Political Connections and Voluntary Disclosure: Evidence From Around The World
Political Connections and Voluntary Disclosure: Evidence From Around The World
Political Connections and Voluntary Disclosure: Evidence From Around The World
Presented by
Dr Yongtae Kim
#2016/17-09
The views and opinions expressed in this working paper are those of the author(s) and
not necessarily those of the School of Accountancy, Singapore Management University.
Political Connections and Voluntary Disclosure: Evidence
from Around the World
Mingyi Hung
The Hong Kong University of Science and Technology
Yongtae Kim
Santa Clara University
Siqi Li
Santa Clara University
ABSTRACT
This study investigates whether and how political connections affect managers’ voluntary
disclosure choices. We find that compared to non-connected firms, connected firms issue fewer
management earnings forecasts, and the difference is primarily driven by firms in developed
capital markets. Further, connected firms, especially those in countries with more developed
markets, increase management forecasts subsequent to exogenous political shocks that damage
their political ties. We also find that connected firms enjoy a relatively low cost of debt
regardless of their voluntary disclosure choices, whereas non-connected firms can lower their
cost of debt by providing more voluntary disclosure. Our results are robust to using conference
calls as an alternative measure of voluntary disclosure. Overall, our evidence suggests that
preferential access to credit and weak capital market incentives shape connected firms’ voluntary
disclosure practices.
Data availability: All data are publicly available from sources indicated in the text.
Political Connections and Voluntary Disclosure: Evidence
from Around the World
I. INTRODUCTION
Politically connected firms are economically important and the focus of a large body of
literature. Prior research suggests that political ties are associated with various benefits and costs
to firms and their shareholders.1 However, an important and unanswered question remains as to
whether and how political connections affect managers’ voluntary disclosure choices. Politically
connected firms may provide less voluntary disclosure because their preferential access to capital
simply makes them subject to less capital market pressure, or because their managers have
incentives to maintain opaque information environments in order to mask true performance and
extract political rents (Chen, Ding, and Kim, 2010; Chaney, Faccio, and Parsley, 2011).
Alternatively, politically connected firms may provide more voluntary disclosure because they
are subject to more public scrutiny and their managers have a greater need to convince outside
investors that they do not engage in self-dealing (Watts and Zimmerman, 1983; Guedhami,
Pittman, and Saffar, 2014). In this paper, we address this question by assessing the effect of
important communication channel through which managers convey their expectation of firms’
future performance to the capital market (Hirst, Koonce, and Venkataraman, 2008). Following
Chaney et al. (2011) and Guedhami et al. (2014), we capture political connections using the data
1
Examples of these benefits include access to credit from government-owned banks, lower tax burdens, lax
regulatory enforcement, and receipt of government contracts and financing (Adhikari, Derashid, and Zhang, 2006;
Faccio, Masulis, and McConnell, 2006; Claessens, Feijen, and Laeven, 2008; Correia, 2014). Examples of costs
include rent extraction by politicians and excessive government intervention that results in inefficient investment
(Khwaja and Mian, 2005; Fan, Wong, and Zhang, 2007).
1
from Faccio (2006), who define a firm as politically connected if at least one of its large
shareholders and top directors is a member of parliament, a minister or the head of state, or is
closely related to a top official. Our sample consists of 547 (28,006) firm-year observations
representing 208 (11,466) politically connected (non-connected) firms in 24 countries from 2002
to 2004.2 Because political connections are not random, we implement the two-stage regression
procedure suggested by Heckman (1979) to control for the effect of self-selection. Our selection
model indicates that political connections are more prevalent among firms that are larger, more
mature, and headquartered in a nation’s capital city. Connected firms are also more likely to be
in an industry where a greater percentage of industry peers are connected and less likely to be in
Importantly, after controlling for the self-selection effect, we find that politically connected firms
are associated with less frequent management forecasts. The association is economically
significant. For example, the average number of management forecasts issued by connected
firms in a given year is 77.4% lower relative to forecasts issued by non-connected firms.
To the extent that political connections deter the issuance of management forecasts, we seek
explanations for the negative relation between political connections and the level of management
forecasts. The capital market incentive explanation suggests that relative to non-connected firms,
connected firms issue fewer management forecasts due to a lack of capital market pressure for
external capital. Unlike connected firms that have preferential access to credit, non-connected firms
need to access external capital and greater information transparency helps reduce the cost of capital.
Because the demand for transparency is associated with the development of capital markets, this
2
Faccio (2006) identifies politicians for most countries as of 2001. We begin our sample period in 2002 because the
coverage of Capital IQ’s corporate guidance information begins in 2002. We end the sample period for our primary
analyses in 2004, the first year in which one of our sample countries experiences a political realignment due to
major elections, to avoid measurement errors in our political connection variable.
