View
224
Download
1
Category
Preview:
Citation preview
Whence the privatized firm dividend premium?
Abhinav GoyalUniversity of Liverpool,
Liverpool, UK∗
Shrikant JategaonkarSouthern Illinois University,
Edwardsville, USA†
William L. MegginsonUniversity of Oklahoma,
Norman, USA‡
Cal MuckleyUniversity College Dublin,
Dublin, Ireland§
This Version: Mar. 15th, 2014
Abstract
We find that the major determinants of the payout premium of firms after pri-vatization are improved firm operating performance and a prevalence of agency costswhich are mitigated by higher pay outs. We examine up to 82,612 firm-years (up to409 privatized and 6,193 non-privatized firms) across 26 countries. The privatized firmpayout premium increases substantively in civil law countries and is inversely relatedto the proportion of closely held shares. It also increases with firm earnings, efficiencyand growth opportunities. Our main findings do not materially differ in respect to theinternational variation over time of the dividend tax penalty and across the state ofeconomic development in the country of firm privatization but they are not evidentin industry sectors with high levels of regulation. We therefore provide an economicrationale for the higher pay outs of privatized firms.
Keywords: Agency costs, dividends, privatization, operating performance, dividendtax penalty.JEL Classification: G35, L33, L25
∗E-mail: agoyal@liv.ac.uk†E-mail: sjatega@siue.edu‡E-mail: wmegginson@ou.edu§E-mail: cal.muckley@ucd.ie Corresponding author: Cal Muckley, School of Business, University
College Dublin, Blackrock, Dublin, Ireland. The authors would like to thank for suggestions GurmeetBhabra, Harjeet Bhabra, Angus Chu, Dominique Demougin, Chinmoy Ghosh, Iftekhar Hasan, Olan Henry,Narayanan Jayaraman, Steven Jones, Kathleen Kahle, Jason Laws, Brendan McCabe and Michael McKenzieparticipants at the Finance, Economics and Econometrics seminar in the University of Liverpool (Oct. 11,2013), Financial Management Association annual meeting 2013 in Chicago (Oct. 17, 2013), Indian FinanceConference annual meeting 2013 in the IIM-A, Ahmedabad (Dec. 18, 2013), Mid-West Finance Associationannual meeting 2014 in Orlando (Mar. 8, 2014). This paper is scheduled to be presented at Global FinanceConference annual meeting 2014 in Dubai (Mar. 31, 2014) and INFINITI Conference, 2014 in Prato (June9, 2014). The usual disclaimer applies.
1 Introduction
We undertake a comprehensive analysis of dividend pay outs by up to 409 privatized and
6,193 non-privatized firms from 26 countries and highlight an interesting and important
question: Why do newly privatized firms increase dividends? Previous studies examining
the change in firm characteristics around privatizations document a significant increase
in dividends by newly privatized firms (Megginson, Nash, and van Randenborgh, 1994,
Boubakri and Cosset, 1998, among others).1 The objective of this study is to better un-
derstand the economic motivations behind privatized firms’ tendency to pay a dividend
premium.2 In line with the Miller-Modigliani (1961) payout irrelevance proposition, we
account for operating performances and test whether variables associated with a privatized
firm’s incomplete contracting possibilities, financial life-cycles, information asymmetries or
taxes are of foremost importance in explicating its dividend pay outs.
The high dividend pay outs by privatized firms is evident even relative to non-privatized
firms. The difference in the amount of dividends paid by privatized and non-privatized
firms is, in fact, startling. In 2005, von Eije and Megginson (2008) find that, while the
average cash dividend payment by 4,070 non-privatized firms was euro 21 million, the
average cash dividend payment by 83 privatized firms was euro 308 million. However, the
question regarding the privatized firms’ motives behind paying such high dividends is still
unanswered in the literature. We attempt to fill this gap by asking (i) why privatized firms
pay a dividend premium and (ii) what factors allow them to do so? To find answers to
these questions, we empirically analyze the change in dividend pay outs for privatized firms
around the time of privatization and also compare them to a sample of non-privatized firms.
Changes in the firm’s objective function during privatization can induce uncertainty
in the investor’s mind about the firm’s future direction.3 While privatization is expected
1In our sample, the non-privatized firms are firms that have never been controlled by the government.Privatized firms, on the other hand, are government controlled entities that sell shares or assets to a non-government entity.
2Although we do not expect a share repurchase payout premium on part of privatized firms as thesefirms issue shares as an integral part of the privatization process, we test for it. We repeat the analysis foran aggregate sample of 394 repurchasing firms. In the full sample of firms, there is no evident repurchasepremium once well-known pay out determinants are accounted for. Only a very small number of firms fromMexico and Russia, occasionally, account for a significant amount of repurchases. However, overall, we findno evidence of a repurchase premium by privatized firms.
3While a state-owned enterprise (SOE) may pursue objectives conflicting with profit maximization, pri-vatization leads to a significant change in the firm’s ownership structure, which in turn may lead to a change
2
to bring efficiency, it also means higher capital market scrutiny, harsher product market
competition, and the possibility of increased agency conflicts between stakeholders. We
hypothesize that higher agency conflicts result into higher dividends as the shareholders of
firms that have high potential agency conflicts would demand higher pay outs in the form of
dividends; in the process minimizing the discretionary cash under the management’s con-
trol (Pinkowitz, Stulz and Williamson, 2006 and Harford, Mansi and Maxwell, 2008). We
empirically test various implications of agency theory to examine whether privatized firms
pay higher dividends to mitigate agency costs. First, the free cash flows theory (Jensen
1986) would imply that privatized firms have higher free cash flows and use dividends for
disbursement to avoid the over-investment problem. Similarly, Easterbrook (1984) sug-
gests that dividends can be used as a potential solution to agency conflicts as it subjects
the managers to market scrutiny while raising external funds. Second, we examine how
the level of firm’s ownership concentration can have an effect on the firm’s dividend policy
(Chay and Suh, 2009). While the conflict between the management and firm’s shareholders
implies that a low proportion of ownership by insiders results into higher agency conflicts,
the conflict between the controlling and minority shareholders suggests the opposite. Mi-
nority shareholders will demand higher dividends if they are concerned that the controlling
shareholders might extract rents through other mean such as salaries and perks (Chay and
Suh 2009). The substitution (outcome) hypothesis put forth by La Porta, Lopez-de-Silanes,
Shleifer, and Vishny (2000) predicts that these concerns will be exacerbated (alleviated) in
civil law (common law) countries as they offer lower (higher) protection to minority share-
holders. Therefore, third, we test whether the dividend policy of privatized firms differs
across the civil and common law countries.
Our findings support the predictions of agency theory. Specifically, as we find privatized
firm pay outs decline with an increase in the proportion of closely held shares (a proxy for the
fastidious monitoring of management by shareholders), we show evidence consistent with
the prevalence of agency costs influencing pay outs (Chay and Suh, 2009 and Aggarwal, Erel,
Ferreria and Matos, 2011). Our strongest finding, however, is that as pay outs by privatized
firms, relative to non-privatized firms, are highest in civil law countries our results provide
in the firm’s objective function (Jones, Megginson, Nash and Netter, 1999). For instance, after privatization,firms are more likely to focus on profit maximization (D’Souza, Megginson, and Nash, 2005).
3
economic support for the ‘substitution model’ of La Porta, Lopez-de-Silanes, Shleifer, and
Vishny (2000). In contrast, the findings of La Porta, Lopez-de-Silanes, Shleifer, and Vishny
(2000), in a large international sample of firms, suggests the importance of their ‘outcome
model’. Our results highlight the distinctiveness of privatized firms pay outs. Lastly, we
also find some support for the free cash flows hypothesis. We argue that the changes in
firm’s ownership structure and objective functions around privatization can lead to higher
agency conflicts and create uncertainty about the firm’s future direction. Managers of the
newly privatized firms may use dividends to alleviate the shareholder concerns and mitigate
the agency costs.4
Alternatively, the life-cycle theory of dividends suggests that there is a trade-off between
the costs and advantages of retention of internally generated capital and firms tend to
initiate dividends after reaching a certain maturity level.5 DeAngelo, DeAngelo, and Stulz
(2006) show that firms with high proportions of earned capital as a proportion of total
equity are more likely to pay dividends. We test whether the privatized firms are in a
phase of financial life cycle that makes them better candidates for disgorging the retained
earnings to shareholders. By comparing the retained earnings to total equity (RETE) for
the privatized and non-privatized firms, we show that the higher pay outs of privatized
firms are not accounted for by the life-cycle theory. In our sample, we find that while
non-privatized firms have a higher median RETE, privatized firms pay higher dividends.
In a similar vein, the maturity hypothesis suggests that as the firm moves from the growth
phase to a more mature phase of its financial life-cycle, the firm’s investment opportunity
set starts to contract and it experiences a reduction in growth and capital expenditures
(Grullon, Michaely, and Swaminathan, 2002). We test these predictions by examining the
change in total assets and growth opportunities of the privatized firms. Our findings do
4Under this scenario, managers are using dividends for signalling. However, it is not a traditional signalabout the changes in future earnings (Bhattacharya, 1979, Miller and Rock, 1985, and John and Williams,1985). Instead, managers are paying dividends to signal their willingness and attempts to mitigate theagency conflicts. Therefore, we conclude that our findings support the predictions of agency conflicts andnot the traditional signalling theory.
5It is noteworthy that there is an important theoretical linkage between the financial life-cycle phaseof a firm, agency costs and dividends. The opportunity to over-invest and fritter away free cash flows isheightened as the firm transits to a mature phase of its financial life-cycle and as management concurrentlyseeks to maximise assets under management (Jensen, 1986 and Grullon, Michaely and Swaminathan, 2002).Dividend pay out at this financial life-cycle phase transition can act as a mechanism to mitigate agencycosts.
4
not support the life-cycle theory or the maturity hypothesis in respect to explicating the
privatized firm payout premium. We not only find a significant increase in the total assets
after privatization but also a significant growth in the earnings, sales, and market-to-book
ratio.
Our findings show a strong positive link between the privatized firms’ dividend premium
and the growth in sales, earnings, and firm efficiency following privatization. There is a
paucity of studies examining the actual determinants of the payout policy of privatized
firms or explaining why exactly these firms increase dividends. This study contributes to
the literature by empirically examining exactly which factors influence the dividend policy
of privatized firms. To the best of our knowledge, we are the first to explicitly show that the
higher pay out of post-privatization firms is principally associated with improved operating
performance and firm efficiency combined with the ‘substitution model’ of the agency costs
hypothesis, and is off-set by the higher level of closely held shares in privatized firms relative
to non-privatized firms. We test the robustness of our findings by examining the sub-samples
based on (i) the extent of privatization; (ii) competitiveness of the industry; and (iii) the
level of economic development of the firm’s domestic country. We find qualitatively similar
findings across these sub-samples.
Our paper proceeds as follows: Section 2 contains a brief review of the literature.
Our sample selection process, variable definitions, and summary statistics are discussed in
Section 3. Empirical results are in Section 4, while Section 5 concludes.
2 Literature review
2.1 Payout policy
Since Miller and Modigliani’s (1961) irrelevance proposition of dividends, theories based
on information asymmetry, agency conflicts, and a firm’s financial life-cycle stage, among
others, have been put forth and empirically tested by researchers.6 It is important to note
that these theories are not mutually exclusive and may co-exist with different extents of
6See the survey article by Allen and Michaely (2003) for a detailed summary of both the theoretical andempirical literature on payout policy. In our estimation of the privatized firm dividend payout premium wedo not account for the catering theory of dividend payout determination, since catering incentives have beenshown to lose their significance when accounting for life-cycle and risk variables (DeAngelo, DeAngelo, andStulz, 2006, Denis and Osobov, 2008, von Eije and Megginson, 2008).
5
influence in different settings.
2.1.1 Agency costs theory
The agency costs theory suggests that the costs associated with prospective agency conflicts
can affect the payout policy of the firm. If shareholders can minimize the free cash flows that
management controls, for instance by a limited disciplinary action, it becomes more difficult
for management to pursue negative net present value investments (Jensen and Meckling,
1976, Easterbrook, 1984, and Jensen, 1986). The free cash flow hypothesis implies that firms
disburse cash to shareholders to mitigate the potential over-investment by management and
to increase share price, for example, to reduce the cost of raising capital in the market in
future. This relation between payout and investment policies is a clear infringement of an
assumption of the Miller and Modigliani’s (1961) irrelevance proposition of dividends.
The evidence on agency theory is mixed with respect to the importance of its influence on
pay outs. While Lang and Litzenberger (1989) and Grullon, Michaely, and Swaminathan
(2002) find evidence supporting the predictions of the Jensen and Meckling (1976) and
Jensen (1986) free cash flows theory, more recently, Chay and Suh (2009) do not find support
for the agency theory of pay out, when accounting for cash flow uncertainty. Another aspect
of agency costs theory is to examine the effect of shareholders’ rights on the firm’s payout
policy. La Porta, Lopez-de-Silanes, Shleifer, and Vishny (2000) compare the strength of
corporate governance mechanisms across 33 countries and test its effect on dividend policies
in these countries. They conclude that firms in common law countries are more likely to
pay dividends than those in civil law countries because the common law system provides a
better investor protection and stronger corporate governance. Taking these points together,
with respect to the agency theory of pay outs, it is clear that dividends in themselves can
be good news as they can serve to allay agency costs which might otherwise serve to fritter
away a firm’s free cash flows.
An alternative to altering firm pay outs and a possibly more effective mechanism for
limiting free cash flows is to increase the level of debt (Jensen, 1986). This is especially
the case when an increase in leverage can act as a substitute for an expensive decrease
in dividends to finance an improved investment opportunity set. Another mechanism for
the mitigation of management shareholder agency costs is the extent of closely held shares.
6
The greater the proportion of closely held shares, especially by large firms, can act as
a monitoring mechanism and can also substitute for firm pay outs (Chay and Suh, 2009
and Aggarwal, Erel, Ferreria and Matos, 2011). Finally, with respect to the free cash flow
hypotheses, it is expected to find a positive relation between a privatized firm’s cash holding
and pay out (Pinkowitz, Stulz and Williamson, 2006 and Harford, Mansi and Maxwell,
2008). A higher cash holding is consistent with a greater scope to fritter away free cash
flows which is offset by higher pay outs.
2.1.2 Life-cycle theory
The theory that has received strong empirical support in recent times is the life-cycle theory
of dividends. DeAngelo, DeAngelo, and Stulz (2006) argue that there is a trade-off between
the costs and advantages of retention of internally generated capital, which evolves with
respect to the financial life cycle of the firm. Using the earned/contributed capital mix,
they measure the extent to which the firm is self-financing or reliant on external capital.
DeAngelo, DeAngelo, and Stulz (2006) suggest that higher levels of retained earnings to
total equity indicate that the firm has become a better candidate to initiate dividends and
show that a large fraction of such firms actually pay dividends.7 Using a sample of world-
wide firm-level data, Denis and Osobov (2008), Chay and Suh (2009) and Brockman and
Unlu (2011) report findings that further corroborate the life-cycle theory. They find that
the earned/contributed capital mix is an important determinant of payout policy in many
countries. Furthermore, Chay and Suh (2009) test the effect of cash flows uncertainty,
proxied by stock return volatility, on dividends by using worldwide firm-level data.8 Con-
sistent with the predictions, they find a strong negative impact of cash flow uncertainty,
independent of retained earnings to total equity, on the amount of dividends as well as the
probability of paying dividends across countries. Finally, Brockman and Unlu (2011) show
a firm’s disclosure environment plays a significant role in dividend pay outs through its
effect on agency costs. They confirm an agency-cost inclusive life-cycle theory of dividends.
7However, in a recent paper, Banyi and Kahle (2013), provide a criticism of the earned to contributedcapital mix as a life-cycle proxy variable for firms listed in the United States.
8Lintner’s (1956) survey study indicates that managers view stability of earnings as an important factorin dividend decisions. More recently, Brav, Graham, Harvey, and Michaely (2005) also find that two-thirdsof the CFOs of dividend-paying firms consider stability of future cash flows as a significant determinant ofdividend policy.
