases (Lee and Suh, 2011). We
add a dummy variable for the alternative payout method in order to capture whether there is
17
an association between the two types of distribution. Earlier research indicates that capital
structure adjustment is a primary motivation for share repurchases, as well as finding a strong
negative association between firm debt and the likelihood of a dividend distribution (Dixon et
al., 2008; Oswald and Young, 2008) in the UK, so we incorporate a leverage variable
(LEVERAGE). In order to test our hypotheses, we then include two dummy variables to
capture the regulatory regime: REGIME2 is a dummy variable which equals one for all
observations after 2003; REGIME3 is also a dummy variable which equals one for all
observations after 2009. Our sample period includes the global financial crisis and periods of
recession for the UK. We reflect these macroeconomic factors in our analyses by including
the calendar year measure of growth in UK gross domestic product (GGDP). Finally, we
control for industry effects by categorising our firms into one of nine industries and
incorporating eight industry dummy variables in our multivariate analyses. Our base models
for the determinants of a dividend distribution or share repurchase are therefore as presented
in equations (3) and (4), respectively:
𝐷𝐼𝑉𝐷𝑈𝑀𝑡 =∝ +𝛽1𝑆𝐼𝑍𝐸𝑡 + 𝛽2𝑉𝑡
𝑇𝐴𝑡⁄ + 𝛽3
𝑑𝑇𝐴𝑡𝑇𝐴𝑡
⁄ + 𝛽4𝐸1𝑡
𝑇𝐴𝑡⁄ + 𝛽5
𝑅𝐸𝑡𝑇𝐴𝑡
⁄ +
𝛽6𝐴𝐺𝐸 + 𝛽7𝐶𝐴𝑆𝐻𝑡
𝑇𝐴𝑡⁄ + 𝛽8𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽9𝐿𝑎𝑔𝐷𝐼𝑉𝐷𝑈𝑀 + 𝛽10𝑆𝑃𝐷𝑈𝑀 +
𝛽11𝑅𝐸𝐺𝐼𝑀𝐸2 + 𝛽12𝑅𝐸𝐺𝐼𝑀𝐸3 + 𝛽13𝐺𝐺𝐷𝑃 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦
(3)
𝑆𝑃𝐷𝑈𝑀𝑡 =∝ +𝛽1𝑆𝐼𝑍𝐸𝑡 + 𝛽2𝑉𝑡
𝑇𝐴𝑡⁄ + 𝛽3
𝑑𝑇𝐴𝑡𝑇𝐴𝑡
⁄ + 𝛽4𝐸1𝑡
𝑇𝐴𝑡⁄ + 𝛽5
𝑅𝐸𝑡𝑇𝐴𝑡
⁄ + 𝛽6𝐴𝐺𝐸
+ 𝛽7𝐶𝐴𝑆𝐻𝑡
𝑇𝐴𝑡⁄ + 𝛽8𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽9𝐷𝐼𝑉𝐷𝑈𝑀 + 𝛽10𝐿𝑎𝑔𝑆𝑃𝐷𝑈𝑀
+ 𝛽11𝑅𝐸𝐺𝐼𝑀𝐸2 + 𝛽12𝑅𝐸𝐺𝐼𝑀𝐸3 + 𝛽13𝐺𝐺𝐷𝑃 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦
(4)
18
Our next two equations substitute the binary dependent variables from (3) and (4) with the
amount disbursed to shareholders via dividends or repurchases, as follows:
log(1 + £𝐷𝐼𝑉𝑡)
=∝ +𝛽1𝑆𝐼𝑍𝐸𝑡 + 𝛽2𝑉𝑡
𝑇𝐴𝑡⁄ + 𝛽3
𝑑𝑇𝐴𝑡𝑇𝐴𝑡
⁄ + 𝛽4𝐸1𝑡
𝑇𝐴𝑡⁄ + 𝛽5
𝑅𝐸𝑡𝑇𝐴𝑡
⁄
+ 𝛽6𝐴𝐺𝐸 + 𝛽7𝐶𝐴𝑆𝐻𝑡
𝑇𝐴𝑡⁄ + 𝛽8𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽9𝐿𝑎𝑔𝐷𝐼𝑉𝐷𝑈𝑀
+ 𝛽10𝑆𝑃𝐷𝑈𝑀 + 𝛽11𝑅𝐸𝐺𝐼𝑀𝐸2 + 𝛽12𝑅𝐸𝐺𝐼𝑀𝐸3 + 𝛽13𝐺𝐺𝐷𝑃 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦
(5)
log (1 + £𝑆𝑃𝑡) =∝ +𝛽1𝑆𝐼𝑍𝐸𝑡 + 𝛽2𝑉𝑡
𝑇𝐴𝑡⁄ + 𝛽3
𝑑𝑇𝐴𝑡𝑇𝐴𝑡
⁄ + 𝛽4𝐸1𝑡
𝑇𝐴𝑡⁄ + 𝛽5
𝑅𝐸𝑡𝑇𝐴𝑡
⁄
+ 𝛽6𝐴𝐺𝐸 + 𝛽7𝐶𝐴𝑆𝐻𝑡
𝑇𝐴𝑡⁄ + 𝛽8𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽9𝐷𝐼𝑉𝐷𝑈𝑀
+ 𝛽10𝐿𝑎𝑔𝑆𝑃𝐷𝑈𝑀 + 𝛽11𝑅𝐸𝐺𝐼𝑀𝐸2 + 𝛽12𝑅𝐸𝐺𝐼𝑀𝐸3 + 𝛽13𝐺𝐺𝐷𝑃 + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦
(6)
4.2 Multinomial models of the determinants of dividends and share repurchases
As the results of testing Equations (3) to (6) will reveal, there is evidence that dividends and
share repurchases are simultaneously determined. We therefore follow the method employed
by Renneboog and Trojanowski (2011) and estimate multinomial logit regressions of payout
choice, allocating observations to payout types as follows: no distribution is coded as
PAYTYPE0; dividend only is PAYTYPE1; share repurchase only is PAYTYPE2; and both
dividends and repurchases are coded as PAYTYPE3. These classifications are consistent with
those employed by Renneboog and Trojanowski (2011). The multinomial models explain
both the likelihood that a firm makes a distribution and the factors affecting its choice of a
19
particular payout method. We use the same independent variables as in our earlier tests
(Equations 3-6). All the variables used in our tests are described in Table 1.