2
explanation predicts that the negative relation between political connections and the level of
management forecasts is more pronounced in countries with more developed capital markets. The
political rent seeking explanation posits that relative to non-connected firms, connected firms issue
fewer management forecasts due to greater incentives to mask their rent seeking activities. Because
rent extractions of politicians and connected firms are associated with the levels of corruption in a
country, this explanation predicts that the negative relation between political connections and the
We find that the capital market incentive explanation dominates the political rent seeking
explanation. Specifically, supporting the capital market incentive explanation, we find that the
negative relation between political connections and management forecasts is more pronounced in
countries with more developed capital markets, as proxied by stock market capitalization or
stock market efficiency. Contrary to the political rent seeking explanation, we find that that the
negative relation between political connections and management forecasts is more pronounced in
countries with less corruption, as proxied by the corruption indexes compiled by the International
Next, to shed light on the causal link between political connections and management
forecasts, we investigate whether an exogenous shock to political ties affects the frequency of
management forecasts for connected firms. We take advantage of worldwide presidential and
legislative elections that result in political realignment to capture the shock to the existing
political ties. Specifically, we use a sample of firms in thirteen countries that experience
elections resulting in power turnovers (i.e., elections that involve a change of the ruling party in
countries with a parliamentary system and a change of the president in countries with a
presidential system) between 2004 and 2010. Our difference-in-differences analysis finds that
3
connected firms increase the frequency of management forecasts subsequent to these elections,
compared to non-connected firms in the same countries. This finding supports our inference that
political connections cause less frequent management forecasts. In addition, consistent with the
capital market incentive explanation, we find that changes in management forecasts after the
elections are primarily driven by connected firms in countries with more developed markets and
less corruption. We also find that connected firms are more likely to forecast additional line
items and provide accompanied explanations for their forecasts subsequent to the elections. In
sum, these findings confirm our capital market incentive explanation and suggest that connected
firms respond to market pressure for transparency upon the loss of political ties.
Additional analyses reveal that while non-connected firms can lower their cost of debt
through more voluntary disclosure, creditors provide connected firms with relatively cheap
capital regardless of their disclosure choices. This finding is consistent with the notion that
politically connected firms’ easy access to cheap credit provides disincentive for voluntary
disclosure. Further, our results continue to hold using conference calls as an alternative measure
of voluntary disclosure.
checks. Faccio (2006) classifies political connections into more or less objective categories. We
find that the effect of political connections on management earnings forecasts is greater among
firms whose political ties are measured more objectively (i.e., where top shareholders and
director serve as a member of parliament or a minister), relative to firms whose political ties are
measured more subjectively (i.e., where top shareholders and director have close relationship
with a top official). In addition, we consider state ownership as another form of political
connections and find that state-owned enterprises are associated with a lower level of
4
management earnings forecasts. More importantly, we find that the effect of high-level political
connections remains significant even after we control for connections through state ownership.
Further, we find that our results are robust to controlling for earnings quality, analyst forecast
accuracy, and family ownership, using a matched sample, excluding preliminary earnings
announcements, restricting firms with analyst following, dropping influential countries, and
Our study contributes to the literature in two important ways. First, we add to the
literature of corporate reporting environments by being the first to examine whether and how
political connections influence voluntary disclosure. While several recent studies examine
corporate reporting environments of politically connected firms (Chen et al., 2010; Chaney et al.,
2011; Guedhami et al., 2014), they yield mixed evidence on politically connected firms’
preference for information transparency. For example, Chaney et al. (2011) find that connected
firms have poorer earnings quality than non-connected firms, but Guedhami et al. (2014) find
that connected firms are more likely to appoint Big N auditors than non-connected firms. Further,
the findings of these studies speak only indirectly to the relation between political connections
and firms’ voluntary disclosure choices. 3 Our finding, in line with Chaney et al. (2011) but
contrary to Guedhami et al. (2014), suggests that politically connected firms prefer information
opacity. In addition, we extend this prior work by exploring two different managerial incentives
3
The finding of earnings quality is not necessarily generalizable to voluntary disclosure practices because poorer
earnings quality, as typically measured by higher discretionary accruals, may be driven by various motivations
including: (1) meeting or beating markets’ expectations of earnings, (2) improving the informational value of
earnings, and (3) opportunistic earnings management to increase managers’ compensation or mask true performance
(Subramanyam, 1996; Ayers, Jiang, and Yeung, 2006). While the first two motivations can be associated with more
management forecasts to facilitate expectation management or convey managerial private information, the third
motivation can be associated with fewer management forecasts to hide political favors. In addition, politically
connected firms may appoint Big N auditors as a way to (1) improve information transparency and credibility, or (2)
leverage the insurance value of large auditors. It is also possible that firms self-select into building their political
connections and expanding their political networks, which may correlate with their choices of Big N auditors due to
these auditors’ extensive networks and connections.
