9 Introduction 9
Introduction
Dividends are of great importance in establishing a company’s value. In particular, models like
the Dividend Discount Model (Gordon (1959)) are widely used in finance to value financial
companies. DDM model is based on the theory that a stock’s price is the present value of its
future dividends. Also Fama and French (1998) find empirical support for this relationship.
Therefore, having reliable estimates of dividend payments and understanding the key drivers
behind dividend policy decisions is extremely important.
Many works are available in literature about the explanatory factors for the dividend policy.
However, the majority of them aims to explain this phenomenon for industrial companies.
This study wants to explain what drives dividend payments for the largest world listed banks. My
sample is composed by 82 banks from 23 countries. I expect that dividend payments in banking
depend on: size, growth opportunities, profitability and ability of the firm to generate cash.
I measured dividend policy using the ratio dividend over equity. I first created a general model
which has, as dependent variable, the ratio between dividends and equity, and includes basic
explanatory factors: ln[total assets] (size), market to book ratio (growth opportunities), ROA
(profitability) and free cash flows divided by total assets (cash generation ability). I expect to find
positive relationships for all the explanatory factors.
Then I added risk factors to this general model. I wanted to investigate if risk influences dividend
payments. For this reason I constructed market risk indicators and other risk factors to measure
default risk. I suppose to have a negative relation between risk and dividends. In fact, the higher
the riskiness of the bank, the lower should be the dividend disbursement.
Moreover, it might be that major shareholders or, more in general, the ownership structure,
influences the management of banks in deciding the dividend distribution strategy. I included in
the general model variables representative of the ownership structure.
I found statistically significant results for both risk and ownership structure factors. This results
permitted me to expand my initial model taking them into account.
The last part of this study is inspired by the work of Acharya, Gujral and Shin (2009). They find
that common equity during the recent financial crisis diminished not only because of big financial
10 Literature Review 10
losses but also because of large scale payments of dividends. I investigated if the banks that have
been saved by the public authorities were among the ones who have paid higher dividends.
Therefore I created a sub-sample of bailed out banks on which I tested my hypothesis.
The remainder of this paper is organized as follows. Section 1 presents a summary of previous
literature related to dividend policy. Section 2 describes the research design, sampling procedures
and data characteristics. Main empirical results about dividends’ explanatory factors are reported
in Section 3. Section 4 summarizes the findings related to the relationship between government
bailout and dividend payments. Appendix 1 explains how the sub-sample of bailed out banks has
been constructed. Annex 1 shows the list and the description of all the variables employed.
Finally, Annex 2 reports bailed out banks’ press releases.
1. Literature Review
Over the past years an extensive literature has examined the determinants that influence dividend
policy of the firms. However, most of these studies constructed dividend determination models
for non-financial and non-utility firms.
Fama and French (2001), after having divided firms in two groups (dividend payers and non-
payers), show evidence that payers and non-payers differ in terms of profitability, investment
opportunities and size. In particular, dividend non-payers demonstrate low level of profitability,
high growth perspective and hence a greater fraction of retained profits. Retained profits can be
seen as a life-cycle variable, that signals the maturity of the firm. In particular, according to the
life-cycle theory, firms pay dividends based on their level of evolution. In their early years, firms
are characterized by low earnings, low level of profitability and focused principally on their
investment strategy. Following Fama and French (2001), also Grullon, Michaely and
Swaminathan (2002) and DeAngelo and DeAngelo (2006) advance life-cycle explanations for
dividends that rely, implicitly or explicitly, on the trade-off between the advantages (e.g.
floatation cost savings) and the cost of retention (e.g. agency costs of free cash flow).
Specifically, firms with low retained earnings compared to total equity (or total assets) tend to be
in the capital infusion stage, while firms with higher ratios tend to be more mature, with ample
cumulative profits that make them largely self-financing, hence good candidates to pay dividends
(DeAngelo, DeAngelo and Stulz (2006)).
11 Literature Review 11
The link between dividends and executive stock holdings has been taken into consideration by
several studies. Blouin, Raedy and Shackelford (2004) report that dividend increase are positively
related to insider ownership. Lambert, Lanen and Larcker (1989), Jolls (1998), Weisbenner
(2000), Fenn and Liang (2001) and Khale (2002) find that when managers hold more of their
wealth in the form of options than in direct shares, they tend to use dividends less heavily.
Following these findings and the dividend tax cut in 2003 in the US, Brown, Liang and
Weisbenner (2007) find that, while there is no relation between executive stock ownership and
the likelihood of a dividend increase in the 10 years before the tax cut (when the dividend tax
rate, and hence the cost of paying out dividends, to the stock-owning executive was much
higher), the relation is quite strong in 2003.