[TABLE 1 HERE]
4.3 Sample and data
Our sample is UK, non-financial firms from 1994-2013. We collect firm-level data from
Datastream via Worldscope, with annual gross domestic product data (GDP) taken from the
Office of National Statistics (ONS). After deleting observations with incomplete data, we are
left with a sample of 19,652 observations. For our pre-change time period, up to the end of
2003, we have 10,154 observations; for our second time period, when repurchased shares could
be kept as treasury shares up to a limit of 10%, there are 6,235 observations; and for our final
regulatory period, from 2010 on, when the cap on treasure shares had been lifted, there are
3,263 observations. Having generated an initial sample of 19,652 firm-year observations with
all the required data, we winsorise continuous variables at 1% and 99% on an annual basis to
mitigate the effects of outlying observations. For variables with many zeros at the bottom end
of the distribution, we only winsorise at 99%. The next section details our analysis of the data.
5. Analysis
5.1 Descriptive Analysis
The first part of our analysis is a simple observation of trends over time, which we compare
to earlier research from the UK and other countries. Table 2 presents descriptive statistics for
our sample firms, split into observations from each regulatory period: 1994-2003 (n=10,154);
2004-2009 (n=6,235); and 2010-2013 (n=3,263). It can be seen from the GDP growth
measure that the first period was the most economically healthy in the UK. The second
20
period includes the recession years of 2008 and 2009 and, unsurprisingly, is the worst time
for economic growth; some recovery is observed in the latest time period. Panel A contains
information about the payout practices of sample firms in each of these periods. DIVDUM is
the proportion of firms which distribute a dividend in the time period – a practice which is
clearly in decline, as observed in earlier work on the UK (Denis and Osobov, 2008;
Renneboog and Trojanowski, 2011). Prior US dividend literature informs us that managers
are reluctant to cut dividends (e.g. Lintner, 1956; Brav, Graham, Harvey and Michaely, 2005)
but we can see a reduction over time in the proportion of UK companies which pay dividends
for two consecutive years (DIVDUM2YR), suggesting dividend payments are becoming less
‘sticky’ in the UK. Over our sample periods, we see substantial growth in the number of
firms which repurchase their own shares (SPDUM), particularly following the first reforms,
which allowed UK companies to retain repurchased shares as treasury shares rather than
cancel them. There is also growth in the proportion of firms repurchasing shares in two
consecutive years, which appears at odds with US findings suggesting repurchases are used
as a vehicle to pay out transitory earnings increases (Skinner, 2008). However, we note that
share repurchase programmes may span two fiscal years so we offer no strong conclusions
here.
The PAYTYPE variables classify firm’s annual distribution policies into four categories. It
can be seen that, over our sample period, a higher proportion of firms make no distribution,
either by dividends or repurchases (PAYTYPE0). As has been observed in the US, but to a
greater extent (Skinner, 2008), distributions to shareholders are becoming more concentrated
in the UK. PAYTYPE1 represents the category of firms which pay only dividends. This drops
from an average of 64% in the time period when repurchased shares had to be cancelled, to
34.6% when they were allowed to be kept in limited number as treasure shares. A
concomitant increase in the proportion of firms which repurchase shares, in isolation
21
(PAYTYPE2) or in conjunction with a dividend distribution (PAYTYPE3) is observed.
Similar, though less extreme, changes occur following the second regulatory change, which
lifted the cap on the holding of treasure shares. More firms made no distribution at all; fewer
paid dividends alone; more made repurchases, with or without dividends.
The amount a firm expends on dividends and repurchases is constrained by the amount that it
holds in retained earnings (RE). Although the proportion of firms issuing regular dividends is
reducing across our sample periods, the proportion of retained earnings being distributed to
shareholders in this manner (£DIV/RE) is increasing. A large proportional increase in the
amount distributed via share repurchases is observed after the first reform, when the
proportion of RE paid for repurchases (£SP/RE) moved from an average of 1% to 1.8%, but
no change is seen following the reform of 2009.
In Panel B, we report the average of the annual median values for our variables for each
regulatory period. As is expected due to the effect of the recession which occurs in the
middle period, the firm value measures (MV, V, TA) are higher for the third period than the
first period, with a dip in the intervening period. Similar patterns are observed for earnings
and retained earnings. Patterns of distributions are reported next in Tables 3 and 4 and in
Figures 1 and 2.
Table 3 and Figure 1 allow us to look at recent trends in dividend distributions in the UK and
compare these to prior studies using earlier samples which several years earlier than the one
employed here (Fatemi and Bildik, 2012; von Eije and Megginson, 2008; Renneboog and
Trojanowski, 2011; Denis and Osobov, 2008).
[Table 3]
[Figure 1]
22
In common with earlier studies, our data indicate a fairly steep decline in in the propensity for
UK firms to pay dividends from 1994, when 88% of our firms paid a dividend, until around
2001, by which time only 59% of sample companies issued a dividend. The trend away from
dividend distributions continues to decline more gradually until the end of our sample period,
2013, when only 44% of sample firms paid a dividend. There is, however, no dramatic
change in the propensity to pay dividends associated with the introduction of more liberal
rules concerning share repurchases in either 2004 or 2009. Figure 1 contains a graphical
comparison of dividend propensities in our sample compared to those of prior papers using
UK data, making it easier to see a levelling off of the declining trend after the end of the
sample periods of these papers.4 Even more steady is the proportion of firms which, since
2006, pay only dividends. Around 29% of companies for the final eight years of our sample
period would be classed as ‘pure dividend payers’ by Skinner (2008). Contrast this to his
observation that only 6.8% of US firms since 2005 fit this description and we can see that
payout policies remain very different in the UK compared to the US.
[Table 4]
[Figure 2]
In Table 4 and Figure 2, we present share repurchase propensities for our sample firms over
time, comparing our results to those of Renneboog and Trojanowski, 2011, and extending
their sample period forward by nine years to 2013, with some interesting results. The trend in
share repurchases increases sharply from 2004 to 2009, following the first set of reforms, and
4 Looking at Figure 1, the pattern observed by Renneboog and Trojanowski (2011) may appear anomalous when
compared to the results of this and other studies until one considers that they require three years of firm data for
sample firms; this likely introduces some element of bias into their sample which differs from any bias in the
samples from other studies.
23
is driven mostly by an increasing tendency for dividend-paying firms to also repurchase their
own shares. There is a small but clear dip in the trend around the time of the UK recession,
then a recovery. The 2013 reduction in the propensity to pay dividends documented in Table
3 and Figure 1 is also accompanied by a reduction in share repurchase behaviour by
dividend-paying firms. Note that this follows a period of very low economic growth, with
UK GDP growth only 0.2% in 2012. The proportion of companies which only repurchase
shares, and do not distribute dividends, continues to slowly increase after the recovery from
the 2008-09 recession.