5
that shape politically connected firms’ disclosure choices. Our results support the view that
managers of connected firms provide less voluntary disclosure because of weaker capital market
incentives rather than greater rent seeking incentives. By differentiating the alternative
explanations behind the disclosure practices of politically connected firms across countries, we
also complement prior work examining the effect of country-level institutions on corporate
Second, we contribute to a growing body of research that examines the effect of political
influence on firm behavior and disclosure choices (Shleifer and Vishny, 1994; Miller and
Skinner, 2015). By taking advantage of changes in political power associated with major
endogeneity concerns and provide insight into how the loss of political ties affects firms’
disclosure practices across different institutional environments. Our evidence indicates that
losing political ties in countries with more developed markets motivates firms to increase
voluntary disclosure, but this effect is weak or nonexistent in countries with less developed
markets. Further, while prior studies suggest that the loss of political ties motivates firms in less
securities (Leuz and Oberholzer-Gee, 2006), our study suggests that these firms’ overall
incentives of voluntary disclosure remain weak, likely because their domestic institutional
arrangement remains relationship-based and they have greater incentive to rebuild political ties.
disclosure practices in global markets, we also extend recent single-country studies that
document the impact of political events on accounting choices and information environments
(Ramanna and Roychowdury, 2010 for the US; Piotroski, Wong, and Zhang, 2015 for China).
6
The rest of the study is organized as follows. Section II discusses the institutional
background and develops the hypothesis. Section III describes the research design, Section IV
discusses the sample and descriptive statistics, and Section V presents the empirical results.
Section VI presents the results of additional analyses and robustness checks. Section VII
A longstanding literature shows that political connections add value to the connected
firms. Fisman (2001) finds that at the time of Indonesian President Suharto's worsening health,
stock prices of firms closely connected with Suharto dropped more than the prices of less well-
connected firms. In a cross-country study, Faccio (2006) finds that stock prices rise when
officers or large shareholders of a firm enter politics. Studies also document that the benefits of
political connections can take various forms including: access to bank finance and lower cost of
capital (Claessens et al., 2008), lower tax burdens (Adhikari et al., 2006), lax regulatory
enforcement (Correia, 2014), and receipt of government contracts and support (Faccio et al.,
2006).
An important implication from this literature is that political connections and political
favors could affect firms’ preference for voluntary disclosure. Disclosure theories suggest that
voluntary disclosures reduce information asymmetry, which in turn, leads to a lower cost of
capital (Diamond and Verrecchia, 1991). There are two channels though which political
connections may reduce firms’ incentives for voluntary disclosure. First, compared to non-
connected firms, connected firms have weaker capital market incentive for voluntary disclosure
because the benefits of disclosure in reducing the cost of capital accrue less to politically
connected firms. We refer to this channel as “the capital market incentive explanation.”
7
Specifically, a firm’s need to access external financing affects the level of investor demand for
its earnings information, which in turn affects the expected benefits of providing voluntary
disclosure. Since politically connected firms have better access to credits and obtain privileged
loans from banks that are influenced by politicians, they have a lesser need to raise capital from
the public and therefore a lower incentive for voluntary disclosure. In addition, because of
political favors that the connected firms enjoy, the cost of capital may be already relatively low
Second, politically connected firms may prefer information opacity to obscure their gains
from politicians. We refer to this channel as “the political rent seeking explanation.” Insiders of
politically connected firms have incentives to divert benefits brought by political connections,
which in turn motivate them to maintain opaque information environment in order to divert
may want to suppress information on true economic performance in order to ensure that their
diversionary practices, largely stemming from political cronyism and corruption, are kept hidden.
Consistent with the notion that politically connected firms prefer opaque information
environments, prior studies find that connected firms have greater analyst forecast errors and
There are, however, reasons to expect a positive relation between political connections and
voluntary disclosure. Specifically, insiders of politically connected firms may want to provide
more disclosure to convince outside investors that they do not engage in self-dealing. In
particular, connections to high-level politicians (i.e., members of parliament and ministers) are
highly visible and subject to great public scrutiny.4 Because more reliable financial reporting and
4
For example, firms connected with members of parliament during our sample period include well-known
companies such as Fiat from Italy, British Petroleum from the UK, and Intel from the US.