7
2.1.3 Traditional signalling theory
The traditional signalling theory, which is based on information asymmetry, implies that
managers use payout policy to convey information regarding the future earnings changes of
the company. The associated signalling models by Bhattacharya (1979), Miller and Rock
(1985) and John and Williams (1985) therefore imply that higher dividend pay outs can
indicate confidence on the part of firm management in the firm’s future earnings improve-
ments to the market.9 In line with findings in Von Eije and Megginson (2008), we show that
firm transparency improves as firms increase the frequency of earnings reporting immedi-
ately following privatization. Hence, newly privatized firms are unlikely to use dividends to
signal future changes in earnings as there is a marked improvement in the available informa-
tion regarding the firms’ expected earnings following privatization. Indeed, the signalling
theory, which is based on information asymmetry, has faced some challenges when put to
empirical tests.
In summary, we do not consider this signalling theory as a likely explanation for the
dividend payout premium associated with privatized firms for five main reasons informed
by previous findings in the literature. First, the relation between dividend changes and
subsequent earnings changes is generally the opposite to what the theory predicts (Watts,
1973, Healy and Palepu, 1988 and Grullon, Michaely, Benartzi and Thaler, 2005). Second,
cross-sectional studies indicate that large profitable firms with the least evident informa-
tion asymmetries pay the vast majority of dividends and are more likely to pay dividends
(DeAngelo, DeAngelo and Skinner, 2004 and von Eije and Megginson, 2008). Third, in
their survey paper, Brav et al. (2005) report that majority of CFOs do not use dividends
as a signaling mechanism. Forth, there is a significant price drift in subsequent years
which is difficult to reconcile with the assumption of rationality in the information asym-
metry based signalling models (Michaely, Thaler and Womack, 1995 and Grullon, Michaely
and Swaminathan, 2002). Finally, we elect not to use the information asymmetry based
signalling model due to a long-standing theoretical rationale. The cited information asym-
metry based signalling models (except John and Williams, 1985 which allows a distinction
9Bhattacharya’s (1979) model takes the cost of share issuance as the cost of the signal. Miller and Rock’s(1985) model assumes that the signalling cost is the positive net present value of investment forgone andJohn and Williams (1985) present a model in which taxes are the dissipative cost.
8
based on tax rates) assume that dividends and repurchases are perfect substitutes. There
is, however, considerable empirical evidence of important distinctions in the information
content of different payout channels in respect to firm risks (von Eije, Goyal and Muckley,
2014).
Instead dividends may act as an inadvertent reflection of the firm’s changing circum-
stances (Miller and Modigliani, 1961 and Miller and Rock, 1985). Dividends, as a by-
product of their residual nature after investment, may reflect the changes in contemporane-
ous earnings of the firm and, independently of the classic signalling models, as earnings can
be positively auto-correlated over time, and this association can link dividends to future
earnings.
2.2 Privatization and pay outs
During the 1980s and 1990s, extremely large companies in the European Union (EU) were
privatized. As a result, the literature on privatization has been rapidly growing over the
last three decades.10 A series of papers have examined the effect of privatization on var-
ious firm characteristics and performance measures. For instance, Megginson, Nash, and
van Randenborgh (1994) compare the pre- and post-privatization financial and operating
performance of 61 companies from 18 countries during the period 1961 to 1990, and report
a strong performance improvement and increase in capital spending in the privatized firms.
They also find that, after being privatized, firms significantly lower their debt and increase
their dividends. Using a sample of 21 developing countries, Boubakri and Cosset (1998)
also examine the change in financial and operating performance of 79 privatized firms dur-
ing the period 1980 to 1992 and they find results similar to those reported by Megginson,
Nash, and van Randenborgh (1994). D’Souza and Megginson (1999) examine a sample of
85 privatizations in 28 industrialized countries during 1990 through 1996 and report similar
results of post privatization performance improvements. They also find that firms in the
noncompetitive industries exhibit significantly greater increases in dividend pay outs, firm
efficiency, profitability, and output and larger reductions in leverage. Boubakri, Cosset and
Guedhami (2005) and D’Souza, Megginson and Nash (2005) corroborate the earlier findings
10See the survey article by Megginson and Netter (2001) for a detailed review of the literature on priva-tization.
9
but also their findings imply that the implications of privatization in developing markets
are influenced by macro economic reforms, financial and trade liberalization, and corporate
governance and thus that ‘privatization in developing countries indeed obeys to particular
constraints and has a dynamic of its own’.
The fact that privatized firms pay significantly higher dividends after privatization and
relative to their non-privatized counterparts is well established in the literature. However,
it is still unclear why these firms pay such high dividends and which factors influence this
difference in the payout policies. We test whether the difference between dividend pay outs
stems principally from differences in signalling motives, agency conflicts or the phase of the
financial life-cycle between the two groups.
3 Data and variable descriptions
Our dataset includes a total of 6,602 listed firms (82,612 firm-years) out of which 409 are
privatized firms (5,406 firm-years). The sample consists of firms listed on exchanges (and
headquartered) in 26 countries globally. The data is primarily obtained from Worldscope.
Our sample commences in 1990 and extends to 2011.11 We apply sample restrictions
consistent with prior studies. Consistent with recent literature on international corporate
pay out determination (e.g. von Eije and Megginson, 2008, Denis and Ososbov, 2008,
Chay and Suh, 2009 and Brockman and Unlu, 2009), we exclude foreign firms, American
Depository Receipts (ADRs) and firms with negative dividends and sales. As a final filter,
we exclude countries with less than three privatized firms.
Worldscope defines privatization as a government or government controlled entity that
sells shares or assets to a non-government entity. This definition of privatization includes
both direct and indirect sales of up to a 100% stake to an identifiable buyer and floatation
of stock on a stock exchange. Non-privatized firms are firms that have never been controlled
by the government. We source a unique identifier for the privatized firms and their year of
privatization in the merger and acquisitions section of Worldscope. Due to data availability
constraints, we have the year of privatization for 302 firms out of the 409 firms that we
examine. Specifically, 245 of these firms were privatized during our sample period, 1990
11The coverage of firm-specific data outside the United States prior to 1990 is limited (Denis and Osobov,2008).
10
to 2011. When we study the pay out determination 3-years pre- and post-privatization,
our dataset is constrained to a sub-sample of 92 privatized firms for which we have data
available for all the necessary variables.
In Appendix 1, we provide a detailed description of the variables we use in our study.
PVT is a dummy variable that indicates a privatized company. Our variable for cash
dividends (DIV) is the total real amount distributed as cash dividends by the firm in 1990
US dollars. Consistent with prior literature on corporate pay out, we adopt several firm-
specific characteristics to estimate the determination of firm pay outs. In line with Fama and
French (2001) and Dennis and Osobov (2008) we use the market capitalization (MV) and
annual percentile ranking based on market capitalization (SIZE) as a proxy to measure firm
size. Next, in order to study firm-level profitability we use earnings before interest and after
tax to total assets; ER (von Eije and Megginson, 2008) and net income; NI (Brockman and
Unlu, 2009). We use retained earnings to total equity; RETE (DeAngelo, DeAngelo, and
Stulz, 2006) and firm-level cash holding; CASH (Aggarwal, Erel, Ferreira, and Matos, 2011)
as a proxy for firm financial life cycle phase and firm liquidity. Following Chay and Suh
(2009) we use the fraction of common stock owned by insiders; CLOSE as a proxy variable
for agency conflicts and change in ownership concentration pre- and post-privatization. To
control for the income risk of the firm, we include the standard deviation of last three years’
net income scaled by each year-specific total assets; NI Risk (von Eije and Megginson 2008).
As a proxy for the firm’s growth opportunities, we construct an annualized real change in
total assets (G TA) and market-to-book value (MTBV) of the firm (Fama and French, 2001
and Denis and Osobov, 2008) and an annualized real change in sales; G Sales (La Porta,
Lopez-de-Silanes, Shleifer, and Vishny, 2000). Following Brockman and Unlu (2009), we
control for the firm-level leverage (LR), which can allay prospective agency costs of free
cash flows due to associated monitoring by the lending institution (Jensen, 1986). We use
the frequency of financial reporting (ERF) as a proxy for firm transparency (von Eije and
Megginson, 2008). Following Megginson, Nash, and van Randenborgh (1994) and Boubakri
and Cosset (1998), we incorporate sales to employees (Sales Emp) and total employment
(Emp) as a parameter to test the firm-level efficiency.
Finally to account for the investor rights, we follow La Porta, Lopez-de-Silanes, Shleifer,
and Vishny (1997) and include a dummy variable (COM=1) for common law countries in
11
our sample. We also include a time trend variable (YEAR) to account for a deterministic
time trend in payout amounts. The country-specific consumer price indices are used to
deflate the nominal firm-specific accounting and financial data into real 1990 US$. We use
US$ as a common currency numeraire by converting the local currency unit values into
US$ using the year-end conversion rate. To adjust for the extreme outliers, we winsorize
variables defined as ratios, namely earnings ratio (ER), retained earnings to total equity
(RETE), cash holding (CASH), ownership concentration (CLOSE), income risk (NI Risk),
growth in total assets (G TA), market-to-book value (MTBV), growth in sales (G Sales),
leverage ratio (LR) and sales- to-employee ratio (Sales Emp) at the top and lower 1% of
their respective distributions (Aggarwal, Erel, Ferreira, and Matos, 2011).
The sample distribution of privatized firms across 26 countries is reported in Panel A of
Appendix 2. The dataset contains privatizations from viz. India and Mexico (3 privatized
firms each), Malaysia and Norway (4 privatized firms each), Chile and China (5 privatized
firms), Argentina, Australia, Greece, Netherlands and New Zealand (6 privatized firms
each), Austria and Portugal (7 privatized firms each), Finland and Turkey (9 privatized
firms each), Peru (10 privatized firms), Spain (14 privatized firms), Sweden (16 privatized
firms), Italy (19 privatized firms), Poland (20 privatized firms), Brazil and Germany (28
privatized firms each), Russia (31 privatized firms), France (41 privatized firms), Canada
(57 privatized firms), and the U.K. (59 privatized firms). The average dividends paid in
each country are also reported.
Following Megginson, Nash, and van Randenborgh (1994) and D’Souza and Meggin-
son (1999), Panel B of Appendix 2 separates the privatized firms into control and revenue
privatization.12 While the average dividends paid by firms privatized to concede control
is $107.38 million, the average dividends for the firms which were privatized for revenue
is $298.12 million. Panel C separates the privatized firms into competitive versus non-
competitive industry groups (Megginson, Nash, and van Randenborgh, 1994, Boubakri
and Cosset, 1998, D’Souza and Megginson, 1999).13 While the average dividends paid by
12A control privatization is where government reduces its stockholding to below 50% and a revenueprivatization is where governments retains majority ownership, i.e., more than 50% ownership.
13A competitive industry is one where the firm is subject to international product market competition. Anon-competitive industry is one where the firm is not subject to international product market competition,i.e., financial institutions, banks and utilities.
12
firms in the competitive industries is $128.02 million, the average for the firms in the non-
competitive industries is $213.32 million. Panel D of Appendix 2 shows that out of the total
409 privatized firms in the sample, 124 firms are from the emerging markets (developing
countries) and the other 285 firms are from developed countries. This disaggregation of our
set of privatized firms across developing and developed markets is motivated by the distinc-
tive implications of privatization in developing markets (Boubakri, Cosset and Guedhami,
2004, D’Souza, Megginson and Nash, 1999). In our sample of privatized firms, while the
average dividend for firms in the developed countries is $155.42 million, the average for the
firms in the developing countries is about $113.88 million. Panel E reports the number of
privatized firms and the average dividends by industry in our sample.
4 Empirical findings
4.1 Pre- and post-privatization
We initially focus on just privatized firms and report the variables used in the study pre-
and post-privatization. Figure 1 shows the change in certain firm characteristics related to
growth opportunities (MTBV, G TA and G Sales), profitability (ER) and dividends (DIV
and DIV EBIAT) for -3 to +3 years relative to the privatization year.14 Table 1 reports
the mean and median for all the firm characteristics during the 3-years pre- and post-
privatization, and whether differences are significant. As shown in figure 1, along with an
increase in dividends, we find an increase in profitability (ER) and growth opportunities
(G TA and G Sales) post privatization and we find no decline in the market to book value
(MTBV). The proportion for dividend payout (DIV) reported in table 1, shows that 77%
of our sample firms increase dividends after privatization.15 Along with increases in payout
(DIV, DIV EBIAT, and DIV NI), a large proportion of firms exhibit a significant increase
in the profitability (ER),16 asset growth (G TA), sales growth (G Sales), and firm efficiency
14We use up to 245 privatized firms in our sample to construct figure 1. The exact number of firms foreach year varies depending upon the data availability.
15However, we follow previous studies such as Boubakri and Cosset (1998), D’Souza and Megginson(1999), Boubakri, Cosset, and Guedhami (2005) and D’Souza, Megginson, and Nash (2005) and applystricter restrictions to calculate the numbers reported in table 1. Specifically, we require the firms to haveat least two years of consecutive data during both pre- and post-privatization periods. The findings in table1 are therefore for a constrained data set of between 92 and 121 privatized firms which satisfy the datarequirements. These firms were privatized between 1992 and 2009.
16EBIAT, NI, and Sales also increase post privatization though the results are not reported here.
13
(Sales Emp).
[Please insert figure 1 about here]
As expected, we find a significant decrease in ownership concentration (CLOSE) after
privatization. While the average ownership concentration is 74.67% before privatization,
it decreases to 52.66% during the 3-years after the privatization. We find an increase in
the firm risk (NI Risk) after privatization. On one hand, this post privatization increase
in the standard deviation of net income could reflect the high uncertainty during firm’s
transition from SOE to a privatized firm. On the other hand, it could be a consequence of
the rapid growth in earnings exhibited by the newly privatized firms. As reported in table
1, unlike in previous studies (Meginnson, Nash, and van Randenborgh, 1994, Boubakri and
Cosset, 1998, D’Souza and Megginson, 1999) we find no significant change in leverage (LR)
following privatization, which is indicative of no new monitoring by lending institutions
after firm privatization. We do find a significant increase in financial reporting frequency
(ERF) from before privatization (1.90) to after privatization (2.68), which is consistent with
the newly privatized firm wanting to reduce the information asymmetry and uncertainty
in the investor’s mind regarding the firm’s future direction.17 Overall, the comparison
between variables during pre- and post-privatization periods suggests that privatization is
associated with increases the amount of cash dividends (DIV, DIV EBIAT, DIV NI), firm
efficiency (Sales Emp), firm size (MV, SIZE), firm transparency (ERF), growth (G TA,
MTBV, G Sales), income risk (NI Risk), profitability (ER) and retained earnings (RETE).
Privatization results in a decrease in firm ownership concentration (CLOSE).
[Please insert table 1 about here]
We extend our analysis by partitioning the sample of 121 privatized firms into subsam-
ples based on three criteria: extent of privatization, the level of regulation in the industry,
and the level of development in the country to compare the performance changes for these
firms during the 3 years pre- and post-privatization. Findings for the subsamples are re-
ported in table 2. Panel A reports the comparison of subsamples based on the extent of
17Alternatively, the increase in ERF could be a result of higher scrutiny by the capital markets andconsequently investors’ expectations for all publicly traded companies to report earnings frequently.
14
privatization. Control privatization refers to cases where the government reduces its stock-
holding to below 50% and the Revenue privatization refers to cases where the government
retains ownership of more than 50%. We have 38 control privatization and 83 revenue pri-
vatization in our sample. We find that while the dividend (DIV) increases for both groups,
the increase in pay out is more pronounced for revenue privatization and this is consistent
with a possible attempt to allay agency costs evident after revenue privatization. We find
a significant increase in firm size (MV, SIZE) and profitability (ER) for both subsamples
following privatization. As expected, we find the ownership concentration (CLOSE) to
be lower after control privatization. We find that the firm risk (NI Risk) increases only
for control privatization, but the increase in growth and frequency of financial reporting
(ERF) following privatization is significant for both groups. We find no significant change
in the leverage ratio (LR) of firms after privatization. The increase in median firm effi-
ciency (Sales Emp) is significant for both groups, but the increase in mean efficiency is
more prominent in the revenue group.