[Table 5]
[Figure 3]
[Figure 4]
In Table 5, and Figures 3 and 4, we display information relating to the number of sample
firms, how many of them pay dividends or repurchase shares, and how much distributing
firms spend on dividends or repurchases, on average, each year. Although fewer firms are
paying regular dividends over time, the amount being distributed to shareholders in this way
is increasing steadily. This result was reported by Renneboog and Trojanowski (2011) for
1992-2004; we show that the trend continues beyond their sample period. The amount being
used to repurchase shares is more volatile, increasing steeply following the first reform but
decreasing sharply around the time of the UK recession. Although the sample period of
Renneboog and Trojanowski stops short of the repurchase reforms and the recession, they too
report more volatility in average share repurchase expenditure relative to dividend payments.
The average expenditure on share repurchases often exceeds the average expenditure on
dividends over our time period, though, as Figure 5 reveals, the aggregate annual amount
24
spent on repurchases is never greater than that used to distribute regular dividends in our
sample. Prior to the crash of 2008, it appeared that share repurchases may have been about to
overtake dividends by value but the gap between the two has widened since the financial
crisis. The final column of Table 5, and Figure 6, provide further descriptive evidence on the
non-replacement of dividends by share repurchases in the UK, using as its measure the
proportion of total payout made using share repurchases. The final column of Table 5
presents the outcome of dividing aggregate share repurchases for each year by the sum of
aggregate dividends and aggregate share repurchase for each year. Although, prior to the
financial crisis, share repurchases did appear to be gaining in importance as a payout vehicle,
their use relative to dividends, decreased sharply once recession hit the UK, and they have yet
to fully recover to their pre-crisis position. Although firms managed to maintain their
dividends throughout the crisis, on average, funds used to repurchase shares declined
dramatically.
5.2 Multivariate Analysis
Our next set of analyses examines the determinants of the payout decisions made by sample
firms. Table 6 reports the Pearson correlation coefficients between the variables used in these
tests. It can be seen that there is a high correlation (0.835) between the likelihood of a current
year dividend (DIVDUM) and whether a firm distributed a dividend the year before
(LDIVDUM). Such collinearity could cause inflation of standard errors, causing either or
both variables to appear to have no significance in the regressions when, in fact, either or
both have a significant association with the dependent variable. We therefore keep this in
mind for our later interpretation of results. No other pairs of variables display troublesome
correlations.
25
Table 7 reports the results of logistic regressions designed to examine the factors which
determine whether a firm distributes a dividend or repurchases its own shares in a particular
year. The models used are from Equations (3) and (4) which were developed earlier, in
Section 3. Model 1 investigates the determinants of a dividend distribution. Consistent with
the UK findings of Denis and Osobov (2008) and Fatemi and Bildik (2012), we find the
likelihood of a firm distributing a regular dividend is positively associated with its size,
current year earnings, retained earnings, and undervaluation measured as V/TA. Different to
to the results in these two papers, in our sample firms there is a positive relationship between
asset growth and the propensity to issue a regular dividend. We find older UK firms are more
likely to pay dividends, similar to the US findings of DeAngelo, DeAngelo and Skinner
(2008). Cash holdings are negatively associated with dividend payments, and we find no
relationship between leverage and the propensity to issue regular dividends. Supporting
previous research which has identified a reluctance of managers to stop dividend payments
(e.g. Lintner, 1956; Brav et al., 2005; Denis and Osobov, 2008), we find a strong, positive
association between the likelihood of issuing dividends in the current year and whether the
firm had used this distribution channel in the year prior (LDIVDUM). This is consistent with
the dividend stability results for the UK reported by Lee and Suh (2011). Different to UK
research on an earlier sample by Oswald and Young (2008), which reports no association
between dividends and share repurchases, we find a significant positive association between
the odds of distributing a dividend and the likelihood a firm repurchases its own shares in the
same year (SPDUM). Finally, our annual measure of the health of the general economy
(GGDP) is positively associated with the likelihood a firm issues a dividend in a year.
Model 2 of Table 7 investigates the determinants of a share repurchase decision. The results
are consistent with prior literature in several respects. Firms with fewer growth opportunities
(DTA/TA) and more cash (CASH/TA), which are undervalued (V/TA) are more likely to
26
repurchase their own shares, consistent with earlier work by Oswald and Young (2008).
Leverage (LEV) is significantly positively associated with share repurchases, in line with the
findings of Dittmar (2000) and Dixon et al. (2008), that capital structure rebalancing
motivates some repurchase decisions. However, this finding is contrary to that of Lee and
Suh (2011) who report a negative relationship between the £ amount of UK repurchases and
leverage. Our results are similar to those of Lee and Suh (2011), however, in respect of the
positive association between the propensity to repurchase shares in the current year, and
having repurchased shares in the prior year. Similar to dividend paying firms, companies
which repurchase shares tend to have higher earnings, more retained earnings, and are older,
than those which do not repurchase. Our measure of macroeconomic wellbeing (GGDP) is
weakly positively associated with the likelihood of a share repurchase.
Turning to our hypotheses, we wish to assess whether changes to regulation affected the
distribution policies of UK firms, with H1 being concerned with reforms from late 2003, and
H2 relating to reforms occurring late in 2009. Our dummy variables REGIME2 and
REGIME3 are designed to capture each of these rule changes, respectively. It can be seen
that, relative to the pre-reform period, the likelihood of issuing a regular dividend is lower in
both later time periods, while the likelihood of a share repurchase is higher in both time
periods. We can therefore reject H1, which posits there will be no effect on distribution
decisions post-2003. In order to reject H2, we need to test the effects of the 2009 regulations
against the pre-2009 period. To do this, we re-run Models 1 and 2 (Eq.3 and 4) from Table 7,
excluding all observations prior to 2004 and excluding REGIME2. The results, which are not
tabulated, reveal a significant reduction in the propensity to issue dividends in our latest time
period compared to the middle period (coeff. = -0.524, p-value <.0001), allowing us to reject
H2 for dividends. However, in our amended Model 2, the coefficient on REGIME3 for share
repurchases is not significant (coeff. = 0.011; p-value = 0.881). Whilst the propensity to
27
issue regular dividends decreases significantly post-2009, there is no significant change in the
propensity to buy back shares.
Table 8 reports the results of testing Equations 5 and 6, where the value of dividends and
repurchases are substituted as the dependent variables, with all independent variables
remaining as in Table 7. The results of Model 1 reveal some differences between the
determinants of a dividend payment and the factors affecting the amount paid. Although
firms with asset growth are more likely to issue a dividend, the cash amount distributed is
lower for growth firms, on average (DTA/TA). Although firms with higher reported retained
earnings (as a proportion of total assets) are more likely to pay a dividend, there is a negative
association between retained earnings (RE/TA) and the amount paid in the dividend. This is
consistent with more profitable firms regularly paying dividends but not to the full extent of
their profits, so the dividend as a proportion of prior undistributed profit, is smaller for
regular dividend payers. All other associations remain of the same sign as in Table 7. The
results of Model 2 in Table 8 are largely similar to those from Table 7, with one exception.