8
information disclosure help prevent expropriation by insiders and their political patrons, there is
a stronger market demand for information transparency for politically connected firms (Watts
and Zimmerman, 1983). Consistent with this reasoning, Guedhami et al. (2014) find that
politically connected firms are more likely to appoint a Big N auditor. Thus, the relation between
political connections and voluntary disclosure is an empirical question. Our hypothesis, stated in
We obtain data on political connections from Faccio (2006), who develops a dataset of
politically connected firms worldwide as of 2001. The data are commonly used in prior studies
examining political connections in global markets (Chen et al., 2010; Chaney et al., 2011;
Guedhami et al., 2014). According to Faccio (2006), a firm is classified as politically connected
if at least one of its large shareholders (anyone directly or indirectly controlling at least 10% of
votes) or top directors (CEO, chairman of the board, president, vice-president, or secretary) is a
define an indicator variable, PC, which takes a value of one for connected firms, and zero
otherwise.
It is worth noting that Faccio (2006) focuses on connections with high-profile politicians,
and does not include connections via campaign contributions or state ownership. Campaign
contributions and cash payments to politicians are generally unobservable and difficult to
identify in the global setting. In addition, the political ties documented in Faccio (2006) are
9
likely to be more durable and thus provide a more powerful setting to detect the impact of
because they are one of the most widely examined forms of voluntary disclosure (Hirst et al.,
2008). Management earnings forecasts are an important form of voluntary disclosures that
provide information about forthcoming earnings. These forecasts represent a key voluntary
disclosure mechanism by which managers establish or alter market earnings expectations and
influence the information environment of a firm (Pownall, Wasley, and Waymire, 1993; Coller
We obtain management forecasts from Standard & Poor’s Capital IQ, which collects
worldwide corporate guidance information in text format starting in 2002. Capital IQ relies on
public data sources including press releases and news wire articles, regulatory files, company
websites, web agents, conference call transcripts, and investor conference organizer websites.
While this database is relatively new to the literature, it has been increasingly used in recent
studies examining management forecasts around the world (Radhakrishnan, Tsang, and Yang
for “earnings”, “Earnings”, or “EPS” in headlines and main texts under the “Corporate Guidance”
event type from Capital IQ’s Key Developments Database. We do not differentiate annual versus
quarterly earnings forecasts because our interest is on how political connections affect voluntary
5
Our additional analyses in Table 8 suggest that more direct and visible political ties are more negatively associated
with management forecasts. In addition, state ownership, an alternative measure of political connections, is also
negatively associated with management forecasts.
6
We address the potential issues regarding Capital IQ’s coverage in sensitivity tests of Section VI.
10
disclosure in general. We treat multiple forecasts by the same firm on the same date as a single
forecast. Our measure of voluntary disclosure, Freq, captures the frequency of the disclosure and
equals the natural logarithm of one plus the number of management forecasts issued in a given
year.
Research Design
Because politically connected firms do not represent a randomly selected sample (Faccio,
2006), we implement the Heckman (1979) two-stage regression procedure to control for the
effect of self-selection in our analyses. In the first stage, we model the determinants of political
connections by estimating a probit model in which the dependent variable is an indicator variable
(PC) with a value of one for connected firms and zero for non-connected firms. The independent
variables are the factors influencing firms’ decisions to establish political connections. In the
second stage, we regress our voluntary disclosure measure, Freq, on the indicator, PC, and a set
of control variables including the inverse Mills ratio (Lambda) estimated from the first-stage
probit model.7
political connections based on prior literature (Schuler, Rehbein and Cramer, 2002; Hillman,
Keim, and Schuler, 2004; Faccio, 2006; Chaney et al., 2011; Guedhami et al., 2014): (1) firm
size (Size), calculated as the natural logarithm of total assets in US dollars, because larger firms
have greater resources and tend to be more politically active; (2) firm age (LnAge), measured as
7
Inverse Mills ratio is calculated differently for treated firms (i.e., connected firms) and untreated firms (i.e., non-
connected firms) (Tucker, 2010).
11
the natural logarithm of number of years since the IPO date, because older firms are more likely
to have political connections; (3) free cash flows (FreeCash), measured as operating income
before depreciation and amortization minus income taxes less changes in deferred taxes, interest
expense, preferred dividends and common dividends, deflated by total assets (Lehn and Poulsen,
1989), because firms with more free cash flows can afford to engage in political activities; (4) a
variable indicating the location of a firm’s headquarters (Capital), which equals one if a firm’s
corporate headquarters is located in the nation’s capital city, and zero otherwise, because the
location of a firm’s headquarters affects the formation of political connections; and (5) industry
Herfindahl index and percentage of connected firms in a two-digit SIC industry, respectively, as
industry characteristics have been shown to correlate with a firm’s political activities.