Panel B reports results for subsamples based on the industry’s competitiveness. The
companies are separated into firms that are subject to international product market com-
petition (90 sample firms) versus firms in non-competitive industries such as financial insti-
tutions and utilities (31 sample firms). While the dividend pay outs (DIV) increase in both
groups, consistent with D’Souza and Megginson (1999), the changes are more prominent
for the non-competitive industry group. We find an increase in firm size (MV, SIZE), prof-
itability (ER), and proportion of retained earnings in total equity (RETE), and ownership
dilution (CLOSE) for both subsamples following privatization. We also find a significant
increase in growth opportunities (G TA, MTBV, G Sales), financial reporting frequency
(ERF), and firm efficiency (Sales Emp) for both groups. In the non-competitive indus-
try sector we find an increase in firm leverage ratios (LR) after privatization. Finally, a
comparison between privatized firms in emerging versus developed countries reveals some
interesting findings. While the pay out (DIV) increases in both groups, the proportion of
firms that increase dividends is larger in the developed countries (84%) as compared to
the emerging countries (64%). The increase in scaled payout (DIV EBIAT and DIV NI) is
only significant in the developed countries. We find an increase in the earnings (ER), net
income risk (NI Risk), and firm efficiency (Sales Emp) in both groups. However, the re-
15
tained earnings (RETE), ownership concentration (CLOSE), and market to book (MTBV)
change only for the firms in the developed countries. Only in emerging markets is there a
significant increase in firm leverage ratios (LR) after privatization. Overall, we find that
the post-privatization increase in dividends (DIV) is accompanied by an improvement in
the firm’s earnings (ER), operating efficiency (Sales Emp), and sales (G Sales) and a de-
crease in ownership concentration regardless of how the sample is arranged into categories
of privatized firms.
[Please insert table 2 about here]
4.2 Comparison between privatized and non-privatized firms
Table 3 reports the annual average and median cash dividends (DIV) paid by privatized
versus non-privatized firms in the sample during 1990 - 2011. The proportion of dividend
payers in both the groups is also reported for each year. Among the privatized firms, the
proportion of dividend payers is as low as 60% in 2004 and as high as 85% in 1990. The
corresponding numbers for the non-privatized firms are 57% in 2004 and 87% in 1990.
Consistent with Fama and French (2001), there is a decline in the proportion of dividend
payers over time. The decline in the proportion of dividend payers over time has been
slightly greater for the non-privatized firms. Over our sample period, 72% of the privatized
firms have been dividend payers. As compared to the non-privatized firms, overall, a
greater proportion of privatized firms pay dividends. The mean cash dividends paid by
privatized firms are significantly higher than those by non-privatized firms in each year of
our sample. There is an evident upward trend in the dividends paid by privatized firms
from 1990 through 2011.18 The average dividends paid by non-privatized firms have also
increased over time, but at a much slower pace. While the average annual dividend by
non-privatized firms is $73.34 million in year 2011, the average amount distributed by a
privatized firm in the same year is $305.47 million. Similarly, over the whole sample period,
the mean dividend for privatized firms is $143.59 million as compared to $35.50 million for
non-privatized firms.19 Table 3 also reports the annual medians. Each year, the median
18Note that these are real dividends in 1990 US Dollars.19Additionally, to show the substantial increase in dividends for privatized firms over time we can com-
pare the dividends for two groups in 1990 and in 2011. In 1990, privatized firm dividends are 1.83 times
16
dividends paid by privatized firms are significantly greater than those paid by non-privatized
firms. The comparison between the two groups clearly indicates that the privatized firms
not only pay significantly higher dividends than the non-privatized firms, but also a larger
proportion of privatized firms tend to pay dividends.
[Please insert table 3 about here]
Table 4 reports summary statistics for the dependent and control variables used in the
study for the privatized and non- privatized firms. It is not surprising that the mean and
median for the unscaled dividend variable (DIV) show that the privatized firms pay much
larger dividends than non-privatized firms. While the median (average) dividend pay out
by privatized firms in our sample is $5.82 ($143.59) million, the median (average) pay out
by firms that have never been state owned is only $1.31 ($35.50) million.20 The median
dividend paid by privatized firms is almost 4.5 times ($5.82 / $1.31) of the median dividend
by the non-privatized firms. Some of the difference between dividend pay outs by these
two groups can be explained by the differences in their sizes. Comparing the market value
of equity (MV) and annual percentile size ranking (SIZE) for the two groups, it is evident
that privatized firms tend to be significantly larger in size. Therefore, we also analyze the
dividend pay outs adjusted for profitability (DIV EBIAT, DIV NI). Again we find that the
median DIV EBIAT and DIV NI are significantly higher for the privatized firms, which
suggests that the privatized firms pay out a significantly higher proportion of their earnings
as cash dividends.21 To study what influences this difference in pay outs across the two
groups, we next compare various firm characteristics and the factors that are known to
affect dividend policy.
(39.22/21.43) higher than for non-privatized firms. The same ratio in 2011 is 4.17 times (305.47/73.34).There is a similar increase in the medians of these ratios of privatized to non-privatized firms’ dividend payouts (from 2.46 times in 1990 to 7.4 times in 2011).
20von Eije and Megginson (2008) also compare the unscaled dividends in their paper (page 357) wherethey examine the impact of privatization on dividend payments and they show substantively similar resultsfor firms in the European Union.
21We acknowledge that the means for the scaled dividend variables are similar across the two groups.However, given the differences in the samples sizes and some of the outliers, we think that comparing themedians is a better approach.
17
First, we examine the profitability across the two groups and find that the mean and
median of the earnings ratio (ER) for the privatized firms are significantly higher.22 This is
an interesting finding as it suggests that dividends paid by privatized firms are high not only
because they pay out a higher proportion of earnings (median DIV NI and DIV EBIAT),
but also because the firms are significantly more profitable. In addition, we compare the
level of firm efficiency across the privatized and non-privatized firms. Following Megginson,
Nash, and van Randenborgh (1994) and Boubakri and Cosset (1998), we incorporate sales-
to-employees (Sales Emp) and total employment (Emp) as a parameter to test the firm-
level efficiency. We find the Sales Emp ratio to be higher (albeit not significantly higher)
for privatized firms. This is our first, albeit tentative, evidence of an association between
the dividends and firm profitability and efficiency of privatized firms.
Next, we use the retained earnings to total equity (RETE) ratio to proxy for liquidity
and to test the effect of the life-cycle theory of dividends. The life-cycle theory of divi-
dends (DeAngelo, DeAngelo, and Stulz, 2006) predicts that firms with higher proportions
of earned equity in their total equity should pay higher dividends. Using a sample of firms
from six developed countries, Denis and Osobov (2008) show that dividend payers exhibit a
higher retained earnings to total equity (RETE) ratio as compared to non-dividend payers.
We compare the retained earnings ratio of privatized versus non- privatized firms from 26
countries and find that while the mean retained earnings ratio is insignificantly higher for
the privatized firms, the median is actually significantly higher for non-privatized firms.
Therefore, the higher dividend pay out by privatized firms does not support the life-cycle
theory predictions. These univariate findings for the retained earnings ratio cannot explain
why the privatized firms tend to pay higher dividends. Further, we use growth in total
assets (G TA), market-to-book ratio (MTBV), and growth in sales (G Sales) to estimate
firm growth opportunities. We find no significant difference in G TA across privatized and
non-privatized firms. However, both the average and median MTBV and G Sales are sig-
nificantly higher for the privatized firms. The life-cycle theory and the maturity hypothesis
predict that firms with lower growth opportunities are more likely to pay dividends. Con-
trary to these predictions, we find that the privatized firms not only pay higher dividends,
22We also find that Earnings before interest and after tax (EBIAT), net income (NI) and Sales are alsosignificantly higher for privatized firms (not reported here).
18
but also have higher growth opportunities.
Following the agency costs theory of dividend determination, if new leverage is con-
sidered to increase the external monitoring of the firm, it reduces the need to disgorge
cash flows to shareholders as dividends (Jensen, 1986). As a result, we can expect, that
the leverage channel can be used either as a substitute for or to work in conjunction with
higher dividends. We find that privatized firms tend to have higher leverage. The higher
leverage ratio for privatized firms could be because government owned firms usually exhibit
higher debt levels. The non-privatized firms in our sample have never been controlled by
government and hence are likely to exhibit lower debt ratios and also tend to pay higher
dividends. It is interesting to note, however, that we do not find a reduction in leverage of
firms in the three year period after privatization.
The variable CLOSE estimates the ownership concentration of a firm. We find that the
ownership in privatized firms is more concentrated as compared to the non-privatized firms.
Specifically, we find the average ownership concentration for privatized and non-privatized
firms to be 62.24 and 46.31, respectively. This is not unexpected as the privatized firms
have been controlled by government in the past and the ownership gets dispersed overtime
after the firm has been privatized. On the contrary, the non-privatized firms in our sample
have never been controlled by the government and hence exhibit lower ownership concen-
tration. Chay and Suh (2009) predict a negative relation between ownership concentration
and dividends. Similarly, Megginson, Nash, and van Randenborgh (1994) suggest that if
the firm’s ownership is dispersed among small investors, none of whom have incentives to
monitor; shareholders are likely to demand higher dividends to reduce agency costs. By that
logic, the group with lower ownership concentration (non-privatized firms in our sample)
would be expected to pay higher dividends, if agency costs across groups are comparable.
However, one can argue that the higher concentration of ownership by insiders is a sign of
management entrenchment and higher agency problems. The privatized firms are, hence,
more likely to pay higher dividends. The difference in ownership concentration across the
two groups can explain why privatized firms pay higher dividends.
Finally, we examine the income risk (NI Risk) and the frequency of financial reporting
(ERF) of our sample firms. We find no significant difference in the medians for these two
19
variables across the two groups. While the average reporting frequency is slightly higher for
privatized firms (2.68) as compared to the non-privatized firms (2.52), the average income
risk is higher for non-privatized firms. Given these findings, it is unlikely that either of
these firm characteristics can explain the difference in the dividends. Therefore, the overall
univariate comparison between the privatized and non-privatized firms indicates that the
higher dividend pay outs by privatized firms might be a function of their significantly
better profitability (ER), efficiency (Sales Emp), and investment opportunities (G Sales),
while they can be simultaneously explained by way of a mechanism, together with closely
held shares (CLOSE) and firm leverage (LR), to offset higher prospective agency costs.
[Please insert table 4 about here]
4.3 Multivariate analysis of the impact of privatization
We initially specify random effect models, for cash dividends paid, DIV (the natural log of
dividends), using the data for only privatized firms during the 3 years before and 3 years
after the privatization.23 The findings, reported in table 5, further give us an insight into
exactly which factors influence the dividend policy of firms post privatization. We report
the coefficient for each factor and also the difference, ‘Difference’, in coefficients for the pre-
and post-privatization periods.24
The results reported in table 5 corroborate our findings that the post-privatization
firm earnings (ER) and growth in sales (G Sales), growth in total assets (G TA), leverage
(LR) and the extent of information asymmetry (ERF) are major determining factors of the
dividends by newly privatized firms. The retained earnings to total equity (RETE) is not
significantly more positive in the post-privatization period and neither is the size (SIZE)
of the firm once the dividend tax penalty is accounted for. These latter findings constitute
evidence of an absence of influence of the financial life-cycle maturation of the privatized
23Due to data restrictions, we use 92 privatized firms for this analysis. In table 1, which contains univariateanalysis, the number of firms varies from 92 to 121. In the multivariate (regression) analysis, however, thefirm count is limited to 92.
24The ‘Difference’ column corresponds to the difference between the 3-year post privatization regressioncoefficients and the coefficients in the 3-year pre privatization period, estimated using interaction variables.The dummy variable in the interactions equals to 1 for the 3-year post privatization period and zero forthe 3-year pre privatization period. All independent variables are multiplied with this dummy and thecorresponding coefficient is indicated as ‘Difference’.
20
firm on its dividends.
The negative influence of a one percent increase in sales growth (G Sales), on divi-
dend pay outs reverses after privatization, from a -0.012% decrease in dividend pay outs
before privatization to a 0.028% greater increase in dividend pay outs after privatization.
Furthermore, this emergent positive association of sales growth is robust to the dividend
tax penalty as illustrated in panel B. As earnings from sales growth increase, firm pay
out increases too. In the same vein, the earnings ratio (ER) exhibits a positive influence
on the dividend pay out after privatization. After privatization, a one percent increase in
the earnings ratio (ER) is associated with a 0.05% increase in dividend pay out, and this
effect is also robust to the dividend tax penalty. Prior to privatization, there is a positive
association between growth in total assets (G TA) and pay out. In the post-privatization
period, however, there is a marked reversal of this influence. For a one percent growth in
total assets (G TA) there is a reduction of 0.04% of dividend pay outs, and this effect is
also robust to the dividend tax penalty. Our dividend tax penalty (DTP) is constructed
following Poterba and Summers, 1984 and Jacob and Jacob, 2013.25 As privatized firms
invest in assets, they pay out less. Turning to information asymmetry, a once per annum
increase in the earnings reporting frequency (ERF) is associated with a 0.26% reduction
in the privatized firm pay out. Hence, a reduction in information asymmetry can substi-
tute for privatized firm pay outs. It is also worthwhile noting that when the dividend tax
penalty is accounted for, leverage (LR) is positively associated with firm pay out. This is
consistent with leverage (LR) which can be used together with high dividends to discipline
management to pursue exclusively positive net present value investments. The dividend tax
penalty is shown to be internationally associated with no statistically significant marginal
influence on dividend pay outs for privatized firms in the post privatization period.
It is finally worthwhile noting that privatized firms in civil law countries pay out more,
even accounting for well-known determinants of pay out, and this effect strengthens post
privatization, albeit not significantly (mere statistical power implication).26 This finding
25The same results hold using other proxies for the dividend tax viz. a dividend tax preference (La Porta,Lopez-De-Silanes, Shleifer, and Vishny, 2000) and weighted average dividend tax preference (Becker, Jacoband Jacob, 2013). The results are available from the authors on request.
26This corroborates our finding in model III and model IV of tables 6 and 7, that the privatizationpayout premium is relatively small in common law countries (COM * PVT) despite the superior minorityshareholder rights in these countries.
21
supports the so-called ‘substitution’ free cash flow hypothesis as it suggests that an improve-
ment in minority shareholder rights, in common law (COM) relative to civil law countries,
implies a reduction in the dividend pay out as these minority shareholder rights substi-
tute for dividend pay outs (La Porta, Lopez-de-Silanes, Shleifer, and Vishny, 2000). Taking
these findings in table 5 together, we have evidence of the importance of improved operating
performance in explicating the privatized firm pay out premium and tentative evidence in
respect to information asymmetries and associated agency costs but no evidence in support
of the life cycle theory.
[Please insert table 5 about here]
To further examine the impact of privatization on the dividend policy, we simultaneously
study the determination of dividends in both privatized and non-privatized firms. We use
random effect panel regression models to regress the natural log of dividend pay out (von
Eije and Megginson, 2008) on a wide set of determinants of dividend policy established in
the literature.27 The model includes all the determinants and the interactions used in the
model specifications of table 5. We include both the privatized and non-privatized firms for
this analysis, with a dummy (PVT) variable to identify privatized firms. This also enables
us to empirically test, using a difference-in-differences parametric regression methodology
(Ashenfelter and Card, 1985), whether interactions between the privatization dummy and
other variables significantly impact the dividend policy of our sample firms. The findings
are reported in table 6. In Model I, we test the impact of privatization on the cash dividend
pay out. Consistent with the findings reported in earlier tables, we find a significant positive
relation between privatization and dividend pay outs. This confirms that privatization has
a first order effect on the dividend policy.
In Model II, reported in table 6, we add the variables that proxy for different factors
that have been shown in the literature to have an impact on a firm’s dividend policy. We
find a positive relation between dividends and firm size (SIZE) and cash holdings (CASH),
which is consistent with the notion that larger firms with higher cash holdings pay more
27We also use scaled, to total assets and sales, adjustments of dividend pay out and we show similarfindings. The determination of scaled dividend pay outs is also principally associated with changes inprivatized firm earnings and firm efficiency and with a view to allaying prospective agency costs post-privatization. The results are available from the authors on request.