Table 7 reports a significant negative association between our market to book ratio (V/TA)
and the propensity to repurchase shares, which is consistent with earlier US literature
(Dittmar, 2000) and suggests companies buy back their own shares when they are cheap.
Table 8, however, reports no significant association between the amount repurchased,
log(1+£SP), and V/TA. In terms of H1, we find a significant reduction (increase) in the
average amount spent on dividends (share repurchases) following the first regulatory reform,
allowing us to reject H1 according to this test. The models tested in Table 8 explain 78% of
the variation in the amount paid in dividends and 33.5% of the variation in share repurchase
expenditure. Share repurchases are therefore less well predicted by variables generally
associated with dividend payments.
28
We again re-run the models using only observations from after 2003 to examine whether
REGIME3 is significant compared to REGIME2. We find that the average amount paid in
dividends is lower following the second reform than during our second regulatory period,
with the coefficient on REGIME3 being -0.122 (p = 0.012). For Model 2, the coefficient on
REGIME3 is also negative (-0.092) and marginally significant (p = 0.109). These results do
not support the null hypothesis that there is no change to payout policy following the second
regulatory reform of 2009. We also run Models 1 and 2 on the full sample, substituting the
lag of log(1+£DIV) for (LDIVDUM), and the lag of log(1+£SP) for LSPDUM as independent
variables. We also substitute log(1+£DIV) for DIVDUM and log(1+£SP) for SPDUM on the
right-hand side of the models. We do this to test the strength of the association between the
amount paid by either payout channel in the current year, and the amount distributed in that
way in the previous year, and the amount paid by the alternative channel in the current year.
The coefficients on the lagged amount variables are all positive and highly significant (p
<.0001). Although our univariate tests indicate that dividend stickiness is reducing over time
in the UK, current year dividends remain strongly associated with the amount paid last year.
More interestingly, there is a strong positive association between the amount paid by one
payout method and the amount expended on the other payment method, for the current year.
The higher the dividend, the more is spent on share repurchases, and vice versa. Further, the
more was spent on share repurchases last year, the more will be spend the current year, by the
average firm.
It is apparent from the results in Table 7 and Table 8 that our sample firms consider dividends
and share repurchases simultaneously. We therefore employ a multinomial approach, similar
to that taken by Renneboog and Trojanowski (2011) and present our results in Table 9. In
Panel A, we examine the factors which differentiate firms with any type of payout compared
to firms which pay neither dividends nor repurchase shares. It can be seen that larger firms,
29
and those with lower market values (V/TA) are more likely to issue dividends, or repurchase
shares, or both, as are firms with higher current earnings (E/TA), retained earnings (RE/TA)
and older firms (AGE). As is the case in the US (Dittmar, 2000), it is the more mature firms
which are making distributions to shareholders. Although cash as a proportion of total assets
(CASH/TA) is negatively associated with the likelihood of paying dividends, having more
cash on the balance sheet is positively associated with the chance that a firm will repurchase
its own shares. For firms making both type of distribution, there is no apparent relationship
between this and the amount of cash held. Firms with more debt (LEVERAGE) are less likely
to issue dividends or repurchase shares alone, but there is no association with debt and the
decision to make both types of distribution in the same year.
Turning to variables which provide some information about payout trends, we find that firms
which only distribute a dividend in the current year (PAYTYPE1) are more likely to have paid
a dividend last year (LDIVDUM) but less likely to have repurchased shares. However, last
year’s repurchase decision (LSPDUM), is positively associated with more repurchases in the
current year, whether or not they are accompanied by a dividend (i.e. PAYTYPE2 and
PAYTYPE3). In fact, a dividend payment last year is positively associated with all types of
payout in the current year. The results for growth in GDP (GGDP) suggest that firms are
more likely to pay dividends (alone or along with share repurchases) when the economy as a
whole is performing well, but this does not have a significant association with the decision to
repurchase shares but pay not dividend.
What effect do we observe for our share repurchase regulatory regimes? Looking at the
results for the variable REGIME2 in Panel A reveals that firms are less likely to distribute
dividends alone (PAYTYPE1), and more likely to repurchase shares, either alone or along
with dividends (PAYTYPE2 and PAYTYPE3) following the first regime change, when firms
30
were first allowed to retain repurchased shares as treasury shares, compared to the earlier
period when firms had to cancel any repurchased shares. Following the second change in
regulation, which lifted the cap on the proportion of equity which could be held as treasury
stock, even fewer companies issued dividends alone (PAYTYPE1), while more issued
dividends in conjunction with repurchasing their own shares (PAYTYPE2), and no difference
is observed in the practice of making share repurchases without paying dividends
(PAYTYPE3). Moving on to Panel B, we compare firms which either make a repurchase
alone (PAYTYPE2) or make a repurchase and also issue regular dividends (PAYTYPE3) to
companies which issue dividends alone, with some interesting results. LDIVDUM is
negatively associated with both PAYTYPE2 and PAYTYPE3 so firms which repurchase in the
current year are much less likely to have paid a dividend in the prior year, compared to firms
which issue only a dividend in the current year (PAYTYPE1). Compared to dividend alone
firms, there are more repurchasing firms, whether PAYTYPE2 or PAYTYPE3 following each
share repurchase reform.
We therefore reject our null hypotheses that there has been no change to firm distribution
policies associated with the liberalising of regulation relating to share repurchases in the UK.
On the contrary, allowing UK companies to repurchase shares and hold them as treasury
stock is associated with a contemporaneous decrease in the use of regular dividends and an
increase in the use of share repurchases as vehicles for returning funds to shareholders.
6. Supplementary Analyses and Limitations of the Study
It may be argued that managers do not know current year earnings (and therefore retained
earnings) by the time they declare current year dividends. We therefore repeat the tests in
Table 7, replacing current year profitability (E/TA) and retained earnings (RE/TA) with one-
year lags of the same variables. The lagged variables are positively and significantly
31
associated with the propensity to issue regular dividends and to repurchase shares, though
their coefficients are not as large as those recorded for current year income or retained
earnings. All other variables are qualitatively unaffected. We repeat the exercise with the
models from Table 8, with the same outcome.
Our analyses have assumed that share repurchases are used to distribute part of firm earnings
to shareholders. As is pointed out by Fama and French (2001) and Skinner (2008), not all
repurchases are for this purpose. An alternative reason is to finance transactions such as
those relating to stock option schemes or acquisitions. Once the retention of treasury shares
was allowed after the first reform of November, 2003, this applies to the UK as well as the
US. We make no attempt to differentiate between repurchases for distribution purposes and
those for transactional reasons and this is a limitation of our work.