Lopez-de-Silanes, Shleifer, and Vishny (1998) and Faccio (2006): (1) corruption (ICRGCorrupt),
securities or outward direct investment by citizens, because capital restrictions ensure connected
firms’ access to domestic capital; (3) regulatory environment (RegScore), measured as the
regulatory score constructed in Faccio (2006), to capture regulations that prohibit or set limits on
the business activities of public officials; (4) economic development (LnGDP), measured as the
natural logarithm of GDP per capita in 2001; and (5) three legal-origin indicators (German,
French, and Scandinavian), to capture the German, French, and Scandinavian civil-law traditions.
Finally, we include year fixed effects. Appendix A provides detailed definitions of the variables.
12
Capital, IndustryPC, and RegScore work as exogenous variables to satisfy exclusion restrictions
because a firm’s location in the capital city, industry-level political connections, and limits on the
business activities of public officials are unlikely to directly affect firm-specific voluntary
disclosure levels.
We estimate an OLS model with our voluntary disclosure variable, Freq, as the dependent
variable. Our variable of interest is the political connection indicator, PC. A negative (positive)
coefficient on PC indicates that politically connected firms have a lower (higher) level of
management forecasts than non-connected firms. Following Chaney et al. (2011), we cluster
We control for various factors that prior literature identifies to affect firms’ voluntary
disclosure choices (Chen et al., 2008; Li and Yang, 2016). Our control variables include: (1) firm
size (Size), defined as the natural logarithm of total assets in US dollars, (2) return on assets
(ROA), defined as net income deflated by total assets, (3) market-to-book ratio (MTB), calculated
as market capitalization divided by book value of equity, (4) leverage (LEV), calculated as the
long-term debt deflated by total assets, (5) earnings volatility (EarnVol), calculated as the
standard deviation of earnings over assets for the past five years, (6) return volatility (RetVol),
calculated as the standard deviation of annual stock returns over the past five years, (7) the
number of analyst following (NAnalyst), (8) an indicator variable that equals one if a firm
8
As shown in Table 8, our results are robust to alternative clustering schemes.
13
experiences a negative earnings change in the year (BadNews), (9) a variable indicating equity
issuance in the subsequent year (EquityIssue), defined as a dummy variable equal to one if a
firm’s total number of common shares outstanding after adjusting for stock splits and dividends
increases by 20% or more in the next year, (10) a variable indicating whether a firm is cross-
listed in the US (Cross), (11) an indicator variable that equals one for Big N auditors (BigN), (12)
an indicator variable that equals one for the use of International Accounting Standards (IAS),9
and (13) a measure of a firm’s ownership structure (Closeheld), defined as the number of closely
held shares divided by total shares outstanding. We also include year, industry, and country fixed
effects to control for the variation of management forecasts across different years, industries, and
countries. Throughout our analyses, we winsorize all scaled variables, including ROA, MTB, LEV,
EarnVol, RetVol, and Closeheld, at the top and bottom 1% of their distribution to mitigate the
influence of outliers.
Sample
We start with a list of 541 politically connected firms from 34 countries in 1997-2001 as
identified in Faccio (2006), and match these connected firms with data on management earnings
forecasts from Capital IQ. Since complete corporate guidance information in Capital IQ starts
from 2002, we restrict our sample period to 2002-2004, by which we assume that the political
connections established in 1997-2001 continue to hold in the subsequent three years.10 We next
9
We use the term IAS to refer to both the International Accounting Standards issued by the International
Accounting Standards Committee (IASC), and the International Financial Reporting Standards (IFRS) issued by its
successor, the International Accounting Standards Board (IASB).
10
We end our sample period in 2004 because this is the first year in which one of our sample countries experiences
a political realignment from major elections. Our results (untabulated) remain qualitatively the same if we extend the
sample period to 2005, as in Chaney et al. (2011) and Guedhami et al. (2014). We also find that our results
14
construct a sample of non-connected control firms in these 34 countries over the same period.