22
dividends. The life-cycle theory of dividends suggests that firms with higher proportion of
retained earnings in their total equity are more likely to pay dividends. However, for our
sample, we find a negative (albeit small) relation between retained earnings to total equity
(RETE) and dividends. This suggests that the firms in our sample do not pay dividends
because they have reached a certain stage in their life-cycle in which they are more likely to
distribute the cash flows to the shareholders. While we find a significant negative relation
between ownership concentration (CLOSE) and cash dividends paid by our sample firms,
we find a positive relation between the income risk and dividends. The coefficient for
market to book (MTBV) suggests a negative relation between growth opportunities and
dividends. Following von Eije and Megginson (2008), we use the frequency of financial
reporting (ERF) as a proxy for firm transparency. Increased frequency of financial reporting
should increase transparency and hence reduce the information asymmetry for the firm,
and thereby increase investors‘ capacity to monitor the firm. Wood (2001) suggests that
improvement in reporting and corporate governance would make investors less focused on
dividends. Consistent with these predictions, we find a significant negative relation between
dividends and the frequency of financial reporting (ERF). We find that dividends increase
with increase in the earnings (ER) and firm efficiency (Sales Emp). Consistent with La
Porta, Lopez-de-Silanes, Shleifer, and Vishny (2000), we find that non-privatized firms in
common law countries (COM), relative to civil law countries, pay higher dividends. An
important finding to note in Model II is that even after adding all these factors to the
regression model, the coefficient for the privatization dummy (PVT) remains positive and
significant.
Next, we add interaction variables to the model. After adding the interactions between
the privatization dummy and the other determinants of dividend policy, in Model III,
we find that the coefficient on privatization dummy (PVT) is no longer significant. The
loss of significance for the dummy (PVT) coefficient indicates that the positive relation
between dividends and privatization is likely to be driven by one of the other determinants
in the model. A closer examination of the results in Model III indicates a significant
positive relation between dividends and the earnings ratio (ER * PVT) and growth in sales
(G Sales * PVT) of privatized firms. These results show a strong relation between the
privatized firm’s decision to increase dividends and its improvement in performance and
23
efficiency post privatization. On the other hand, our results in Model III further show
that arguments based on the life-cycle or maturity hypothesis do not fit the privatized
firms in our sample. The interaction coefficients for the retained earnings to total equity
(RETE * PVT) is insignificant. The results are, however, consistent with the free cash
flows hypothesis. We find a significant relation between the privatized firm’s dividends
and the interaction coefficients for closely held shares (CLOSE * PVT). Furthermore, we
find a significant negative coefficient on the interaction between the privatization dummy
and the dummy for the common law countries (COM * PVT). It suggests that, in our
sample, privatized firms in the common law countries pay significantly lower dividends.
This finding for the privatized firms is consistent with the ‘substitute model’ suggested
by La Porta, Lopez-de-Silanes, Shleifer, and Vishny (2000). This hypothesis suggests that
dividends are a substitute for legal protection. Therefore, firms in countries with lower level
of protection to shareholders should pay higher dividends. As reported in Model IV, our
findings in respect to Model III are robust to the international variation over time of the
dividend tax penalty (Poterba and Summers, 1984, Jacob and Jacob, 2013).28 Specifically,
the dividend tax penalty (DTP) is associated with a significant, large (-0.328), and negative
influence of dividend pay outs internationally, however, there is no significant difference in
the magnitude of this effect across privatized and non-privatized firms (DTP * PVT).
[Please insert table 6 about here]
Last, for robustness check, we repeat the analysis by separating the sample into sub-
samples based on extent of privatization, the level of regulation in the industry, and the
level of development in the country. The results for these sub-samples are reported in table
7. The findings for the sub-samples based on extent of privatization are reported in Panel A.
While we still find that the closely held shares (CLOSE * PVT) negatively affect the firms’
dividends, the coefficient on the common law dummy (COM * PVT) interaction variable is
negative but not significant. We find that the coefficient for the privatization dummy stays
significant even after adding the interactions with other factors in the model. In Panel
B, however, we show that while closely held shares (CLOSE * PVT) and growth in sales
28The same result holds using other proxies for the dividend tax viz. dividend tax preference (La Porta,Lopez-De-Silanes, Shleifer, and Vishny, 2000) and weighted average dividend tax (Becker, Jacob and Jacob,2013). The results are available from the authors on request.
24
(G Sales * PVT) are significant for privatizations in competitive industries (sectors with
relatively low levels of regulation), only the effects of earnings reporting frequency (ERF
* PVT) are significant for privatizations in the non-competitive industries. We do find a
significant and negative coefficient for the common law dummy (COM * PVT), reported
in Panel B (for privatizations in competitive industry sectors) and in Panel C. In non-
competitive industry sectors, therefore, our analysis suggest little by way of explanation for
the privatized firm payout premium. In Panel C, we also show that while cash (CASH *
PVT) positively influences pay out in emerging markets, the negative influence of closely
held shares arises principally in developed markets. The growth in sales (G Sales * PVT)
retains a positive influence on dividend pay outs in privatized firms in developed markets.
As reported in Model IV of each panel, our findings are also robust, across a sample of 17
countries, to the international variation over time of the dividend tax penalty; DTP * PVT
(Poterba and Summers, 1984, Jacob and Jacob, 2013).29
[Please insert table 7 about here]
4.4 Additional tests
In the results discussed so far, we recognize that there is a significant difference in the sample
sizes of privatized and non-privatized firms. Therefore, we ask ourselves whether the results
are influenced by the difference in the sample sizes of these two groups. As a robustness
check, we create another control sample of non-privatized firms that is comparable to the
privatized firms in sample size. Specifically, for each privatized firm, we find one matching
non-privatized firm. We construct the one-to-one matched sample sequentially at the year
of privatization on the following criteria: country of origin, firm size (+/- 10%), cash
holdings (+/- 5%), and growth in total assets. We repeat the difference-in-differences
analysis reported in tables 6 and 7 using this control sample. Our difference-in-differences
findings in the matched sample are substantively identical to findings in the full sample of
firms. The findings are reported in Appendix 3 and 4 respectively.
29The same result holds using other proxies for the dividend tax; viz. dividend tax preference (La Porta,Lopez-De-Silanes, Shleifer, and Vishny, 2000) and weighted average dividend tax (Becker, Jacob and Jacob,2013). The results are available from the authors on request.
25
5 Conclusions
Since the rapid growth in privatizations of European firms during the 1980s and 1990s, the
effect of privatization on the firm’s financial performance, operating efficiency, and payout
decisions has been of great interest to researchers. Governments usually expect privatization
to increase the profitability and the operational efficiency of the firms. Consistent with those
expectations, prior studies document an improvement in firm performance, an increase in
capital spending, and a decrease in debt and ownership concentration post privatization.
Another significant impact of privatization is on the dividend policies of the firms. While
there are no explicit theoretical explanations as to why privatized firms exhibit higher
dividends pay outs, prior studies suggest that it could be a consequence of changes in
the ownership structure, shareholder preferences, and the resulting agency conflicts. The
pay outs by privatized firms increase markedly around the privatization event and are
significantly higher as compared to the non-privatized firms and hence the topic warrants
further research. Although the literature on privatization has grown rapidly, the question
as to why privatized firms pay such high dividends and what factors enable them to do so
are still unanswered. We attempt to fill this gap in the literature. From the viewpoints of
corporate officials who must set the payout policy, investors in respect to capital allocation
decisions, and economists seeking to understand the functioning of the capital markets,
an important question arises in respect to the determination of the privatized firm payout
decisions: Does the difference between the dividend pay outs of pre- and post-privatized
firms stem principally from differences in information asymmetries, incomplete contracting
possibilities, the phase of the financial life-cycle or taxes between the two groups?
When compared to non-privatized firms, we find that the privatized firms are not only
more profitable but also pay a higher proportion of their profits as dividends. Our find-
ings show a strong positive relation between the firm’s decision to pay dividends and its
profitability, growth in sales, and improvement in firm efficiency. Our findings show no
reduction in the privatized firm’s growth opportunities, sales growth, earnings growth,
market-to-book ratio, or cash reserves. We conclude that the life-cycle theory maturity
hypothesis does not explain the dividend premium paid by privatized firms. Instead, we
propose an agency costs type signalling based argument. Privatization leads to harsher
26
product market competition, higher capital market scrutiny, and a likely change in a firm’s
objective function along with significant change in the ownership structure, which in turn
could lead to an increase in the agency conflicts between various stakeholders. The man-
agement can send a costly signal, in a higher dividend pay out, to the market to mitigate
potential over-investment and other agency costs. We find a strong and consistent positive
relation between the firm’s dividends and its earnings, growth in sales, and operating effi-
ciency. Our results show that, after privatization, firms increase pay outs more in civil than
in common law countries which suggests a signal of reduced prospective agency costs to
protect minority shareholder, which is less necessary in common law countries. In addition,
we find a negative relation between the extent of closely held shares in privatized firms and
dividend pay outs, which is consistent with dividends potentially substituting for the moni-
toring activities of certain major shareholders. These findings are robust to different model
specifications. Therefore, we conclude that the commonly observed increase in dividends
immediately following privatization is mainly driven by improvements in profitability, firm
efficiency, growth opportunities, and a new incentive on firm management to reduce agency
costs.
27
References
[1] Aggarwal, R., Erel, I., Ferreira, M., Matos, P., 2011. Does governance travel aroundthe world? Evidence from institutional investors. Journal of Financial Economics 100,154-181.
[2] Allen, F., Michaely, R. 2003. Payout Policy. In: Constantinides, G., Harris, M., Stulz,R. (Eds.), Handbook of the Economics of Finance. Elsevier Science, Amsterdam.
[3] Ashenfelter, O., Card, D., 1985. Using the Longitudinal Structure of Earnings to Es-timate the Effect of Training Programs. The Review of Economics and Statistics, 67,648-660.
[4] Banyi, M., Kahle, K.M., 2013. Declining propensity to pay? A re-examination of thelife cycle theory. University of Arizona working paper series.
[5] Bhattacharya, S., 1979. Imperfect information, dividend policy, and the bird In thehand fallacy. Bell Journal of Economics 10, 259-70.
[6] Becker, B., Jacob, M., Jacob, M., 2013. Payout taxes and the allocation of investment.Journal of Financial Economics 107, 1-24.
[7] Benartzi, S., Grullon, G., Michaely, R., Thaler, R., 2005. Dividend changes do not signalchanges in future profitability. Journal of Business 78, 1659-1682.
[8] Boubakri, N., Cosset, J-C., 1998. The financial and operating performance of newlyprivatized firms: Evidence from developing countries. Journal of Finance 53, 1081-1110.
[9] Boubakri, N., Cosset, J-C., Guedhami, O., 2005. Liberalization, corporate governanceand the performance of privatized firms in developing countries. Journal of CorporateFinance 11, 767-790.
[10] Brav, A., Graham, J.R., Harvey, C.R., Michaely, R., 2005. Payout policy in the 21stcentury. Journal of Financial Economics 77, 483-527.
[11] Brockman, P., Unlu, E., 2009. Dividend policy, creditor rights and the agency costs ofdebt. Journal of Financial Economics 92, 276-299.
[12] Brockman, P., Unlu, E., 2011. Earned/contributed capital, dividend policy, and dis-closure quality: An international study. Journal of Banking and Finance 35, 1610-1625.
[13] Bushman, R., Piotroski, J., Smith, A., 2004. What determines corporate transparency?Journal of Accounting Research 42, 207-252.
[14] Chay, J.B., Suh. J, 2009. Payout policy and cash-flow uncertainty. Journal of FinancialEconomics 93, 88-107.
[15] DeAngelo, H., DeAngelo, L., Skinner, D.J., 2004. Are dividends disappearing? Div-idend concentration and the consolidation of earnings. Journal of Financial Economics72, 425-456.
28
[16] DeAngelo, H., DeAngelo, A., Stulz, R.M., 2006. Dividend policy and theearned/contributed capital mix: A test of the lifecycle theory. Journal of Financial Eco-nomics 81, 227-254.
[17] Denis, D.J., Osobov, I., 2008. Why do firms pay dividends? International evidence onthe determinants of dividend policy. Journal of Financial Economics 89, 62-82.
[18] D’Souza, J., Megginson, W.L., 1999. The financial and operating performance of pri-vatized firms during the 1990s. Journal of Finance 54, 1397-1438.
[19] D’Souza, J., Megginson, W.L., Nash, R., 2005. Effect of institutional and firm-specificcharacterstics on post-privatization performance: Evedence from developed countries.Journal of Corporate Finance 11, 747-766.
[20] Easterbrook, F.H., 1984. Two agency-cost explanations of dividends. The AmericanEconomic Review 74, 650-659.
[21] von Eije, H.J., Goyal, A., Muckley, C., 2014. Does the information content of payoutinitiations and omissions influence firm risks? Journal of Econometrics, forthcoming.
[22] von Eije, H.J., Megginson, W.L., 2008. Dividends and share repurchases in the Euro-pean Union. Journal of Financial Economics 89, 347-374.
[23] Fama, E.F., French, K.R., 2001. Disappearing dividends: Changing characteristics orlower propensity to pay?. Journal of Financial Economics 60, 3-43.
[24] Grullon, G., Michaely, R., Swaminathan, B., 2002. Are dividend changes a sign of firmmaturity. Journal of Business 75, 387-424.
[25] Harford, J., Mansi, S.A., Maxwell, W.F., 2008. Corporate governance and firm cashholdings in the US, Journal of Financial Economics 87, 535-555.
[26] Healy, P.M., Palepu, K.G., 1988. Earnings information conveyed by dividend initiationsand omissions. Journal of Financial Economics 21, 149-176.
[27] Jacob, M., Jacob, M., 2013. Taxation, dividends, and share repurchases: Taking evi-dence global. Journal of Financial and Quantitative Analysis forthcoming.
[28] Jensen, M.C., 1986. Agency costs of free cash flow, corporate finance and takeovers.American Economic Review 76, 323-329.
[29] Jensen, M.C., Meckling, W.H., 1976, Theory of the firm: managerial behavior, agencycosts and ownership structure. Journal of Financial Economics 3, 305-360.
[30] John, K., Williams, J., 1985. Dividends, dilution and taxes: A signaling equilibrium.Journal of Finance 40, 1053-1070.
[31] Jones, S.L., Megginson, W.L., Nash, R.C., Netter, J.M., 1999. Share issue privatiza-tions as financial means to political and economic ends. Journal of Financial Economics53, 217-253.
[32] La Porta, R., Lopez-de Silanes, F., Shleifer, A., Vishny, R.W., 1997. Legal determinantsof external finance. Journal of Finance 52, 1131-1150.
29
[33] La Porta, R., Lopez-de Silanes, F., Shleifer, A., Vishny, R.W., 2000. Agency problemsand dividend policies around the world. Journal of Finance 55, 1-33.
[34] Lang, L.H.P., Litzenberger, R.H., 1989. Dividend announcement: Cash flow signallingvs. free cash flow hypothesis? Journal of Financial Economics 24, 181-191.
[35] Lintner, J., 1956. Distribution of incomes of corporations among dividends, retainedearnings, and taxes. American Economic Review 46, 97-113.
[36] Megginson, W.L., Nash, R.C., van Randenborgh, M., 1994. The financial and operatingperformance of newly privatized firms: An international empirical analysis. Journal ofFinance 49, 403-452.
[37] Megginson, W.L., Netter, J.M., 2001. From state to market: A survey of empiricalstudies on privatization. Journal of Economic Literature 39, 321-389.
[38] Michaely, R., Thaler, R.H., Womack, K.L., 1995. Price reactions to dividend initiationsand omissions: Overreaction or drift? Journal of Finance 50, 573-608.
[39] Miller, M.H., Modigliani, F., 1961. Dividend policy, growth, and the valuation ofshares. Journal of Business 34, 411-433.
[40] Miller, M.H., Rock, K., 1985. Dividend policy under asymmetric information. Journalof Finance 40, 1031-1051.
[41] Pinkowitz, L., Stulz, R., Williamson, R., 2006. Does the contribution of corporate cashholdings and dividends to firm value depend on governance? A cross-country analysis.Journal of Finance 61, 2725-2751.
[42] Poterba, J.M., Summers, L.H., 1984. New evidence that taxes affect the valuation ofdividends. Journal of Finance 39, 13971415.
[43] Watts, R., 1973. The information content of dividends. Journal of Business 46, 191-211.
[44] Wood, A., 2001. Death of the dividend? CFO Europe research report. The EconomistGroup, London.
30
Fig
ure
1:T
he
grow
thop
por
tun
itie
sof
new
lyp
riva
tize
dfi
rms
are
pro
xie
dby
the
mar
ket
tob
ook
valu
e(M
TBV
),gro
wth
into
tal
asse
ts(G
TA
)an
dsa
les
grow
th(G
Sales)
and
pro
fita
bil
ity
ofth
ese
firm
sis
pro
xie
dby
the
scal
edea
rnin
gs
(ER
).T
he
tim
esc
ale
isfr
om3
yea
rsp
re-
to3
year
sp
ost-
the
year
ofp
riva
tiza
tion
,ye
ar0.