7. Summary and Conclusions
We track and analyse the distribution policies, with respect to regular dividends and share
repurchases, for a sample of 19,652 UK listed companies from 1994-2013. This time period
includes two regulatory changes, each of which relaxed previously strict restrictions on share
repurchases in the UK. The first, late in 2003, allowed firms to retain repurchased shares as
treasury stock, where the former requirement was for them to be cancelled. The second, in
October 2009, removed a 10% ceiling on the amount of issued equity that could be held as
treasury shares. Our long sample period enables us to comment on the effects of these
changes. The study provides information about: (a) trends in distribution practices in the UK
for a twenty-year time period; (b) the determinants of payout policy in respect of dividends
and share repurchases; and (c) whether, as in the US, repurchases are replacing dividends as
the primary vehicle for returning cash to shareholders in the UK.
32
First, in common with prior UK research, we observe a decrease in the number of dividend-
issuing firms over time. However, we document that the decline has slowed and it does not
appear that dividends are disappearing in the UK, as some previous scholars have suggested.
Unlike in the US, a stable core of around 29% of UK firms pay dividends and only dividends
in recent years. Share repurchases increase in number and value for most of our sample
period, with a particular increase following the first reform which allowed firms to retain
repurchased shares in treasury. However, our evidence does not indicate that share
repurchases are replacing dividends in the UK. Although, as a proportion of total payout,
share repurchases gained in importance steadily up until 2007, the advent of the financial
crisis saw a rapid decrease in the value of share repurchases in the UK, and their recovery
since then has only been partial. In contrast, UK firms have maintained dividends even
through the recession and financial crisis. A reduction in the proportion of dividend-paying
UK firms, but an increase in the average dividend paid by those firms, suggests greater
concentration of dividends in the UK, as has been observed in the US.
We build on earlier models which analysed the determinants of payout policy with some
interesting results. Many of the factors which are associated with the propensity to issue
dividends, also seem to influence firms’ repurchase policies. Larger, more profitable and
mature companies are more likely to make both types of distribution. Firms which are
undervalued are also more likely to return funds to shareholders, while asset growth firms are
more likely to issue dividends but less likely to repurchase shares. Both types of distribution
are more likely when there is positive growth in GDP. Our multivariate analyses support the
indications from descriptive tests that firms which make one type of distribution are more
likely to make the other – there is a positive association between the propensity to issue
dividends and to repurchase shares. There is also a strong positive association between
current payout practice and that of the previous year, suggesting a degree of ‘stickiness’
33
prevails in payout decisions. We therefore use multinomial models to allow for the apparent
simultaneity of the dividend/repurchase decision. They validate the results from our binomial
tests and provide strong evidence that the regulatory changes have affected payout policy in
the UK. Relative to firms which make no distribution, following the first reform of 2003, UK
firms are less likely to issue regular dividends alone, and more likely to repurchase their own
shares, with or without an accompanying dividend. Following the second reform, companies
are less likely to issue dividends alone, and more likely to make repurchases alone, but there
is no change in the likelihood a firm will make both types of distribution, relative to zero
payout companies. Relative to firms which only issue dividends, companies are more likely
to make repurchases, with or without an accompanying dividend, after each reform.
Compared to firms which repurchase only, firms are less likely to issue accompanying
dividends after each reform, but this is only significant following the second rule change,
which lifted the ceiling on treasure shares.
Overall, our study concludes that, although recent reforms have led to an increase in the use
of share repurchases as a vehicle for returning funds to shareholders in the UK, dividends are
not disappearing, and they are not being replaced by repurchases.
34
REFERENCES
Andriosopolous, D., & Lasfer, M. (2015). The market valuation of share repurchases in Europe,
Journal of Banking and Finance, 55, 327-339.
Bank of England (1988), Share repurchase by quoted companies: a report by the City Capital
Markets Committee. Bank of England Quarterly Bulletin (August).
Bell, L., & Jenkinson, T. (2002). New evidence of the impact of dividend taxation and on the
identity of the marginal investor, Journal of Finance, 57(3), 1321-1346.
Brav, A., Graham, J. R., Harvey, C. R., & Michaely, R. (2005). Payout policy in the 21st entury,
Journal of Financial Economics, 77, 483-527.
DeAngelo, H., DeAngelo, L., & Skinner, D. J. (2004). Are dividends disappearing? Dividend
concentration and the consolidation of earnings. Journal of Financial Economics, 72(3), 425-456.
DeAngelo, H., DeAngelo, L., & Skinner, D. J. (2008). Corporate payout policy. Foundations
and Trends in Finance, 3(2-3), 95-287.
Denis, D. J., & Osobov, I. (2008). Why do firms pay dividends? International evidence on the
determinants of dividend policy. Journal of Financial Economics, 89(1), 62-82.
Dhanani, A., & Roberts, R. (2009). Corporate share repurchases: the perceptions and practices
of UK financial managers and corporate investors. Edinburgh: The Institute of Chartered
Accountants of Scotland.
Dittmar, A. K. (2000). Why do firms repurchase stock? The Journal of Business, 73(3), 331-355.
Dixon, R., Palmer, G., Stradling, B., & Woodhead, A. (2008). An empirical survey of the
motivation for share repurchases in the UK. Managerial Finance, 34(12), 886-906.
Eije, H. von, & Megginson, W. L. (2008). Dividends and share repurchases in the European
Union. Journal of Financial Economics, 89(2), 347-374.
Fama, E. F., & French, K.R. (2001). Disappearing dividends: changing firm characteristics or
lower propensity to pay? Journal of Financial Economics, 60(1), 3-43.
Fatemi, A., & Bildik, R. (2012). Yes, dividends are disappearing; worldwide evidence. Journal of Banking and Finance, 36(2), 662-677.
Grullon, G., & Ikenberry, D. L. (2000). What do we know about stock repurchases? Journal of
Applied Corporate Finance, 13(1), 31-52.
35
Grullon, G., & Michaely, R. (2002). Dividends, share repurchases and the substitution
hypothesis. The Journal of Finance, 57 (4), 1649-1684.
Ikenberry, D. L., Lakonishok. J., & Vermaelen, T. (1995). Market underreaction to open market
share repurchases. Journal of Financial Economics, 39(2-3), 181-208.
Julio, B., & Ikenberry, D. L. (2004). Reappearing dividends. Journal of Applied Corporate
Finance, 16(4), 89-100.
Lee, B. S., & Suh, J. (2011). Cash holdings and share repurchases: international evidence. Journal of Corporate Finance, 17(5), 1306-1329.
Lintner, J. (1956). Distribution of income of corporations among dividends, retained earnings,
and taxes. American Economic Review, 46(2), 97-113.
Oswald, D., & Young, S. (2008). Share requisitions, surplus cash, and agency problems.
Journal of Banking and Finance, 32(5), 795-806.