We exclude companies in Japan because their management forecasts are mandated (Kato,
Skinner, and Kunimura, 2009). After requiring financial data to be available in Compustat and
Worldscope, our final sample consists of 547 (28,006) firm-year observations representing 208
Descriptive Statistics
Table 1, Panel A reports the sample distribution by country. It indicates a wide cross-
country variation in both the connected and non-connected firms. For example, firms from the
UK and Malaysia dominate the politically connected sample, accounting for 32% and 26% of all
connected firm-year observations in this study, respectively; while the rest of the countries each
represents less than 7% of the connected firms. This pattern is consistent with prior studies such
as Chaney et al. (2011) and Guedhami et al. (2014). Among the non-connected control firms, the
US has the largest number of observations (12,893) and Hungary has the smallest (34).11
Panel B of Table 1 presents the institutional characteristics of our sample countries. Nine
out of our 24 sample countries have at least one restriction on cross-border capital flows
(XborderRestrict). Philippines has the highest regulatory score (RegScore), indicating the most
stringent regulatory environment that prohibits or sets limits on the business activities of public
officials, while six countries have no such regulations (i.e., Belgium, India, Indonesia, Malaysia,
Mexico, and Taiwan). Panel B also shows that Switzerland has the highest GDP per capita
(LnGDP) in 2001 while India has the lowest. As for the legal origin (Legal Origin), eleven, six,
five, and two of our sample countries have English, German, French, and Scandinavian legal
(untabulated) are robust to removing the firm-year observations in countries that experience elections resulting in
political realignment.
11
As reported in Table 8, our results are robust to using a control sample matched on country, year, and two-digit
SIC industry, as well as excluding countries with the largest number of connected firms.
15
origin, respectively. Hong Kong has the highest stock market capitalization over GDP (MktCap)
in our sample period and Mexico has the lowest. The US has the highest stock market efficiency
score (StockmktEff) while Mexico has the lowest. Based on the ICRG corruption index
(ICRGCorrupt), Philippines is the most corrupt country with the highest score in our sample and
Canada, Finland, Sweden, and Switzerland are the least corrupt countries with the lowest score.
Based on the German corruption index, on the other hand, Indonesia and Thailand have the
highest value of five in our sample and 14 countries have the lowest value of zero.
Table 2 presents descriptive statistics across connected and non-connected firms. We find
that for politically connected firms, the average number of management earnings forecasts issued
over a year is 0.25, which is significantly lower than the number for non-connected firms (0.47),
and the difference is statistically significant at the 0.01 level. In comparison to non-connected
firms, we also find that connected firms, on average, are larger (Size) and older (LnAge), have
more free cash flows (FreeCash), are more likely to have their headquarters located in the
nation’s capital city (Capital), are in an industry with less competition (Herf) and more
connected peer firms (IndustryPC), are more leveraged (LEV), less likely to issue equity in the
subsequent year (EquityIssue), more likely to cross-list in the US (Cross), more likely to hire a
Big N auditor (BigN), more likely to adopt International Accounting Standards (IAS), have better
accounting performance (ROA), less volatile earnings and returns (EarnVol and RetVol), greater
analyst coverage (NAnalyst), and more closely held shares (Closeheld). These differences are
generally consistent with prior research (Chen et al., 2010; Guedhami et al., 2014).
V. EMPIRICAL ANALYSIS
16
Table 3 reports the Heckman two-stage regression results of the effect of political
disclosure. For brevity, we suppress reporting of the coefficients on fixed effects in this and all
subsequent tables.
Column (1) of Table 3 reports the results of the first-stage probit model. It shows that
political connections are more prevalent in larger and more mature firms, consistent with the
findings in Hilman et al. (2004). As in prior studies (Schuler et al., 2002; Hillman et al., 2004;
Chaney et al., 2011), we also find that firms are more likely to be politically connected when
their headquarters are located in a nation’s capital city, or when there is a greater percentage of
industry peers that are connected. In addition, consistent with Faccio (2006), political
connections are less common in countries with stricter regulations limiting business activities of
government officials. Finally, political connections are less prevalent in countries with German
legal origin.
Column (2) of Table 3 presents the results of the second-stage OLS regression.12 We find
that the coefficient on the political connection indicator, PC, is negative and statistically
significant at the 0.01 level. To gauge the economic significance of the result, we take the
exponential value of the coefficient on PC and then subtract one. We find that the average
number of management forecasts issued by connected firms in a given year is 77.4% lower
relative to forecasts issued by non-connected firms.13 Thus, the effect of political connection on
the frequency of management forecasts is economically significant. Column (2) shows that
several control variables are significant at the 0.10 level or better. Specifically, the frequency of
12
In untabulated analysis, we use a negative binomial regression to model the effect of political connections on the
frequency of management forecasts in the second stage and continue to find a negative coefficient on PC (significant
at the 0.10 level).