Th
ed
ata
issa
mp
led
from
Sta
teO
wn
edE
nte
rpri
sers
in26
cou
ntr
ies
from
1990
to20
11fo
rfi
rms
pri
vati
zed
bet
wee
n19
92an
d20
09.
31
Table 1This table presents summary statistics for the proxy variables (firm characteristics) used in this studyto compare the firm-specific characteristics 3-years before and 3-years after privatization for 121 newlyprivatized firms across 26 countries from 1990 to 2011 for the firms privatized between 1992 and 2009. Alldata are sourced in Worldscope. N refers to the number of firm year observations available for the respectivevariables. Expected change refers to the anticipated change in proxy variable after privatization. It relatesto both Sign and Proportion. Sign refers to the expected sign of the difference in mean and median proxyvariable values after privatization. Proportion refers to the percentage of firms whose proxy values changeas expected as well as a test of significance of this change (Z-statistics). Before and After refers to themean and median values of the proxy variables for the three-year periods before and after privatization.Difference refers to the difference in mean and median values for 3-years after privatization minus mean andmedian values for 3-years before privatization. Difference in mean between the pre and post privatizationfirm-specific characteristics for privatized firms is calculated by using a two-sample mean-comparison test(T-statistics). Difference in median between the pre and post privatization firm-specific characteristics forprivatized firms is calculated by using a Wilcoxon signed rank test (Z-statistics). We use the country specificconsumer price indices to deflate the nominal firm specific accounting and financial data into real 1990 prices.The proxy variables have been converted from local currency to US$ by using the year-end conversion rate.For a definition of the proxy variables please refer to Appendix 1.
Variables N Expected change Mean MedianSign Proportion Before After Difference Before After Difference
Payout
DIV 117 (+) 0.77a 102.12 205.59 103.47a 11.67 30.23 18.56a
DIV EBIAT 117 (+) 0.64a 0.22 0.27 0.05 0.14 0.24 0.10b
DIV NI 117 (+) 0.60b 0.00 0.48 0.48c 0.22 0.36 0.14b
SizeMV 92 (+) 0.71a 7368.34 10103.89 2735.55c 1466.09 1697.85 231.75a
SIZE 92 (+) 0.61b 75.80 78.80 3.00a 83.40 86.66 3.26a
Profitability
ER 121 (+) 0.66a 4.20 6.06 1.86a 5.79 6.46 0.67a
Liquidity
RETE 111 (+) 0.64a -6.29 20.52 26.81c 11.76 20.88 9.12a
CASH 121 (+) 0.48 25.97 25.64 -0.33 22.46 21.97 -0.49Ownership
CLOSE 94 (-) 0.68a 74.67 52.66 -21.99a 66.85 52.16 -14.69a
RiskNI Risk 121 (+) 0.60b 0.07 0.11 0.04c 0.03 0.04 0.01a
GrowthG TA 105 (+) 0.63b 10.30 14.64 4.34a 7.94 11.25 3.31a
MTBV 92 (+) 0.53 3.26 2.77 0.49 1.67 1.69 0.02G Sales 105 (+) 0.68b 11.51 16.70 5.19a 7.71 10.25 2.54b
Leverage
LR 121 (-) 0.46 22.21 22.76 0.56 19.85 20.24 0.39Reporting
ERF 120 (+) 0.88a 1.90 2.68 0.78a 1.59 2.67 1.08a
Efficiency
Sales Emp 99 (+) 0.76a 0.39 0.55 0.16c 0.19 0.22 0.03a
Emp 99 (-) 0.48 32505.10 31743.24 -761.86c 6374.67 6580.00 205.33b
a, b, c represents significance at the 1%, 5% and 10% levels respectively.
32
Tab
le2
This
table
pre
sents
sum
mary
stati
stic
sfo
rpro
xy
vari
able
sfo
rpri
vati
zed
firm
sunder
diff
eren
tca
tegori
es:
contr
ol
and
reven
ue
pri
vati
zati
ons,
com
pet
itiv
eand
non-c
om
pet
itiv
ese
ctors
,and
emer
gin
gand
dev
elop
edm
ark
ets.
The
sam
ple
isacr
oss
26
countr
ies
from
1990
to2011
for
the
firm
spri
vati
zed
bet
wee
n1992
and
2009.
All
data
are
sourc
edin
Worl
dsc
op
e.P
anel
Apre
sents
per
form
ance
changes
for
Contr
ol
pri
vati
zati
on
-w
her
egov
ernm
ents
reduce
thei
rst
ock
hold
ing
tob
elow
50%
and
Rev
enue
pri
vati
zati
on
-w
her
egov
ernm
ent
reta
ins
majo
rity
owner
ship
,i.e.
,m
ore
than
50%
owner
ship
.W
eco
mpare
the
firm
-sp
ecifi
cch
ara
cter
isti
cs3-y
ears
bef
ore
and
3-y
ears
aft
erpri
vati
zati
on
for
121
new
lypri
vati
zed
firm
s-
38
Contr
ol
pri
vati
zati
ons
(Cont.
)and
83
Rev
enue
pri
vati
zati
ons
(Rev
.).
Panel
Bpre
sents
firm
chara
cter
isti
cch
anges
for
firm
sop
erati
ng
inco
mp
etit
ive
indust
ries
-th
ose
that
are
sub
ject
toin
tern
ati
onal
pro
duct
mark
etco
mp
etit
ion
and
non-c
om
pet
itiv
ein
dust
ries
(pri
nci
pally
financi
al
inst
ituti
ons,
banks,
and
uti
liti
es)
-th
ose
indust
ries
that
are
rela
tivel
yfr
eeof
pro
duct
mark
etco
mp
etit
ion.
We
com
pare
the
firm
spec
ific
chara
cter
isti
cs3-y
ears
bef
ore
and
3-y
ears
aft
erpri
vati
zati
on
for
121
new
lypri
vati
zed
firm
s-
90
com
pet
itiv
ein
dust
rypri
vati
zati
ons
(Com
p.)
and
31
non-c
om
pet
itiv
ein
dust
rypri
vati
zati
ons
(Non-C
om
p.)
.P
anel
Cpre
sents
firm
sop
erati
ng
inem
ergin
gco
untr
ies
ver
sus
firm
sop
erati
ng
indev
elop
edco
untr
ies.
We
com
pare
the
firm
-sp
ecifi
cch
ara
cter
isti
cs3-y
ears
bef
ore
and
3-y
ears
aft
erpri
vati
zati
on
for
121
new
lypri
vati
zed
firm
s-
42
emer
gin
gm
ark
etpri
vati
zati
on
(Em
er.)
and
79
dev
elop
edm
ark
etpri
vati
zati
on
(Dev
.).
Nre
fers
toth
enum
ber
of
firm
yea
robse
rvati
ons
available
.E
xp
ecte
d(P
osi
tive
/N
egati
ve)
refe
rsto
the
anti
cipate
dch
ange
inpro
xy
vari
able
aft
erpri
vati
zati
on,
i.e.
,th
eex
pec
ted
sign
of
the
diff
eren
cein
mea
nand
med
ian
valu
esaft
erpri
vati
zati
on.
Pro
p.
refe
rsto
the
pro
port
ion
of
firm
sw
hose
pro
xy
valu
esch
ange
as
exp
ecte
das
wel
las
ate
stof
signifi
cance
of
this
change
(Z-s
tati
stic
s).
Diff
eren
cere
fers
toth
ediff
eren
cein
mea
nand
med
ian
valu
esfo
r3-y
ears
aft
erpri
vati
zati
on
min
us
mea
nand
med
ian
valu
esfo
r3-y
ears
bef
ore
pri
vati
zati
on,
resp
ecti
vel
y.D
iffer
ence
inm
ean
bet
wee
nth
epre
-and
post
-pri
vati
zati
on
firm
-sp
ecifi
cch
ara
cter
isti
csfo
rpri
vati
zed
firm
sis
calc
ula
ted
by
usi
ng
atw
o-s
am
ple
mea
n-c
om
pari
son
test
(T-s
tati
stic
s).
Diff
eren
cein
med
ian
bet
wee
nth
epre
and
post
pri
vati
zati
on
firm
-sp
ecifi
cch
ara
cter
isti
csfo
rpri
vati
zed
firm
sis
calc
ula
ted
by
usi
ng
aW
ilco
xon
signed
rank
test
(Z-s
tati
stic
s).
We
use
the
countr
ysp
ecifi
cco
nsu
mer
pri
cein
dic
esto
defl
ate
the
nom
inal
firm
spec
ific
acc
ounti
ng
and
financi
al
data
into
real
1990
pri
ces.
All
the
pro
xy
vari
able
shav
eb
een
conver
ted
from
loca
lcu
rren
cyto
US$
by
usi
ng
the
yea
r-en
dco
nver
sion
rate
.F
or
adefi
nit
ion
of
the
pro
xy
vari
able
sple
ase
refe
rto
App
endix
1.
Pan
elA
:C
ontr
ol
an
dR
even
ue
Pri
vati
zati
on
Pan
elB
:C
om
pet
itiv
ean
dN
on
-Com
pet
itiv
eF
irm
sP
an
elC
:E
mer
gin
gan
dD
evel
op
edC
ou
ntr
ies
Vari
ab
les
NP
rop
.D
iffer
ence
Diff
eren
ceN
Pro
p.
Diff
eren
ceD
iffer
ence
NP
rop
.D
iffer
ence
Diff
eren
ce(E
xp
ecte
d)
of
mea
nof
med
ian
of
mea
nof
med
ian
inm
ean
inm
edia
n
Payou
t
DIV
Cont.
38
0.6
5b
52.5
1a
10.2
9b
Com
p.
87
0.7
8a
81.8
3a
18.5
9a
Em
er.
41
0.6
6b
54.0
7a
8.8
0a
(Posi
tive)
Rev
.79
0.8
3a
127.9
9a
19.1
2a
Non
-Com
p.
30
0.7
7a
166.2
5c
43.0
7a
Dev
.76
0.8
4a
130.1
3a
21.6
5a
DIV
EB
IAT
Cont.
38
0.6
6b
0.0
30.0
8C
om
p.
87
0.6
6a
0.0
50.0
9b
Em
er.
41
0.5
90.0
90.1
2(P
osi
tive)
Rev
.79
0.6
3b
0.0
60.1
1c
Non
-Com
p.
30
0.6
00.0
40.1
0D
ev.
76
0.6
7a
0.0
30.1
0a
DIV
NI
Cont.
38
0.6
8b
0.3
40.1
2C
om
p.
87
0.6
1b
0.6
5c
0.1
6b
Em
er.
41
0.5
10.0
30.1
6(P
osi
tive)
Rev
.79
0.5
70.5
50.1
2c
Non
-Com
p.
30
0.6
00.0
10.0
8D
ev.
76
0.6
6a
0.7
3c
0.1
2b
Siz
eM
VC
ont.
27
0.5
6-1
786.4
083.1
1C
om
p.
69
0.7
2a
2481.4
7369.6
3a
Em
er.
32
0.4
7-1
502.5
4-2
76.7
1(P
osi
tive)
Rev
.65
0.7
8a
4613.9
0a
323.7
7a
Non
-Com
p.
23
0.7
0b
3497.7
91445.2
2b
Dev
.60
0.8
5a
4995.8
6a
923.8
4a
SIZ
EC
ont.
27
0.5
92.6
18.5
7C
om
p.
69
0.6
1c
3.1
8a
5.0
2b
Em
er.
32
0.4
40.5
41.7
9(P
osi
tive)
Rev
.65
0.6
3b
3.1
6a
4.1
5a
Non
-Com
p.
23
0.6
52.4
7b
2.9
2c
Dev
.60
0.7
2a
4.3
1a
6.0
9a
Pro
fita
bilit
y
ER
Cont.
38
0.7
0a
4.5
1b
0.8
3a
Com
p.
90
0.6
3b
1.2
1c
0.5
5b
Em
er.
42
0.5
8c
3.7
3b
0.3
7c
(Posi
tive)
Rev
.83
0.6
5b
2.6
4b
0.5
7a
Non
-Com
p.
31
0.7
8a
2.0
8a
0.7
3a
Dev
.78
0.6
8b
4.8
3a
0.5
1b
a,
b,
cre
pre
sents
sign
ifica
nce
at
the
1%
,5%
an
d10%
level
sre
spec
tivel
y.
33
Tab
le2
contd
. Pan
elA
:C
ontr
ol
an
dR
even
ue
Pri
vati
zati
on
Pan
elB
:C
om
pet
itiv
ean
dN
on
-Com
pet
itiv
eF
irm
sP
an
elC
:E
mer
gin
gan
dD
evel
op
edC
ou
ntr
ies
Vari
ab
les
NP
rop
.D
iffer
ence
Diff
eren
ceN
Pro
p.
Diff
eren
ceD
iffer
ence
NP
rop
.D
iffer
ence
Diff
eren
ce(E
xp
ecte
d)
of
mea
nof
med
ian
of
mea
nof
med
ian
inm
ean
inm
edia
n
Liq
uid
ity
RE
TE
Cont.
37
0.6
5c
23.0
513.1
0c
Com
p.
81
0.6
5a
32.4
3c
8.1
9a
Em
er.
37
0.5
7-1
.02
12.6
4(P
osi
tive)
Rev
.74
0.6
5a
28.7
07.2
4a
Non
-Com
p.
30
0.6
311.6
7c
15.8
3c
Dev
.74
0.6
9a
40.7
4b
8.3
5a
CA
SH
Cont.
38
0.5
02.0
45.1
0C
om
p.
90
0.5
00.6
00.9
8E
mer
.42
0.5
20.0
51.3
1(P
osi
tive)
Rev
.83
0.4
8-1
.41
-3.6
0N
on
-Com
p.
31
0.4
5-3
.01
-3.7
5D
ev.
79
0.4
7-0
.53
-0.5
0O
wn
ersh
ip
CL
OS
EC
ont.
28
0.7
2a
-18.4
0a
-20.9
4b
Com
p.
70
0.6
7a
-9.0
3a
-8.9
9a
Em
er.
28
0.6
8b
-9.4
3a
-2.3
2c
(Neg
ati
ve)
Rev
.66
0.6
7b
-5.3
9a
-10.5
6a
Non
-Com
p.
24
0.7
1b
-18.6
0a
-4.8
2b
Dev
.66
0.6
8a
-12.6
9a
-14.9
7a
Ris
kN
IR
isk
Cont.
38
0.6
1b
0.1
0c
0.0
1c
Com
p.
90
0.5
8b
0.0
5c
0.0
1b
Em
er.
42
0.3
80.0
20.0
3b
(Posi
tive)
Rev
.83
0.6
0b
0.0
10.0
1N
on
-Com
p.
31
0.6
8a
0.0
10.0
2D
ev.
79
0.4
1c
0.0
40.0
0b
Gro
wth
GT
AC
ont.
34
0.6
9a
1.9
8b
4.0
1a
Com
p.
79
0.6
8b
3.6
0b
0.1
5E
mer
.35
0.8
3a
14.2
0b
8.7
3a
(Posi
tive)
Rev
.71
0.7
1a
4.8
2a
3.1
1a
Non
-Com
p.
26
0.7
7a
7.1
3a
9.8
5a
Dev
.70
0.5
41.0
42.2
4M
TB
VC
ont.
27
0.5
22.8
50.0
0C
om
p.
69
0.5
40.5
5-0
.05
Em
er.
32
0.4
4-0
.39
-0.1
7(P
osi
tive)
Rev
.65
0.5
4-0
.49
0.0
2N
on
-Com
p.
23
0.5
20.3
2-0
.11
Dev
.60
0.5
80.9
60.0
6c
GS
ale
sC
ont.
34
0.6
1a
7.2
2a
5.2
5a
Com
p.
79
0.6
6b
4.2
6a
3.8
3b
Em
er.
35
0.7
6a
17.1
3a
15.9
4a
(Posi
tive)
Rev
.71
0.6
2b
8.8
2a
0.3
1N
on
-Com
p.
26
0.6
9b
8.8
4a
7.3
9a
Dev
.70
0.5
00.3
12.7
7L
ever
age
LR
Cont.
38
0.3
9-0
.82
-2.6
4C
om
p.
90
0.5
0-1
.18
-1.7
4E
mer
.42
0.2
97.6
1a
8.0
5a
(Neg
ati
ve)
Rev
.83
0.4
81.1
91.7
3N
on
-Com
p.