Dividends as a signal of the economic health of the firm are proposed in Bhattacharya (1979),
John and Williams (1985) and Miller and Rock (1985). In their view, management pays and
increases dividends to signal private information about the quality of the firm’s earnings to the
investing public. Nissim and Ziv (2001) find that a dividend change contains information about
future earnings that cannot be found in other market data.
Worth to be mentioned is the work, inspired by Kumar and Lee (2001), of Cohen and Yagil
(2009) who investigate the extent to which financially distressed firms pay dividends in order to
attract investors. Higher dividend yield and payout ratio are found for financially distressed firms
than for financially stable firms. They also find that dividend flows are more unstable for
troubled firms and more linked to their earnings flows. The reason why distressed firms act in a
similar way has been identified in the importance that these kind of firms ascribe to the dividend
payment or to the aggressive dividend policy undertaken. This erodes firms’ financial stability
and force them to reduce dividend per share rapidly.
Both firm size and industry regulation can affect dividend payments. Barclay, Smith, and Watts
(1995), in explaining dividend policy for non-financial and non-utility firms use the natural log of
real sales to account for the size-effect. In particular, firms with higher revenues should have a
remote bankruptcy probability and hence a greater propensity for paying dividends. They
introduce the regulation-effect creating a dummy variable that assumes values equal one for a
regulated industry and zero otherwise. Regulated industries lead firms to make regular dividend
payments and, thus, should be associated with higher dividend yields.
12 Data sample 12
Financial institutions differ from industrial companies in their core business but also in regulatory
aspects. For this reason dividend models for financial institutions should differ from the one for
non-financial companies. Dickens, Casey and Newman (2003), starting from the work of Barclay
et al., adapt the model for the financial industry. Barclay et al. used a regulation variable to signal
if the industry considered was strongly regulated or not. Dickens et al. removed this dummy
variable since financial industry is heavily regulated and they find empirical evidence for 5
factors that influence bank dividend policy: investment opportunities, size, agency cost, dividend
history and risk.
Differences between dividend policy of bank holding companies during steep recessions and
good economic conditions have been found by Theis, Yesilyaprak, Jauregui and Dutta (2008),
even if their limited sample reduce the relevance of their findings. A future study, examining data
covering several years surrounding the recession is needed in order to increase the understanding
of this phenomenon.
2. Data sample
The construction of the list of the biggest world listed banks by total assets has been the first step
of this analysis. Different data sources have been analyzed and mixed together to obtain the final
result. I started by looking at the rankings published by “The Banker” review from 2006 to 2010.
Every year The Banker’s “Top 1000 World Banks” ranking tracks the world’s largest bank
holding companies ranked by Tier 1 capital together with a size ranking based on total assets.
Wherever possible, consolidated figures are used. The size ranking is constructed by using banks’
total assets and excludes third-party item such as acceptances, guarantees and securities held with
third parties, and off-balance-sheet assets. For the same time period (2006-2010) I constructed a
size ranking, based on total assets value at the end of each fiscal year, using Bloomberg, Bureau
van Dijk’s Bankscope database and SNL financial institutions database. For each ranking
obtained, non listed banks have been dropped from the list.
Once four rankings for each year considered have been obtained, I merged them together in order
to produce a modified final ranking. It contains the biggest world listed banks by total assets
13 Data sample 13
based on the values they assumed in the period considered (2006-2010) for constructing my final
ranking
1
.
From this analysis I ended up having 110 banks from 28 different countries. However, not all the
banks in the sample respect my search criteria. Particularly, I believe that exists a relation
between the amount of dividends paid and the government bailout. I suppose that financial
institutions in a situation of financial distress increased their dividend payments just before the
government bailout. For this reason, I decided to drop from my sample those banks where the
government’s participation was relevant also before the crisis. In order to perform this task, I
investigated the ownership structure of each bank in the database. I looked at this information
historically from 2003 to 2010. Using historical information I avoided to drop banks where the
government step into the capital structure during the crisis to provide financial support (i.e.
Commerzbank). In this way I found the banks where the government’s participation has been
predominant for a stable period of time. I found information about the capital structure from two
different data sources: Bureau van Dijk’s Bankscope database and Thomson Financial. I ended
up dropping 19 banks, 6 of them were Chinese banks.
Moreover, I required that banks, in the final database, need to have available market information
for all the period covered. In particular, the market value has to be available for each fiscal year
together with other accounting information as the value of debt, of assets and of total equity. In
addition, the yearly data on free cash flow and performance indicators as ROA (Return on
Assets) have to be largely available. After this further analysis, 9 banks have been dropped and
again the majority of them has been represented by Asian banks.
My final dataset contains 82 banks (see table 1) from 23 different countries (see table 2) for a
total of 574 bank-year observations for which I have both ownership, accounting and market
data. The latter are always taken from Bloomberg, apart from the dividend variable that need a
specific explanation (see paragraph 2.2).
1
This is the reason why banks like Lehman Brothers Inc. are included in my final list