13
-77.4% = (exp(-0.457)-1)/0.474, where -0.457 is the coefficient on PC in column (2) of Table 3 and 0.474 is the
average number of forecasts for non-connected firms in Table 2.
17
management forecasts increases in firm size, return on assets, market-to-book ratio, earnings
volatility, and analyst coverage, and decreases in leverage and the percentage of closely held
shares. These findings are generally consistent with prior studies such as Chen, Chen, and Cheng
(2008). Finally, the coefficient on Lambda is significantly positive at the 0.01 level, indicating a
significant self-selection effect in our sample.14 In summary, the results in Table 3 suggest that
relative to non-connected firms, politically connected firms issue less frequent voluntary
disclosure.
management forecasts documented in Table 3 is driven by the lack of capital market incentive or
political rent extraction. We perform this analysis by re-estimating equation (2) after partitioning the
sample based on the median values of the degree of country-level capital market development and
the extent of country-level corruption. If the capital market incentive is the primary driver of the
different voluntary disclosure practices between politically connected firms and non-connected firms,
we expect our results to be more pronounced in countries with more developed markets than in
countries with less developed markets. Alternatively, if political rent seeking is the primary driver of
the different voluntary disclosure practices between connected firms and non-connected firms, we
expect our results to be more pronounced in countries with more corruption than in countries with
less corruption.
We measure capital market development using two proxies: (1) stock market
capitalization, MktCap, which equals the market capitalization of listed firms scaled by GDP,
and (2) stock market efficiency, StockmktEff, which indicates whether stock markets provide
14
We also estimate the regression in column (2) of Table 3 without controlling for the effects of self-selection. The
results are qualitatively the same as those reported in Table 3, with the coefficient on PC statistically significant at
the 0.05 level.
18
adequate financing to companies (Ghoul, Guedhami, and Kim, 2016). 15 Both measures are
averaged over 2002-2004 and higher values indicate more developed capital markets. To
measure corruption, we follow Faccio (2006) and use the International Country Risk Guide’s
assessment of the corruption in governments based on La Porta et al. (1998) (ICRGCorrupt), and
a corruption index based on interviews with German exporters as developed by Neumann (1994)
country.17
Panel A of Table 4 reports the regression results of analyses conditional on capital market
development. It shows that the coefficient on PC is more negative in countries with larger market
capitalization or with greater stock market efficiency, and the difference across partitions is
significant at the 0.01 level. This result supports the capital market incentive explanation, which
predicts the difference in management forecast frequency between connected and non-connected
firms to be more pronounced in countries with more developed markets. Panel B of Table 4
reports the regression results of analyses conditional on corruption. It shows that the coefficient
on PC is more negative in countries with less corruption, and the difference in the coefficient on
PC across partitions is significant at the 0.01 level. This result is inconsistent with the political
rent seeking explanation, which predicts the difference in management forecast frequency
between connected and non-connected firms to be more pronounced in countries with more
corruption. In summary, the results in Table 4 suggest that lower financing need and weaker
capital market incentives, rather than greater political rent seeking and the desire to mask
15
The correlation between the two capital market development proxies is 0.60 and significant at the 0.01 level.
16
The correlation between the two corruption proxies is 0.93 and significant at the 0.01 level.
17
Additional (untabulated) analysis suggests that the results in Table 4 are robust to alternative measures of capital
market development (e.g., stock market turnover scaled by GDP or the number of listed companies scaled by
population) and corruption (e.g., the corruption index, which captures the exercise of public power for private gains,
from Kaufmann, Kraay, and Mastuzzi (2007) or the control of corruption index from the Worldwide Governance
Indicators).
19
political favors, drive connected firms to make less voluntary disclosure than non-connected
firms.
While our inferences are that relative to non-connected firms, politically connected firms
have less incentive to issue management forecasts to improve information transparency, one
could argue that this difference is driven by unobservable factors that determine voluntary
disclosure. To bolster our inferences, we adopt an event study approach by examining the change
in voluntary disclosure practice for connected firms subsequent to political events that damage
their political ties. If political connections are indeed the reason for a lower level of management
forecasts, we expect connected firms to increase voluntary disclosure subsequent to these events.
To capture political events that damage the political ties of connected firms, we use
elections that result in power turnovers. We utilize the World Bank Database of Political
Institutions (Beck et al., 2001), which collects information on worldwide presidential and
legislative elections from 1975 to 2012. We define elections with power turnovers as those
electing a new president in countries with presidential systems (e.g., Taiwan and the US) or a
new parliament under the leadership of a different party in countries with parliamentary systems
(e.g., Austria and Canada). Since political connections reflect the ties with the existing head of
state or members of parliament, elections that result in turnover of the incumbent president and
ruling parties should damage such a connection. Thirteen countries in our sample experience
in 2002 and we wish to examine firms’ voluntary disclosure behavior two years before and after
18
During our sample period (i.e., 2002-2004), only Indonesia has an election with a power turnover. Our results in
Table 3 are robust to excluding the politically connected firms from Indonesia in the election year, 2004.