Rau, P. R., & Vermaelen, T. (2002). Regulation, taxes and share repurchases in the United
Kingdom. Journal of Business, 75(2), 245-282.
Renneboog, L., & Trojanowski, G. (2011). Patterns in payout policy and payout channel choice.
Journal of Banking and Finance, 35(6), 1477-1490.
Scott, G. (2014). England and Wales treasury shares guide. IBA Corporate and M&A Law
Committee Report.
Siems, M. M., & Cesari, A. (2012). The law and finance of share repurchases in Europe. Journal
of Corporate Law Studies, 12(1), 33-57.
Skinner, D. J. (2008). The evolving relation between earnings, dividends and stock repurchases. Journal of Financial Economics, 87(3), 582-609.
36
Table 1: Description of Variables
REGIME A set of 3 dummy variables which indicate the regulatory period during which
an observation occurs: REGIME1 represents observations from 1994 to 2003;
REGIME2 indicates an observation is from 2004 to 2009; and REGIME3
denotes observations from 2010 to 2013.
GGDP This is the annual measure of growth in gross domestic product (GDP)
reported by the Office of National Statistics (ONS).
DIVDUM A dummy variable, coded 1 if the firm pays a regular dividend in the year,
otherwise 0.
LDIVDUM One year lag of DIVDUM.
DIVDUM% Percentage of sample firms paying a regular dividend in the year.
DIVDUM2YR
%
Percentage of sample firms paying a dividend this year and also last year.
£DIV The total value of the common dividends declared for the year (Worldscope item
WC18192 - Dividends Provided for or Paid).
SPDUM A dummy variable, coded 1 if the firm reports it has repurchased its own
shares in the year, otherwise 0.
LSPDUM One year lag of SPDUM.
SPDUM% Percentage of sample firms repurchasing shares in the year.
SPDUM2YR% Percentage of sample firms repurchasing shares this year and also last year.
£SP We measure share repurchases for the year as funds used to decrease the
outstanding shares of common stock (Worldscope item WC04751 -
Common/Preferred Redeemed, Retired, Converted etc). (It is assumed that the
buying back of preference shares is negligible.)
MV Market value represents the market capitalisation at the closing price of the
firm’s share at their fiscal year end (Worldscope item WC08001 – Market
Capitalisation).
SIZE The percentile ranking of the firm in the sample for the year, based upon MV.
BE Book equity is calculated as common equity (Worldscope item WC03501 – Common
Equity) plus non-equity reserves equity if reported (Worldscope item WC03401 –
Non-Equity Reserves).
V Following Denis and Osobov (2008), entity market value is calculated as book
value of total assets (Worldscope item WC02999- Total Assets) minus book
value of equity (BEt) plus market value of equity, where market value of
equity is measured as the closing stock price at fiscal year-end times the
number of shares outstanding (Worldscope item WC08001 – Market
Capitalisation).
TA Total assets is represented by the sum of total current assets, long term
receivables, investment in unconsolidated subsidiaries, other investments, net
property plant and equipment and other assets (Worldscope item WC02999 -
37
Total Assets).
BV Book value is measured as the sum of preferred stock and common
shareholders’ equity at the financial year end (Worldscope item WC03995 -
Total Shareholder’s Equity).
E
Our measure of earnings is net income before extraordinary items and
preference dividends (Worldscope item WC01551 - Income before
extraordinary items and preferred and common dividends, but after operating
and non-operating income and expense, reserves, income taxes, minority
interest and equity in earnings).
RE Retained earnings represents the accumulated after tax earnings of the firm
which have not been distributed as dividends to shareholders and allocated to
a reserve account for the financial year ending in year t (Worldscope item
WC03495 – Retained Earnings).
AGE The number of years since the incorporation of the firm. It is calculated as the
firm’s financial year-end (Worldscope item WC05350– Date of Fiscal Year
End) minus its incorporation date (Worldscope item WC18273– Date of
Incorporation).
CASH Cash is measured by cash and cash equivalents (Worldscope item WC02001 -Cash &
Equivalent).
LEVERAGE
(LEV)
Leverage is measured by total debt (Worldscope item WC03255 – Total debt, divided
by total assets (Worldscope item WC02999 - Total Assets).
PAYTYPE Observations are classed as one of four types: PAYTYPE0 = neither dividend
nor repurchase; PAYTYPE1 = dividend only; PAYTYPE2 = share repurchase
only; PAYTYPE3 = both dividend and repurchase.
38
Table 2: Descriptive Statistics by Regulatory Time Period
Sample period Period 1 Period 2 Period 3
Years 1994-2003 2004-2009 2010-2013
Observations 10,154 6,235 3,263
GDP growth 3.089 0.933 1.663
Panel A: sample period average percentages % % %
DIVDUM 71.174 50.264 46.705
DIVDUM2YR 67.674 45.952 42.472
SPDUM 8.144 19.166 22.464
SPDUM2YR 3.116 11.428 15.344
PAYTYPE0 27.812 46.223 49.035
PAYTYPE1 64.044 34.611 28.501
PAYTYPE2 1.014 3.512 4.260
PAYTYPE3 7.130 15.654 18.204
£DIV/RE 6.5 8.4 14.7
£SP/RE 1.0 1.8 1.8
Panel B: means of median values for each sample period (Age is in years; all undeflated
variables are in £mil)
MV 48.521 41.325 55.944
V 78.390 65.825 91.148
V/TA 1.383 1.407 1.310
TA 50.694 47.002 72.883
E 1.507 0.900 1.421
E/TA 0.044 0.028 0.030
RE 4.064 3.308 8.887
RE/TA 0.136 0.090 0.145
AGE 21.398 12.332 15.708
CASH 3.742 4.565 7.354
CASH/TA 0.074 0.105 0.103
LEVERAGE 0.144 0.115 0.090
(Variables are defined in Table 1.)
39
Table 3: Trends in UK dividend payout policy - % of firms paying a dividend in a year
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
This study: all
div. 88 89 89 72 73 71 65 59 56 57 56 53 49 48 50 46 45 49 48 44
This study: div.
alone 85 86 83 67 63 62 56 51 48 48 46 40 32 30 29 30 29 29 28 28
F&B 84 87 82 76 76 74 64 58 55 53 50 46 43
vE&M 84 86 76 75 76 70 56 54 48 46 45 44
R&T 84 86 85 86 86 84 80 77 76 76 77
D&O 85 87 88 85 85 82 71 61 56
F&B = Fatemi and Bildik (2012). Data extracted from their Table 2A. vE&M = von Eije and Megginson (2008). Data extracted from their Figure 3. R&T = Renneboog and
Trojanowski (2011). Data extracted from their Table 3. D&O = Denis and Osobov (2008). Data extracted from their Table 5.