31
0.3
25.5
9a
6.7
0b
Dev
.79
0.5
4-3
.19
-4.1
8R
eport
ing
ER
FC
ont.
37
0.9
4a
0.9
6a
1.0
0a
Com
p.
90
0.8
7a
0.6
8a
0.8
3a
Em
er.
42
0.8
3a
0.8
8a
1.6
7a
(Posi
tive)
Rev
.83
0.8
5a
0.7
0a
0.6
7a
Non
-Com
p.
30
0.9
3a
1.1
0a
1.6
7a
Dev
.78
0.9
1a
0.7
3a
0.8
4a
Effi
cien
cy
Sale
sE
mp
Cont.
30
0.7
6a
0.1
20.0
4a
Com
p.
75
0.7
5a
0.1
40.0
3a
Em
er.
26
0.7
3a
0.2
70.0
5c
(Posi
tive)
Rev
.69
0.8
6a
0.1
7c
0.0
2a
Non
-Com
p.
24
0.8
3a
0.2
3b
0.1
1a
Dev
.73
0.7
8a
0.1
20.0
4a
Em
pC
ont.
30
0.5
01845.4
417.3
3C
om
p.
75
0.4
7-2
09.4
0-6
04.3
3E
mer
.26
0.5
8-1
795.0
4-8
59.1
7(N
egati
ve)
Rev
.69
0.4
6-1
895.4
7984.0
0N
on
-Com
p.
24
0.5
0-2
488.3
2639.0
0D
ev.
73
0.4
4-3
93.8
8-1
25.0
0
a,
b,
cre
pre
sents
sign
ifica
nce
at
the
1%
,5%
an
d10%
level
sre
spec
tivel
y.
34
Table 3This table presents the annual time-series of the number of usable observations (N), the proportion ofdividend payers (Prop.) as a fraction of firms that disclose their dividend payout, the arithmetic mean(Mean) and the median (Median) values of dividend, DIV, pay out for each year - 1990 to 2011 for theprivatized and non-privatized firms. All data are sourced in Worldscope. All the DIV observations havebeen converted from local currency to US$ by using the year-end conversion rate. We test the significanceof the changes in the proportions, arithmetic means, and medians of cash dividend pay out for each yearbetween the privatized and non-privatized firms. We employ the two-sample mean-comparison tests (withT-statistics) as our test for significance for the difference in proportions and mean payout amounts of payers.We employ the Wilcoxon signed rank test (with its Z-statistics) as our test of significance for the change inmedian values of DIV between the privatized and non-privatized firms.
Year Privatized firms Non-privatized firms Difference inN Prop. Mean Median N Prop. Mean Median Prop. Mean Median
1990 128 0.85 39.22 6.42 1899 0.87 21.43 2.61 -0.02 17.79b 3.81a
1991 133 0.83 38.59 6.55 2047 0.85 20.95 2.44 -0.02 17.64b 4.11a
1992 141 0.80 40.34 6.41 2170 0.83 19.48 2.22 -0.03 20.85a 4.19a
1993 151 0.79 41.24 5.23 2274 0.79 16.55 1.69 0.00 24.68a 3.54a
1994 158 0.77 46.00 5.72 2392 0.80 18.81 2.02 -0.03 27.18a 3.70a
1995 175 0.81 61.45 9.81 2559 0.81 22.59 2.50 -0.01 38.85a 7.31a
1996 202 0.80 65.76 7.76 2966 0.78 23.26 2.16 0.02 42.49a 5.60a
1997 216 0.79 69.09 7.84 3162 0.77 21.94 1.88 0.02 47.15a 5.95a
1998 233 0.79 76.58 7.72 3299 0.75 25.83 1.73 0.03 50.75a 5.99a
1999 254 0.76 89.19 7.84 3474 0.70 27.79 1.31 0.05c 61.39a 6.53a
2000 272 0.71 84.36 5.01 3801 0.67 21.69 1.06 0.04 62.66a 3.95a
2001 279 0.68 95.33 4.43 4012 0.64 25.34 0.82 0.05 69.98a 3.61a
2002 294 0.65 93.17 2.75 4166 0.59 23.23 0.53 0.06 69.94a 2.21a
2003 311 0.61 100.55 2.06 4345 0.57 27.00 0.50 0.04c 73.54a 1.56a
2004 329 0.60 143.85 2.44 4527 0.57 33.12 0.57 0.03 110.73a 1.87a
2005 331 0.62 158.79 3.60 4624 0.58 33.20 0.72 0.04 125.58a 2.88a
2006 328 0.66 210.93 5.89 4665 0.60 44.39 0.87 0.06b 166.53a 5.01a
2007 322 0.67 254.29 8.35 4518 0.61 58.84 1.29 0.05b 195.44a 7.06a
2008 310 0.65 273.34 8.50 4380 0.63 58.66 1.54 0.02b 214.68a 6.96a
2009 303 0.64 250.81 4.25 4238 0.59 55.30 0.91 0.05b 195.51a 3.34a
2010 288 0.68 249.37 6.03 4101 0.58 54.91 0.89 0.09a 194.46a 5.14a
2011 248 0.73 305.47 13.91 3587 0.63 73.34 1.88 0.09a 232.13a 12.02a
1990 - 5406 0.72 143.59 5.82 77206 0.64 35.50 1.31 0.08a 108.09a 4.51a
2011
a, b, c represents significance at the 1%, 5% and 10% levels respectively.
35
Table 4This table presents summary statistics for the set of proxy variables (firm characteristics) in privatized andnon-privatized firms in 26 countries from 1990 to 2011. All data are sourced in Worldscope. N refers to thenumber of firm-year observations available for the respective variable in each category. Mean and medianare the arithmetic average and median value for each proxy variable. We use the country specific consumerprice indices to deflate the nominal firm specific accounting and financial data into real 1990 prices. Allthe proxy variables have been converted from local currency to US$ by using the year-end conversion rate.Difference in mean between the mean of the privatized and non-privatized firm-specific characteristics iscalculated by using a two-sample mean-comparison test (T-statistics). Difference in median between themedian of the privatized and non-privatized firm-specific characteristics is calculated by using a Wilcoxonsigned rank test (Z-statistics). For a definition of the proxy variables please refer to the Appendix 1.
Variables Privatized firms Non-privatized firms Difference inN Mean Median N Mean Median Mean Median
PayoutDIV 5406 143.59 5.82 77206 35.50 1.31 108.09a 4.51a
DIV EBIAT 5401 0.24 0.15 76903 0.26 0.14 -0.01 0.01a
DIV NI 5403 0.49 0.22 77036 0.49 0.18 0.00 0.04a
SizeMV 5208 5127.32 439.36 75534 4171.05 129.13 956.28 310.23a
SIZE 5208 64.73 71.42 75534 48.98 48.68 15.75a 22.74a
ProfitabilityER 5579 4.56 6.00 79571 3.22 5.72 1.35a 0.28a
LiquidityRETE 5291 -8.96 18.12 75444 -11.54 20.56 2.58 -2.44a
CASH 5552 25.95 19.86 78665 26.57 18.75 -0.62b 1.10a
OwnershipCLOSE 4405 62.24 62.03 63899 46.31 48.79 15.92a 13.24a
RiskNI Risk 6020 0.25 0.03 86249 1.11 0.03 -0.86a 0.00GrowthG TA 5280 10.99 7.55 75101 11.70 7.71 -0.70 -0.16MTBV 5189 2.31 1.56 75312 1.94 1.52 0.37 0.04c
G Sales 5246 11.47 9.97 73524 10.63 9.16 0.84 0.81a
LeverageLR 5581 22.78 21.30 79635 21.05 18.64 1.73a 2.66a
ReportingERF 5607 2.68 2.00 80778 2.52 2.00 0.16a 0.00EfficiencySales Emp 4668 0.57 0.20 63799 0.47 0.17 0.10 0.03Emp 4669 19588.00 3061.00 63915 7238.00 1021.00 12349.44a 2040.00a
a, b, c represents significance at the 1%, 5% and 10% levels respectively.
36
Table 5This table presents the difference in coefficients in the 3-year pre and 3-year post firm characteristics on(the natural log of) cash dividends paid, DIV, by 92 firms (due to a data limitation) from 26 countriesthat were privatized from 1990 to 2011. A random effects panel regression model is employed to estimatethe coefficients. The values in ‘Coeff.’ column corresponds to the regression coefficient of each explanatoryvariable and ‘P-Val.’ corresponds to the level of significance of the Z-value calculated using a robust standarderrors at the firm-level. The dummy variable equals to 1 for the 3-year post privatization period and zerofor the 3-year pre privatization period. All independent variables are multiplied with this dummy and thecorresponding coefficient is indicated as ‘Diff.’ along with the respective p-value ‘P-Val.’. Hence, the ‘Diff.’column corresponds to the difference of 3-year pre privatization from 3-year post privatization regressioncoefficients and the adjacent p-value column indicates the level of significance in difference between 3-yearpre and 3-year post privatization. The latter model is extended in panel B to include a dividend tax penaltyvariable (Poterba and Summers, 1984) and this reduces sample size due to the exclusion of firms in certaincountries detailed in Appendix 1. We use the natural logarithm of certain firm-specific proxy variablesdenoted by ‘Ln ’. We control for the firm-level industry fixed effects and year fixed effects in the regressionmodel. Observation is the number of firm-average observations. Firms is the number of firms for whichobservations are available. R2 overall is the overall R-squared statistic. For a definition of the proxy variablesplease refer to Appendix 1.
Panel A: Privatized Firms Panel B: Countries with dataon Dividend Tax Penalty (DTP)
Variables Coeff. P-Val. Diff. P-Val. Coeff. P-Val. Diff. P-Val.
SIZE 0.022 0.022 0.027 0.029 0.018 0.119 0.019 0.206ER -0.016 0.435 0.051 0.028 -0.004 0.853 0.055 0.052RETE 0.006 0.022 -0.004 0.273 0.007 0.038 -0.005 0.243CASH 0.004 0.509 -0.004 0.649 0.010 0.235 0.000 0.982CLOSE -0.006 0.277 -0.007 0.263 -0.006 0.437 -0.007 0.377NI Risk 0.718 0.003 -0.305 0.788 0.762 0.003 -0.218 0.872G TA 0.014 0.014 -0.041 0.000 0.013 0.054 -0.040 0.001MTBV -0.039 0.398 0.006 0.898 -0.040 0.511 0.044 0.544G Sales -0.012 0.018 0.028 0.010 -0.010 0.097 0.022 0.074LR -0.014 0.088 0.011 0.318 -0.011 0.277 0.027 0.043ERF 0.166 0.177 -0.260 0.079 0.212 0.195 -0.189 0.318Ln Sales Emp 0.563 0.001 -0.150 0.455 0.497 0.015 -0.276 0.328Ln EMP 0.405 0.004 -0.209 0.204 0.440 0.006 -0.166 0.414COM -0.384 0.346 -0.168 0.653 0.341 0.515 -0.431 0.439DTP -1.881 0.040 -0.301 0.807Constant -8.402 0.002 2.629 0.399 -7.988 0.008 3.893 0.329
Observation 182 140Firms 92 70R2 overall 0.727 0.800Ind. fixed effects Yes Yes
37
Table 6This table presents results for the random effects panel regressions for the (natural log of the) real amountspaid as cash dividends, DIV, by privatized and non-privatized firms (26 countries, 1990 to 2011) on a wideset of payout determinants. The values in the Coeff. column correspond to the regression coefficients of eachexplanatory variable and P-Val. corresponds to the level of significance of the Z-value calculated using robuststandard errors at the firm-level. In Model I, only the privatization dummy, PVT is used as an explanatoryvariable. In Model II, the full set of determinants are included. In Model III, interaction variables withthe privatization dummy variable, PVT, are also included. In Model IV, the latter model is extended toinclude a dividend tax penalty variable (Poterba and Summers, 1984) and this reduces sample size due tothe exclusion of firms in certain countries detailed in Appendix 1. To reduce the endogeneity problem theindependent variables, except for the time invariant dummies (COM & PVT) and the YEAR variables, arelagged by one year. Independent variables succeeded by ‘* PVT’ refer to the interaction between firm-specificcharacteristics and the PVT dummy. Hence, we adopt a parametric dummy variable difference-in-differencesprocedure. We use the natural logarithm of the firm-specific proxy variables denoted by ‘Ln ’. We controlfor the firm-level industry fixed effects and year fixed effects in the four regression models. Observation isthe number of firm-year observations. Firms is the number of firms for which observations were available.R2 overall is the overall R-squared statistic. For a definition of the proxy variables please refer to Appendix1.
Model I Model II Model III Model IVVariables Coeff. P-Val. Coeff. P-Val. Coeff. P-Val. Coeff. P-Val.
SIZE 0.031 0.000 0.031 0.000 0.032 0.000ER 0.010 0.000 0.010 0.000 0.009 0.000RETE -0.001 0.000 -0.001 0.000 -0.001 0.000CASH 0.002 0.000 0.002 0.000 0.001 0.003CLOSE -0.003 0.000 -0.003 0.000 -0.003 0.000NI Risk 0.051 0.000 0.049 0.000 0.046 0.000G TA 0.000 0.121 0.000 0.144 -0.001 0.009MTBV -0.022 0.000 -0.024 0.000 -0.021 0.000G Sales 0.000 0.105 0.000 0.034 0.000 0.111LR -0.008 0.000 -0.008 0.000 -0.007 0.000ERF -0.024 0.002 -0.027 0.000 -0.014 0.123Ln Sales Emp 0.129 0.000 0.125 0.000 0.094 0.000Ln EMP 0.256 0.000 0.253 0.000 0.250 0.000COM 0.164 0.000 0.176 0.000 0.117 0.001YEAR 0.034 0.000 0.033 0.000 0.036 0.000DTP -0.329 0.000PVT 0.696 0.000 0.213 0.001 -0.659 0.516 0.058 0.949SIZE * PVT 0.004 0.238 0.001 0.827ER * PVT 0.008 0.060 0.006 0.142RETE * PVT 0.000 0.529 0.000 0.690CASH * PVT 0.002 0.280 0.001 0.601CLOSE * PVT -0.003 0.065 -0.004 0.038NI Risk * PVT 0.051 0.401 0.034 0.547G TA * PVT -0.001 0.412 -0.001 0.158MTBV * PVT 0.022 0.267 0.028 0.080G Sales * PVT 0.002 0.053 0.002 0.028LR * PVT -0.003 0.352 -0.004 0.347ERF * PVT 0.034 0.348 0.021 0.624Ln Sales Emp * PVT 0.042 0.545 -0.010 0.871Ln EMP * PVT 0.020 0.707 0.046 0.395COM * PVT -0.177 0.018 -0.299 0.052DTP * PVT -0.249 0.376Constant 0.954 0.000 -70.853 0.000 -70.324 0.000 -74.056 0.000
Observation 82607 46503 46503 38589Firms 6602 5916 5916 4606R2 overall 0.060 0.642 0.643 0.672Year fixed effects Yes Yes Yes YesInd. fixed effects Yes Yes Yes Yes
38
Tab
le7
Th
ista
ble
pre
sents
resu
lts
for
the
ran
dom
effec
tsp
an
elre
gre
ssio
ns
for
the
(natu
ral
log
of
the)
real
am
ou
nts
paid
as
cash
div
iden
ds,
DIV
,by
pri
vati
zed
an
dn
on
-pri
vati
zed
firm
s(2
6co
untr
ies,
1990
to2011)
on
aw
ide
set
of
payou
td
eter
min
ants
.T
he
regre
ssio
ns
are
per
form
edacc
ord
ing
todiff
eren
tca
tegori
zati
on
sof
pri
vati
zati
on
:co
ntr
ol
an
dre
ven
ue
pri
vati
zati
on
s,co
mp
etit
ive
an
dn
on
-com
pet
itiv
ese
ctors
,an
dem
ergin
gan
dd
evel
op
edm
ark
ets.
Th
ese
cate
gori
zati
on
sare
as
det
ailed
inta
ble
2.
Th
evalu
esin
the
Coeff
.co
lum
nco
rres
pon
dto
the
regre
ssio
nco
effici
ents
of
each
exp
lan
ato
ryvari
ab
lean
dP
-Val.
corr
esp
on
ds
toth
ele
vel
of
sign
ifica
nce
of
the
Z-v
alu
eca
lcu
late
du
sin
gro
bu
stst
an
dard
erro
rsat
the
firm
-lev
el.
Th
eM
od
els
can
be
des
crib
edas
per
tab
le6.
To
red
uce
the
end
ogen
eity
pro
ble
mth
ein
dep
end
ent
vari
ab
les,
exce
pt
for
the
tim
ein
vari
ant
du
mm
ies
(CO
M&
PV
T)
an
dth
eY
EA
Rvari
ab
les,
are
lagged
by
on
eyea
r.In
dep
end
ent
vari
ab
les
succ
eed
edby
‘*P
VT
’re
fer
toth
ein
tera
ctio
nb
etw
een
firm
-sp
ecifi
cch
ara
cter
isti
csan
dth
eP
VT
du
mm
y.W
eu
seth
en
atu
ral
logari
thm
of
the
firm
-sp
ecifi
cp
roxy
vari
ab
les
den
ote
dby
‘Ln
’.W
eco
ntr
ol
for
the
firm
-lev
elin
du
stry
fixed
effec
tsan
dyea
rfi
xed
effec
tsin
the
fou
rre
gre
ssio
nm
od
els.
Ob
serv
ati
on
isth
enu
mb
erof
firm
-yea
rob
serv
ati
on
s.F
irm
sis
the
nu
mb
erof
firm
sfo
rw
hic
hob
serv
ati
on
sw
ere
availab
le.
R2
over
all
isth
eover
all
R-s
qu
are
dst
ati
stic
.F
or
ad
efin
itio
nof
the
pro
xy
vari
ab
les
ple
ase
refe
rto
Ap
pen
dix
1.
Pan
elA
Pan
elA
1:
Contr
ol
pri
vati
zati
on
Pan
elA
2:
Rev
enu
ep
rivati
zati
on
Model
IM
od
elII
Mod
elII
IM
od
elIV
Mod
elI
Mod
elII
Mod
elII
IM
od
elIV
Vari
ab
les
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
PV
T1.0
64
0.0
00
0.6
01
0.0
19
0.5
60
0.0
37
0.7
31
0.0
02
1.7
82
0.0
00
0.6
15
0.0
00
0.4
46
0.0
13
0.6
23
0.0
01
SIZ
E*
PV
T0.0
00
0.9
30
-0.0
04
0.2
50
0.0
05
0.0
87
0.0
00
0.8
91
ER
*P
VT
0.0
08
0.0
77
0.0
05
0.2
16
0.0
06
0.1
18
0.0
04
0.3
33
RE
TE
*P
VT
0.0
00
0.8
81
0.0
00
0.6
32
0.0
00
0.4
95
0.0
00
0.5
47
CA
SH
*P
VT
0.0
03
0.1
84
0.0
02
0.3
82
0.0
01
0.6
03
0.0
00
0.9
40
CL
OS
E*
PV
T-0
.002
0.2
13
-0.0
03
0.0
53
-0.0
03
0.0
79
-0.0
04
0.0
91
NI
Ris
k*
PV
T0.0
30
0.5
96
0.0
21
0.6
84
0.0
39
0.4
92
0.0
34
0.5
30
GT
A*
PV
T-0
.002
0.0
77
-0.0
02
0.0
67
-0.0
01
0.4
54
-0.0
01
0.2
75
MT
BV
*P
VT
0.0
36
0.0
85
0.0
27
0.1
17
0.0
15
0.4
38
0.0
37
0.0
21
GS
ale
s*
PV
T0.0
01
0.1
19
0.0
01
0.1
19
0.0
02
0.0
26
0.0
02
0.0
66
LR
*P
VT
0.0
03
0.4
23
0.0
03
0.4
25
-0.0
02
0.5
68
-0.0
04
0.3
11
ER
F*
PV
T-0
.013
0.7
38
-0.0
17
0.6
67
0.0
32
0.3
81
0.0
13
0.7
63
Ln
Sale
sE
mp
*P
VT
-0.0
07
0.7
70
-0.0
02
0.9
32
0.0
03
0.9
16
-0.0
13
0.6
17
Ln
EM
P*
PV
T0.0
02
0.9
62
0.0
34
0.4
32
-0.0
27
0.4
74
0.0
44
0.3
01
CO
M*
PV
T-0
.042
0.7
47
-0.1
54
0.3
19
-0.1
20
0.3
77
-0.1
58
0.2
99
DT
P*
PV
T-0
.527
0.1
20
-0.3
42
0.2
27
Ob
serv
ati
on
81299
45659
45659
37939
82258
46308
46308
38443
Fir
ms
6519
5838
5838
4550
6575
5893
5893
4592
R2
over
all
0.0
45
0.6
36
0.6
36
0.6
66
0.0
64
0.6
42
0.6
44
0.6
71
39
Tab
le7
contd
.
Panel
B
Panel
B1:
Com
peti
tive
pri
vati
zati
on
Panel
B2:
Non-c
om
peti
tive
pri
vati
zati
on
Model
IM
odel
IIM
odel
III
Model
IVM
odel
IM
odel
IIM
odel
III
Model
IVV
ari
able
sC
oeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
PV
T0.6
55
0.0
00
0.1
61
0.0
19
0.3
82
0.7
04
0.4
63
0.6
06
0.9
17
0.0
00
0.4
96
0.0
04
4.4
20
0.0
16
5.8
59
0.0
03
SIZ
E*
PV
T-0
.001
0.8
63
-0.0
02
0.6
43
0.0
21
0.0
20
0.0
20
0.1
58
ER
*P
VT
0.0
07
0.1
01
0.0
05
0.2
05
0.0
13
0.4
80
0.0
45
0.1
17
RE
TE
*P
VT
0.0
01
0.2
64
0.0
00
0.4
54
0.0
05
0.0
01
-0.0
04
0.1
31
CA
SH
*P
VT
0.0
03
0.1
96
0.0
01
0.6
95
-0.0
01
0.7
56
0.0
00
0.9
88
CL
OSE
*P
VT
-0.0
05
0.0
24
-0.0
05
0.0
19
0.0
02
0.5
88
0.0
03
0.5
31
NI
Ris
k*
PV
T0.0
42
0.4
93
0.0
20
0.7
18
0.4
70
0.6
00
2.3
37
0.1
03
GT
A*
PV
T-0
.001
0.5
87
-0.0
01
0.3
05
-0.0
01
0.7
37
-0.0
02
0.4
21
MT
BV
*P
VT
0.0
23
0.2
64
0.0
32
0.0
48
0.0
92
0.1
05
0.0
38
0.4
03
GSale
s*
PV
T0.0
02
0.0
32
0.0
02
0.0
13
0.0
00
0.8
62
0.0
00
0.9
87
LR
*P
VT
-0.0
06
0.1
10
-0.0
06
0.1
30
0.0
06
0.2
85
0.0
12
0.0
13
ER
F*
PV
T0.0
12
0.7
70
0.0
05
0.9
10
0.0
90
0.0
55
0.0
69
0.0
60
Ln
Sale
sE
mp
*P
VT
0.0
19
0.7
69
-0.0
27
0.6
23
0.2
23
0.1
03
0.2
32
0.1
03
Ln
EM
P*
PV
T0.0
73
0.2
11
0.0
60
0.2
90
-0.0
23
0.8
08
0.0
84
0.4
52
CO
M*
PV
T-0
.248
0.0
90
-0.3
92
0.0
15
0.4
16
0.1
05
0.2
75
0.3
73
DT
P*
PV
T-0
.273
0.3
59
-0.0
69
0.8
88
Obse
rvati
on
81620
45975
45975
38269
78190
43913
43913
36328
Fir
ms
6529
5850
5850
4569
6266
5612
5612
4356
R2
overa
ll0.0
46
0.6
37
0.6
38
0.6
67
0.0
52
0.6
42
0.6
44
0.6
71
Panel
C
Panel
C1:
Em
erg
ing
countr
ies
Panel
C2:
Develo
ped
countr
ies
Model
IM
odel
IIM
odel
III
Model
IVM
odel
IM
odel
IIM
odel
III
Model
IVV
ari
able
sC
oeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
PV
T0.6
51
0.0
00
0.0
67
0.0
02
-5.9
31
0.6
55
1.0
47
0.8
61
0.7
25
0.0
00
0.2
77
0.0
00
0.5
87
0.5
14
0.3
51
0.6
94
SIZ
E*
PV
T-0
.004
0.5
61
0.0
01
0.9
69
0.0
03
0.4
65
0.0
01
0.8
09
ER
*P
VT
0.0
20
0.1
18
-0.0
04
0.8
44
0.0
05
0.2
01
0.0
06
0.1
29
RE
TE
*P
VT
0.0
06
0.0
16
0.0
00
0.8
98
0.0
00
0.6
43
0.0
00
0.6
89
CA
SH
*P
VT
0.0
09
0.0
18
0.0
23
0.0
21
0.0
00
0.9
38
0.0
01
0.7
68
CL
OSE
*P
VT
0.0
04
0.2
31
0.0
08
0.2
42
-0.0
05
0.0
22
-0.0
05
0.0
28
NI
Ris
k*
PV
T-0
.632
0.5
18
-9.0
68
0.0
10
0.0
34
0.5
59
0.0
32
0.5
72
GT
A*
PV
T0.0
03
0.2
61
-0.0
08
0.3
14
-0.0
01
0.1
47
-0.0
01
0.2
39
MT
BV
*P
VT
0.0
31
0.5
58
-0.0
59
0.4
35
0.0
33
0.0
46
0.0
32
0.0
51
GSale
s*
PV
T-0
.002
0.3
66
0.0
08
0.4
07
0.0
02
0.0
13
0.0
02
0.0
24
LR
*P
VT
-0.0
11
0.1
41
-0.0
19
0.0
03
-0.0
02
0.5
73
-0.0
03
0.4
45
ER
F*
PV
T-0
.053
0.3
30
0.1
67
0.1
37
0.0
51
0.2
42
0.0
28
0.5
29
Ln
Sale
sE
mp
*P
VT
0.4
29
0.0
01
-0.0
87
0.8
19
-0.0
46
0.4
26
-0.0
22
0.6
97
Ln
EM
P*
PV
T0.0
76
0.4
45
-0.1
88
0.5
80
0.0
36
0.5
29
0.0
43
0.4
44
CO
M*
PV
T-0
.826
0.0
01
N/A
-0.3
79
0.0
09
-0.4
32
0.0
07
DT
P*
PV
T-0
.769
0.4
75
-0.2
99
0.2
95
Obse
rvati
on
78742
44040
44040
36126
81068
45848
45848
38471
Fir
ms
6317
5650
5650
4340
6478
5812
5812
4585
R2
overa
ll0.0
44
0.6
33
0.6
35
0.6
64
0.0
54
0.6
45
0.6
46
0.6
73
40
Apppendix 1This table presents a description of the firm characteristics and non-firm specific contextual factorsused in the study.
Variables Definition
Privatized A dummy variable, which indicates whether a company is privatized;(PVT) PVT=1 otherwise zero. Privatization is defined as a government or
government controlled entity which sells shares or assets to anon-government entity (Worldscope). Privatization includes both indirectand direct sales of up to a 100% stake to an identifiable buyer andfloatation of stock on a stock exchange.
Non-privatized Firms that have not been and are not controlled by the state.
PayoutCash Dividends The total real (1990 prices) amount of common cash dividend distributed(DIV) by the firm, in millions of US$. DIV EBIAT and DIV NI is cash dividend
(DIV) scaled by earnings before interest but after tax (EBIAT) and netincome (NI), respectively.
SizeMarket Value The total real (1990 prices) amount of market value (capitalization)(MV) of the firm, in millions of US$.
Size of Firm The country-specific market value percentile ranking of a firm on an(SIZE) annual basis.
ProfitabilityEarnings Ratio The firm earnings before interest but after tax (EBIAT) as a percentage(ER) of total assets.
EBIAT The total real (1990 prices) earnings before interest but after tax inmillions of US$.
Net Income (NI) The total real (1990 prices) net income of the firm in millions of US$.
LiquidityRetained Earnings The retained earnings as a percentage of the market value of firm equity.(RETE)
Cash Holding The sum of cash and short term investments as a percentage of total(CASH) assets of the firm.
OwnershipClose The number of shares held by insiders (shareholders who hold 5% or more(CLOSE) of the outstanding shares, such as officers, directors or their immediate
family members, other corporations or individuals) as a percentage of thetotal number of outstanding common shares.
41
Appendix 1 contd.
Variables Definition
RiskIncome Risk The standard deviation of net income as a fraction of total assets over(NI Risk) the most recent three years including the current fiscal year.
GrowthTotal Assets The relative (percentage) change of the total assets in real (1990 prices)Growth (G TA) millions of US$. G TAt = ln(TAt/TAt−1), where ln is natural
logarithm.
MTBV The market value of equity divided by the book value of the equity.
Sales Growth The relative (percentage) change of the total sales in real (1990 prices)(G Sales) millions of US$. G Salest = ln(Salest/Salest−1), where ln is
natural logarithm.
LeverageLeverage Ratio The sum of short-term and long-term debt as a percentage of the total(LR) assets of the firm.
ReportingEarning Reporting The frequency at which earnings are reported per annum. (1 to 4 times).Frequency (ERF) 1 = Annual, 2 = Biannual and 4 = Quarterly Reporting.
EfficiencySales to employees The total real (1990 prices) sales of the firm in millions of US$ as a fractionratio (Sales Emp) of the total number of employees working in a firm.
Employees The total number of both full-time and part-time employees working in(Emp) a firm.
InterceptConstant The intercept of the regression equation.
Non-firm specific contextual factorsCommon Law A dummy variable, which indicates whether a company originates from a(COM) common law country; COM = 1, otherwise zero.
Dividend Tax Dividend tax penalty is attributable to Poterba and Summers (1984) and
Penalty (DTP) defined as δDiv =τDiv−α
1−α −τCG
1−τCG , where τDiv is the dividend tax rate,
τCG is the capital gains tax rate and α is the imputation rate (α variesfrom 0% to 33%). DTP is calculated in all countries except Argentina,Brazil, Chile, China, India, Malaysia, Peru, Russia and Turkey due todata availability limitations.
YEAR Year of observation of the firm-level characteristics in the regressionanalysis, from 1990 to 2011.
42
Appendix 2This table presents a description of the sample of privatized firms and their average cash dividend pay out(in millions of 1990 real US$), 1990 to 2011. Firms refers to the number of privatized firms. Dividendrefers to the average value of DIV. All the DIV observations have been converted from local currency toUS$ by using the year-end conversion rate. Panel A gives the country by country DIV. Panel B givesthe DIV based on control versus revenue privatizations. Panel C gives the DIV based on the nature of theindustry, a competitive versus uncompetitive industry. Panel D gives the DIV based on the level of economicdevelopment of the country where the firm is incorporated. Panel E reports the industry breakdown of theDIV. All panels except panel B contain observations on 409 firms. Panel B is limited to 214 firms due to apaucity of data in the change in shareholdings pre- and post-privatization.
Countries Firms Dividend Category Firms Dividend
Panel A: Country by country Panel B: By level of privatization
Argentina 6 190.08 Control 71 107.38Australia 6 27.09 Revenue 143 298.12Austria 7 12.99Brazil 28 238.08Canada 57 25.78Chile 5 63.81 Panel C: By nature of industry
China 5 39.48Germany 28 291.80 Competitive 336 128.02Spain 14 349.09 Non - competitive 73 213.32Finland 9 111.39France 41 222.16UK 59 45.69Greece 6 154.69 Panel D: By level of development
India 3 50.38Italy 19 489.63 Emerging 124 113.88Mexico 3 256.26 Developed 285 155.42Malaysia 4 56.37Netherlands 6 209.87Norway 4 28.00 Panel E: Sector by sector
New Zealand 6 14.69Peru 10 30.50 Telecommunications 32 296.94Poland 20 44.83 Manufacturing 147 108.64Portugal 7 24.04 Financial 6 51.83Russia 31 92.74 Transportation 32 86.32Sweden 16 28.19 Utilities 67 235.65Turkey 9 50.49 Other 125 95.28
43
Appendix 3This table presents results for the random effects panel regressions for the (natural log of the) real amountspaid as cash dividends, DIV, by one-to-one matched sample of privatized and non-privatized firms (26countries, 1990 to 2011) on a wide set of payout determinants. The matched sample of firms is a monotonicone-to-one relation for the same firm-year of observation on the following criteria: country of origin, firm size(+/- 10%), cash holdings (+/- 5%), and growth in total assets. The values in the Coeff. column correspondto the regression coefficients of each explanatory variable and P-Val. corresponds to the level of significanceof the Z-value calculated using robust standard errors at the firm-level. In Model I, only the privatizationdummy, PVT is used as an explanatory variable. In Model II, the full set of determinants are included. InModel III, interaction variables with the privatization dummy variable, PVT, are also included. In ModelIV, the latter model is extended to include a dividend tax penalty variable (Poterba and Summers, 1984)and this reduces sample size due to the exclusion of firms in certain countries detailed in Appendix 1. Toreduce the endogeneity problem the independent variables, except for the time invariant dummies (COM &PVT) and the YEAR variables, are lagged by one year. Independent variables succeeded by ‘* PVT’ referto the interaction between firm-specific characteristics and the PVT dummy. Hence, we adopt a parametricdummy variable difference-in-differences procedure. We use the natural logarithm of the firm-specific proxyvariables denoted by ‘Ln ’. We control for the firm-level industry fixed effects and year fixed effects in thefour regression models. Observation is the number of firm-year observations. Firms is the number of firmsfor which observations were available. R2 overall is the overall R-squared statistic. For a definition of theproxy variables please refer to Appendix 1.