0
10
20
30
40
50
60
70
80
90
100
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Figure 1: Trends in UK Dividend Distributions
This study all dividend This study pure dividend F&B vE&M R&T D&O
40
0
5
10
15
20
25
30
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Figure 2: Trends in UK Repurchases
This study: SP only This study: SP+Div This study: Total SP R&T: SP only R&T: SP + Div R&T: Total SP
Table 4: Trends in UK share repurchases - % of firms repurchasing shares in a year
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
This study
SP only 0.34 0.22 0.56 0.19 0.44 0.48 2.10 1.46 1.56 2.60 2.42 3.32 2.98 3.80 3.83 4.93 4.16 4.05 3.71 5.29
SP+Div 2.84 3.40 5.15 5.06 9.71 9.46 9.26 8.32 7.61 8.90 9.89 12.36 16.41 17.83 21.45 16.21 16.63 20.00 19.78 16.17
Total
SP 3.18 3.62 5.71 5.25 10.15 9.94 11.36 9.78 9.17 11.50 12.31 15.68 19.39 21.63 25.28 21.14 20.79 24.05 23.49 21.46
Renneboog and Trojanowski (2011). (Data extracted from their Table 3.)
SP only 0.34 0.22 0.74 0.42 0.39 0.76 1.19 0.90 1.18 1.71 1.37
SP +
Div 4.20 3.87 5.61 8.7 12.2 11.4 13.6 11.6 12.4 13.9 14.6
Total
SP 4.54 4.09 6.35 9.12 12.59 12.16 14.79 12.5 13.58 15.61 15.97
41
Table 5: Number of Sample/Dividend/Repurchasing Firms, and Amounts Spent on Distributions, by Year
Year
Sample
size
Number of
dividend paying
firms
Average £Div per
dividend-paying
firm
Number of
repurchasing
firms
Average £SP
per
repurchasing
firm
£SP proportion
of total payout
(£SP+£DIV)
1994 879 775 14107 28 20306 0.05
1995 912 816 16811 33 60367 0.13
1996 894 792 19437 51 20196 0.06
1997 1068 774 18883 56 34441 0.12
1998 1123 818 28424 114 50998 0.20
1999 1047 743 24233 104 30437 0.15
2000 1047 682 32145 119 27324 0.13
2001 1094 646 33461 107 51813 0.20
2002 1090 609 34511 100 80899 0.28
2003 1000 572 39913 115 60104 0.23
2004 1031 574 49038 127 90003 0.29
2005 1084 572 56222 170 126450 0.40
2006 1109 538 70895 215 156987 0.47
2007 1105 531 71291 239 149713 0.49
2008 993 498 72727 251 75012 0.34
2009 913 421 93869 193 20947 0.09
2010 890 402 98560 185 25215 0.11
2011 865 423 111645 208 114268 0.33
2012 809 389 136193 190 89025 0.24
2013 699 310 145102 150 88281 0.23
42
0
500
1000
1500
2000
2500
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Figure 3:
Number of Sample/Dividend Paying/Repurchasing Firms by Year
NO. DIVIDEND PAYING FIRMS NO. SP FIRMS SAMPLE SIZE
0
50000
100000
150000
200000
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Figure 4:
Average Spent on Dividends/Repurchases by Dividend Paying/Repurchasing Firms by Year
AVG £DIV PER DIV FIRM AVG £SP PER SP FIRM
43
0
10000000
20000000
30000000
40000000
50000000
60000000
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Figure 5:
Aggregate £ Amount Spent on Dividends/Repurchases by Year
AGGREGATE £DIV AGGREGATE £SP
0
0.1
0.2
0.3
0.4
0.5
0.6
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Figure 6: Share Repurchase Amounts as a Proportion of Annual Aggregate Payout
44
Table 6: Correlations
1 2 3 4 5 6 7 8 9 10 11 12 13 14
1. DIVDUM 1.000
2. SPDUM 0.188 1.000
3. LDIVDUM 0.835 0.181 1.000
4. LSPDUM 0.169 0.508 0.186 1.000
5. SIZE 0.419 0.254 0.402 0.236 1.000
6. V/TA -0.144 -0.030 -0.166 -0.027 0.133 1.000
7. DTA/TA 0.089 -0.013 0.029 -0.020 0.166 0.064 1.000
8. E/TA 0.427 0.136 0.383 0.115 0.302 -0.220 0.484 1.000
9. RE/TA 0.387 0.121 0.379 0.113 0.270 -0.314 0.315 0.615 1.000
10. AGE 0.341 0.104 0.365 0.103 0.175 -0.177 -0.058 0.189 0.189 1.000
11. CASH/TA -0.304 -0.046 -0.316 -0.054 -0.112 0.373 -0.034 -0.287 -0.311 -0.223 1.000
12. LEV 0.129 0.211 0.138 0.218 0.330 -0.039 0.008 0.066 0.063 0.061 -0.088 1.000
13. REGIME2 -0.142 0.101 -0.142 0.081 0.000 0.015 0.036 -0.051 -0.047 -0.105 0.076 0.022 1.000
14. REGIME3 -0.126 0.109 -0.112 0.129 0.000 -0.019 -0.054 -0.034 -0.106 -0.015 0.031 0.096 -0.304 1.000
15. GGDP 0.130 -0.114 0.101 -0.147 0.000 0.081 0.087 0.073 0.063 0.053 -0.036 -0.067 -0.439 -0.144
(Variables are defined in Table 1.)
45
(Variables are defined in Table 1.)