Model I Model II Model III Model IVVariables Coeff. P-Val. Coeff. P-Val. Coeff. P-Val. Coeff. P-Val.
SIZE 0.036 0.000 0.038 0.000 0.038 0.000ER 0.016 0.000 0.014 0.001 0.011 0.008RETE 0.000 0.501 0.001 0.153 0.001 0.389CASH 0.004 0.010 0.005 0.023 0.003 0.195CLOSE -0.006 0.000 -0.005 0.060 -0.005 0.052NI Risk 0.106 0.139 0.057 0.561 0.055 0.586G TA -0.001 0.161 -0.001 0.270 -0.001 0.207MTBV -0.002 0.895 -0.006 0.726 -0.002 0.899G Sales 0.000 0.515 0.000 0.585 0.000 0.875LR -0.008 0.003 -0.004 0.245 -0.005 0.168ERF -0.007 0.772 -0.029 0.370 -0.051 0.176Ln Sales Emp 0.132 0.032 0.106 0.180 0.067 0.385Ln EMP 0.265 0.000 0.212 0.001 0.226 0.000COM 0.107 0.279 0.125 0.351 -0.142 0.376YEAR 0.035 0.000 0.035 0.000 0.038 0.000DTP -0.536 0.060PVT 0.472 0.006 0.267 0.059 1.091 0.491 1.050 0.484SIZE * PVT -0.004 0.392 -0.005 0.297ER * PVT 0.005 0.032 0.004 0.064RETE * PVT -0.001 0.321 -0.001 0.373CASH * PVT -0.002 0.527 -0.001 0.791CLOSE * PVT -0.002 0.462 -0.002 0.578NI Risk * PVT 0.132 0.237 0.108 0.342G TA * PVT 0.000 0.958 -0.001 0.704MTBV * PVT 0.010 0.702 0.015 0.543G Sales * PVT 0.002 0.144 0.002 0.144LR * PVT -0.008 0.112 -0.006 0.240ERF * PVT 0.050 0.042 0.064 0.025Ln Sales Emp * PVT 0.050 0.064 0.022 0.088Ln EMP * PVT 0.119 0.158 0.136 0.101COM * PVT -0.010 0.057 -0.146 0.070DTP * PVT 0.094 0.814Constant 1.776 0.000 -74.571 0.000 -73.620 0.000 -77.666 0.000
Observation 9324 5543 5543 4810Firms 666 610 610 495R2 overall 0.084 0.662 0.662 0.709Year fixed effects Yes Yes Yes YesInd. fixed effects Yes Yes Yes Yes
44
Ap
pen
dix
4T
his
tab
lep
rese
nts
resu
lts
for
the
ran
dom
effec
tsp
an
elre
gre
ssio
ns
for
the
(natu
ral
log
of
the)
real
am
ou
nts
paid
as
cash
div
iden
ds,
DIV
,by
on
eto
on
em
atc
hed
sam
ple
of
pri
vati
zed
an
dn
on
-pri
vati
zed
firm
s(2
6co
untr
ies,
1990
to2011)
on
aw
ide
set
of
payou
td
eter
min
ants
.T
he
matc
hed
sam
ple
of
firm
sis
am
on
oto
nic
on
e-to
-on
ere
lati
on
for
the
sam
efi
rm-y
ear
of
ob
serv
ati
on
on
the
foll
ow
ing
crit
eria
:co
untr
yof
ori
gin
,fi
rmsi
ze(+
/-
10%
),ca
shh
old
ings
(+/-
5%
),an
dgro
wth
into
tal
ass
ets.
Th
ere
gre
ssio
ns
are
per
form
edacc
ord
ing
tod
iffer
ent
cate
gori
zati
on
sof
pri
vati
zati
on
:co
ntr
ol
an
dre
ven
ue
pri
vati
zati
on
s,co
mp
etit
ive
an
dn
on
-com
pet
itiv
ese
ctors
an
dem
ergin
gand
dev
elop
edm
ark
ets.
Th
ese
cate
gori
zati
on
sare
as
det
ailed
inta
ble
2.
Th
evalu
esin
the
Coeff
.co
lum
nco
rres
pon
dto
the
regre
ssio
nco
effici
ents
of
each
exp
lan
ato
ryvari
ab
lean
dP
-Val.
corr
esp
on
ds
toth
ele
vel
of
sign
ifica
nce
of
the
Z-v
alu
eca
lcu
late
du
sin
gro
bu
stst
an
dard
erro
rsat
the
firm
-lev
el.
Th
eM
od
els
can
be
des
crib
edas
per
tab
le6.
To
red
uce
the
endogen
eity
pro
ble
mth
ein
dep
end
ent
vari
ab
les,
exce
pt
for
the
tim
ein
vari
ant
du
mm
ies
(CO
M&
PV
T)
an
dth
eY
EA
Rvari
ab
les,
are
lagged
by
on
eyea
r.In
dep
end
ent
vari
ab
les
succ
eed
edby
‘*P
VT
’re
fer
toth
ein
tera
ctio
nb
etw
een
firm
-sp
ecifi
cch
ara
cter
isti
csan
dth
eP
VT
du
mm
y.W
eu
seth
en
atu
ral
logari
thm
of
the
firm
-sp
ecifi
cp
roxy
vari
ab
les
den
ote
dby
‘Ln
’.W
eco
ntr
ol
for
the
firm
-lev
elin
du
stry
fixed
effec
tsan
dyea
rfi
xed
effec
tsin
the
fou
rre
gre
ssio
nm
od
els.
Ob
serv
ati
on
isth
enu
mb
erof
firm
-yea
rob
serv
ati
on
s.F
irm
sis
the
nu
mb
erof
firm
sfo
rw
hic
hob
serv
ati
on
sw
ere
availab
le.
R2
over
all
isth
eover
all
R-s
qu
are
dst
ati
stic
.F
or
ad
efin
itio
nof
the
pro
xy
vari
ab
les
ple
ase
refe
rto
Ap
pen
dix
1.
Pan
elA
Pan
elA
1:
Contr
ol
pri
vati
zati
on
Pan
elA
2:
Rev
enu
ep
rivati
zati
on
Model
IM
od
elII
Mod
elII
IM
od
elIV
Mod
elI
Mod
elII
Mod
elII
IM
od
elIV
Vari
ab
les
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
PV
T0.6
04
0.0
13
0.3
81
0.0
21
0.5
14
0.0
12
0.7
99
0.0
01
1.4
25
0.0
00
0.4
64
0.0
10
0.5
23
0.0
10
0.6
03
0.0
03
SIZ
E*
PV
T-0
.008
0.4
01
-0.0
08
0.5
10
-0.0
03
0.4
98
-0.0
05
0.2
36
ER
*P
VT
0.0
05
0.0
62
0.0
04
0.0
51
0.0
04
0.5
29
0.0
03
0.6
58
RE
TE
*P
VT
-0.0
01
0.0
86
-0.0
01
0.1
24
-0.0
01
0.3
39
-0.0
01
0.4
83
CA
SH
*P
VT
-0.0
01
0.6
79
0.0
00
0.9
97
-0.0
04
0.2
34
-0.0
03
0.3
87
CL
OS
E*
PV
T-0
.001
0.6
98
-0.0
02
0.5
70
-0.0
02
0.4
58
-0.0
02
0.5
64
NI
Ris
k*
PV
T0.0
95
0.3
75
0.0
59
0.5
92
0.1
09
0.3
36
0.0
86
0.4
69
GT
A*
PV
T-0
.001
0.3
56
-0.0
01
0.3
26
0.0
00
0.8
61
0.0
00
0.8
17
MT
BV
*P
VT
0.0
27
0.2
94
0.0
15
0.5
56
0.0
02
0.9
33
0.0
21
0.3
77
GS
ale
s*
PV
T0.0
02
0.2
34
0.0
02
0.2
25
0.0
02
0.0
92
0.0
02
0.0
83
LR
*P
VT
-0.0
03
0.6
10
0.0
00
0.9
75
-0.0
07
0.1
75
-0.0
06
0.2
40
ER
F*
PV
T-0
.006
0.9
16
0.0
20
0.7
21
0.0
53
0.3
16
0.0
53
0.3
74
Ln
Sale
sE
mp
*P
VT
0.0
20
0.0
52
0.0
26
0.0
41
0.0
13
0.0
89
0.0
45
0.0
85
Ln
EM
P*
PV
T-0
.085
0.0
35
-0.0
75
0.0
79
-0.0
54
0.0
22
-0.0
95
0.0
92
CO
M*
PV
T0.1
04
0.5
88
0.1
48
0.5
14
0.0
45
0.8
20
0.2
10
0.3
57
DT
P*
PV
T-0
.215
0.6
27
0.0
08
0.9
85
Ob
serv
ati
on
8181
4798
4798
4204
9063
5383
5383
4682
Fir
ms
595
543
543
445
646
592
592
483
Wald
668
1990
2132
2454
825
2526
2773
3101
Pro
bab
ilit
y0.0
00
0.0
00
0.0
00
0.0
00
0.0
00
0.0
00
0.0
00
0.0
00
R2
over
all
0.0
92
0.6
52
0.6
52
0.7
00
0.1
33
0.6
67
0.6
68
0.7
13
45
Ap
pen
dix
4co
ntd
.
Panel
B
Panel
B1:
Com
peti
tive
pri
vati
zati
on
Panel
B2:
Non-c
om
peti
tive
pri
vati
zati
on
Model
IM
odel
IIM
odel
III
Model
IVM
odel
IM
odel
IIM
odel
III
Model
IVV
ari
able
sC
oeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
PV
T0.0
50
0.0
73
0.0
44
0.0
96
-1.0
30
0.5
27
-0.7
85
0.6
08
0.2
47
0.5
69
0.4
24
0.1
48
3.9
65
0.0
43
5.9
87
0.0
08
SIZ
E*
PV
T-0
.009
0.0
66
-0.0
08
0.1
10
0.0
18
0.0
77
0.0
19
0.0
18
ER
*P
VT
0.0
04
0.5
56
0.0
03
0.5
84
0.0
01
0.9
51
0.0
38
0.2
37
RE
TE
*P
VT
-0.0
01
0.5
43
-0.0
01
0.5
40
0.0
06
0.0
01
0.0
06
0.0
87
CA
SH
*P
VT
-0.0
01
0.6
98
-0.0
01
0.7
35
-0.0
07
0.1
72
-0.0
01
0.9
02
CL
OSE
*P
VT
-0.0
03
0.2
79
-0.0
03
0.3
80
0.0
03
0.5
02
0.0
07
0.1
89
NI
Ris
k*
PV
T0.1
17
0.3
01
0.0
85
0.4
58
0.0
65
0.9
45
1.4
71
0.2
88
GT
A*
PV
T0.0
00
0.7
97
0.0
00
0.9
55
0.0
00
0.9
77
-0.0
02
0.4
07
MT
BV
*P
VT
0.0
10
0.7
11
0.0
17
0.5
10
0.1
16
0.0
96
0.0
52
0.4
40
GSale
s*
PV
T0.0
02
0.0
99
0.0
02
0.0
94
0.0
01
0.8
05
0.0
01
0.6
47
LR
*P
VT
-0.0
12
0.0
27
-0.0
09
0.1
00
0.0
04
0.6
08
0.0
09
0.1
59
ER
F*
PV
T0.0
22
0.7
02
0.0
48
0.4
48
0.1
27
0.0
57
0.1
18
0.0
33
Ln
Sale
sE
mp
*P
VT
0.0
47
0.6
59
0.0
25
0.8
01
0.1
63
0.2
47
0.1
66
0.2
89
Ln
EM
P*
PV
T0.1
73
0.0
45
0.1
44
0.0
93
0.0
51
0.6
79
0.1
81
0.1
48
CO
M*
PV
T-0
.003
0.9
88
0.0
39
0.8
63
-0.5
22
0.1
44
-0.6
94
0.0
51
DT
P*
PV
T0.0
05
0.9
90
0.3
88
0.4
33
Obse
rvati
on
8579
5140
5140
4543
5529
3227
3227
2731
Fir
ms
610
559
559
464
378
348
348
273
5R
2w
ithin
0.0
00
0.2
34
0.2
41
0.2
56
0.0
00
0.2
63
0.2
72
0.2
82
R2
overa
ll0.0
54
0.6
53
0.6
53
0.6
94
0.1
15
0.6
84
0.6
96
0.7
46
Panel
C
Panel
C1:
Em
erg
ing
countr
ies
Panel
C2:
Develo
ped
countr
ies
Model
IM
odel
IIM
odel
III
Model
IVM
odel
IM
odel
IIM
odel
III
Model
IVV
ari
able
sC
oeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
Coeff
.P
-Val.
PV
T0.3
52
0.0
03
0.0
75
0.0
76
4.4
93
0.0
95
6.1
85
0.0
40
0.1
49
0.0
31
0.1
63
0.0
66
-0.2
73
0.8
56
-0.9
06
0.5
43
SIZ
E*
PV
T-0
.011
0.2
11
0.0
32
0.1
30
-0.0
05
0.3
28
-0.0
06
0.2
69
ER
*P
VT
0.0
31
0.0
74
-0.0
37
0.0
19
0.0
01
0.8
14
0.0
04
0.4
66
RE
TE
*P
VT
0.0
05
0.0
97
0.0
09
0.0
50
-0.0
01
0.2
66
-0.0
01
0.3
89
CA
SH
*P
VT
0.0
03
0.6
49
0.0
15
0.1
68
-0.0
03
0.3
49
-0.0
01
0.7
07
CL
OSE
*P
VT
0.0
05
0.3
68
0.0
10
0.3
28
-0.0
03
0.4
25
-0.0
02
0.5
42
NI
Ris
k*
PV
T-0
.041
0.9
71
-8.4
01
0.1
38
0.1
10
0.3
22
0.1
08
0.3
43
GT
A*
PV
T0.0
05
0.2
22
-0.0
12
0.0
84
-0.0
01
0.7
04
0.0
00
0.8
01
MT
BV
*P
VT
0.0
17
0.7
84
-0.2
77
0.0
20
0.0
20
0.3
70
0.0
19
0.4
52
GSale
s*
PV
T-0
.002
0.4
98
0.0
15
0.0
16
0.0
02
0.0
73
0.0
02
0.0
52
LR
*P
VT
-0.0
16
0.1
10
-0.0
35
0.0
01
-0.0
07
0.2
09
-0.0
06
0.3
02
ER
F*
PV
T-0
.052
0.4
94
0.1
47
0.1
99
0.0
62
0.2
69
0.0
73
0.2
23
Ln
Sale
sE
mp
*P
VT
0.3
59
0.0
51
0.6
96
0.1
00
0.0
21
0.8
28
0.0
18
0.8
53
Ln
EM
P*
PV
T0.0
44
0.7
60
-0.8
75
0.0
01
-0.1
50
0.0
70
-0.1
38
0.0
97
CO
M*
PV
T-0
.859
0.0
51
N/A
-0.1
59
0.4
18
-0.0
18
0.9
35
DT
P*
PV
T0.1
13
0.9
75
0.0
27
0.9
45
Obse
rvati
on
5777
3266
3266
2533
8331
5101
5101
4741
Fir
ms
407
370
370
255
581
537
537
482
R2
overa
ll0.0
57
0.6
29
0.6
39
0.7
18
0.0
95
0.6
89
0.6
89
0.7
14
46
Recommended