Table 7: Determinants of Dividends and Share Repurchases
Model 1:
(Eq. 3)
DIVDUM = SIZE V/TA DTA/TA E/TA RE/TA AGE CASH/TA LEV
LDIVDUM SPDUM REGIME2 REGIME3 GGDP INDUSTRY
Model 2:
(Eq. 4)
SPDUM = SIZE V/TA DTA/TA E/TA RE/TA AGE CASH/TA LEV
LSPDUM DIVDUM REGIME2 REGIME3 GGDP INDUSTRY
Model 1: DIVDUM Model 2: SPDUM
Parameter
ML
Estimate
Wald
Chi 2 Pr > ChiSq
ML
Estimate
Wald
Chi2 Pr > ChiSq
Intercept -3.107 406.389 <.0001 -4.411 982.234 <.0001
SIZE 0.021 258.374 <.0001 0.014 154.990 <.0001
V/TA -0.078 9.640 0.002 -0.085 12.386 0.000
DTA/TA 0.307 7.538 0.006 -0.911 78.188 <.0001
E/TA 6.209 436.441 <.0001 2.049 57.156 <.0001
RE/TA 1.234 164.128 <.0001 0.302 21.251 <.0001
AGE 0.005 19.741 <.0001 0.001 3.268 0.071
CASH/TA -0.907 21.078 <.0001 0.870 26.043 <.0001
LEV -0.000 0.594 0.441 0.000 13.791 0.000
DIVDUM 0.530 51.132 <.0001
LDIVDUM 3.921 3682.618 <.0001
SPDUM 0.457 24.611 <.0001
LSPDUM 2.362 1891.976 <.0001
REGIME2 -0.537 45.753 <.0001 0.949 210.897 <.0001
REGIME3 -1.092 146.667 <.0001 0.920 160.794 <.0001
GGDP 0.127 42.362 <.0001 0.024 2.435 0.119
INDUSTRY DUMMIES YES YES
N 19,652 19,652
-2LOG L 8130 11068
p-value of LR <.0001 <.0001
% concordant 97.1 85.3
46
Table 8: Determinants of Amount Expended on Dividends and Share Repurchases
Model 1: log(£DIV)
Model 2: log(£SP)
Parameter t Value Pr > |t| Parameter t Value Pr > |t|
Estimate Estimate
Intercept -0.957 -13.56 <.0001 -0.875 -10.8 <.0001
SIZE 0.051 80.57 <.0001 0.014 18.6 <.0001
V/TA -0.066 -6.79 <.0001 0.001 0.08 0.934
DTA/TA -0.221 -4.52 <.0001 -0.559 -9.96 <.0001
E/TA 1.045 13.43 <.0001 0.602 6.67 <.0001
RE/TA -0.114 -8.11 <.0001 0.011 0.65 0.514
AGE 0.004 8.65 <.0001 0.001 2.43 0.015
CASH/TA -0.389 -4.59 <.0001 0.336 3.45 0.001
LEV 0.000 26.67 <.0001 0.000 24.88 <.0001
DIVDUM 0.292 6.5 <.0001
LDIVDUM 5.051 130.97 <.0001
SPDUM 0.600 13.45 <.0001
LSPDUM 3.412 64.52 <.0001
REGIME2 -0.149 -3.81 <.0001 0.710 15.89 <.0001
REGIME3 -0.301 -6.73 <.0001 0.607 11.77 <.0001
GGDP 0.073 7.41 <.0001 0.043 3.76 0.000
INDUSTRY DUMMIES YES YES
N 19,652 19,652
F-value 3327 472
p>F <.0001 <.0001
Adj. R2 0.7804 0.3349
(Variables are defined in Table 1.)
Model 1:
(Eq. 5)
Log(1+£DIV) = SIZE V/TA DTA/TA E/TA RE/TA AGE CASH/TA LEV
LDIVDUM SPDUM REGIME2 REGIME3 GGDP INDUSTRY
Model 2:
(Eq. 6)
Log(1+£SP) = SIZE V/TA DTA/TA E/TA RE/TA AGE CASH/TA LEV
LSPDUM DIVDUM REGIME2 REGIME3 GGDP INDUSTRY
47
Table 9: Multinomial Probit Regressions of Payout Type
Panel A: Base outcome = No Payout (PAYTYPE = 0)
PAYTYPE1: Dividend Only PAYTYPE2: Repurchase Only PAYTYPE3: Dividend plus Repurchase
Coef. z P>|z| Coef. z P>|z| Coef. z P>|z|
SIZE 0.015 15.01 0.000 0.014 9.04 0.000 0.023 18.62 0.000
V/TA -0.045 -2.32 0.020 -0.048 -1.92 0.055 -0.114 -3.01 0.003
DTA/TA 0.116 1.06 0.290 -0.425 -3.86 0.000 -0.501 -4.14 0.000
E/TA 3.291 6.39 0.000 1.102 4.61 0.000 4.962 8.09 0.000
RE/TA 0.843 10.43 0.000 0.296 4.32 0.000 0.776 7.35 0.000
AGE 0.003 4.28 0.000 0.003 2.85 0.004 0.004 4.54 0.000
CASH/TA -0.536 -3.75 0.000 0.604 3.79 0.000 0.038 0.20 0.844
LEV -0.000 -2.33 0.020 -0.000 -2.04 0.041 0.000 0.01 0.994
LDIVDUM 2.990 55.04 0.000 0.360 3.92 0.000 2.595 34.11 0.000
LSPDUM -0.263 -2.91 0.004 1.484 15.08 0.000 1.592 17.50 0.000
REGIME2 -0.440 -8.15 0.000 0.386 4.40 0.000 0.290 4.41 0.000
REGIME3 -0.844 -13.29 0.000 0.260 2.69 0.007 -0.075 -1.00 0.319
GGDP 0.083 5.87 0.000 0.025 1.40 0.161 0.096 5.99 0.000
CONSTANT -2.291 -22.03 0.000 -3.366 -17.64 0.000 -4.491 -29.65 0.000
INDUSTRY DUMMIES YES #CLUSTERS 2,689
LOG LIKELIHOOD -9692 *** N 19,652
48
Panel B: Base outcome = Dividend Only (PAYTYPE = 1)
Panel C: Base outcome =
Repurchase Only (PAYTYPE = 2)
PAYTYPE2:
Repurchase Only
PAYTYPE3:
Dividend plus Repurchase
PAYTYPE3:
Dividend plus Repurchase Coef. z P>|z| Coef. z P>|z| Coef. z P>|z|
SIZE -0.001 -0.51 0.610 0.008 7.43 0.000 0.009 4.87 0.000
V/TA -0.003 -0.11 0.916 -0.069 -2.15 0.032 -0.066 -1.60 0.109
DTA/TA -0.540 -3.88 0.000 -0.617 -5.84 0.000 -0.077 -0.53 0.596
E/TA -2.190 -4.16 0.000 1.671 3.12 0.002 3.860 6.37 0.000
RE/TA -0.547 -5.57 0.000 -0.067 -0.71 0.479 0.480 4.07 0.000
AGE 0.000 0.09 0.927 0.001 1.03 0.304 0.001 0.51 0.608
CASH/TA 1.140 5.97 0.000 0.574 3.35 0.001 -0.566 -2.45 0.014
LEV 0.000 0.17 0.862 0.000 3.64 0.000 0.000 2.73 0.006
LDIVDUM -2.630 -26.89 0.000 -0.394 -5.38 0.000 2.235 21.38 0.000
LSPDUM 1.748 17.88 0.000 1.855 34.41 0.000 0.107 1.15 0.251
REGIME2 0.826 9.10 0.000 0.730 14.35 0.000 -0.096 -0.97 0.331
REGIME3 1.104 11.06 0.000 0.769 12.35 0.000 -0.335 -3.11 0.002
GGDP -0.058 -3.09 0.002 0.013 1.02 0.308 0.071 3.62 0.000
CONSTANT -1.075 -5.56 0.000 -2.200 -15.56 0.000 -1.125 -5.17 0.000
(Variables are defined in Table 1.)