20
the elections.
Table 5 presents the results of this analysis. Panel A describes the realigning elections
and shows that seven sample countries have parliamentary systems and six have presidential
systems. It also suggests a wide variation in the election years across our sample countries, which
helps strengthen our identification strategy by mitigating the undue influence of unobservable
factors common across the nations at a particular time. The sample distribution in Panel B
indicates that the UK dominates the sample of connected firms (128 firm-year observations)
while the US dominates the sample of non-connected firms (10,960 firm-year observations).
Panel C of Table 5 presents the regression analysis comparing the frequency of issuing
management forecasts for connected firms versus non-connected firms in the two years before and
after the realigning elections. We regress the proxy of voluntary disclosure (Freq) on the indicator
for connected firms (PC), the indicator for post-election years (Post), their interaction, and the set
of control variables as in equation (2). We do not include the inverse Mills ratio from the Heckman
approach to control for the confounding factors. However, in untabulated analysis we find that the
tenor of our results remains unchanged if we include the inverse Mills ratio.
Column (1) of Table 5, Panel C shows that the coefficient on PC is negative and significant
at the 0.01 level. This result is consistent with the result in Table 3 and indicates a lower frequency
of management forecasts for connected firms in the two years before the elections. More
importantly, the coefficient on the interaction term PC×Post is significantly positive (at the 0.05
level), consistent with the notion that following the realigning elections politically connected firms
increase management forecasts more than non-connected firms. Columns (2) to (5) report the
results of conditional on the degree of country-level capital market development and corruption.
21
For brevity, we present only the results using MktCap as the proxy for capital market development
and ICRGCorrupt as the proxy for corruption, but the results are robust to using StockmktEff or
GermanCorrupt as alternative proxies. These columns show that the coefficient on PC×Post is
significantly positive only in countries with more developed markets or with less corruption. These
results are consistent with those in Table 4 and suggest connected firms in countries with more
developed capital markets respond more to market pressure for transparency by increasing their
In Panel D of Table 5, we break down political connections into two types: firms connected
to government officials or members of parliaments (GovParli) and firms with close relationships to
a top official (Relation). Faccio (2006) suggests that political connections can be classified into
more or less objective categories. Specifically, a firm’s connection is more objectively measured
through close friendship or other relations (Relation) because the necessity of relying on publicly
column (1), we find a significant and negative coefficient on GovParli (at the 0.01 level) but an
insignificant coefficient on Relation. This result is consistent with Faccio (2006) and indicates that
the negative relation between political connections and the frequency of management forecasts
prior to the elections is significant only for the more objective type of connections (GovParli).
Further, column (1) reports a significantly positive coefficient on GovParli×Post (at the 0.01 level)
19
In untabulated analysis, we also examine the changes in earnings quality (proxied by the standard deviation of
discretionary accruals as in Chaney et al., 2011) and the use of Big N auditors for politically connected firms two
years before and after the elections, relative to the changes for non-connected firms in the same countries over the
same period. We find that compared to non-connected firms, firms with political ties experience an improvement in
earnings quality subsequent to the elections that result in power turnovers. In contrast, there is no significant change
in the use of Big N auditors among connected firms following the elections relative to non-connected firms.
22
and a significantly negative coefficient on Relation×Post (at the 0.10 level), consistent with the
notion that the increase in the number of management forecasts following the elections is present
only among firms connected with government officials or members of parliaments. Further, there
is some evidence suggesting that when political connections are through close relationships with
top officials, connected firms issue even fewer management forecasts after the elections, perhaps
due to added uncertainty after power turnovers. Similar to the results in Panel C, columns (2) to (5)
find that the coefficient on GovParli×Post is significantly positive only in countries with more
connected firms from Capital IQ. We examine the changes in the number of additional line items
(Explanation), whether forecasted earnings is a loss (Loss), and the degree of specificity in
forecast forms such as range or point estimates (Specificity). 20 Results in Table 5, Panel E
suggest that politically connected firms forecast more additional line items and are more likely to
provide accompanied explanations for the earnings forecasts following the elections. This is
consistent with connected firms expanding their disclosure once their political ties are damaged
after the realigning elections. Interestingly, we also find that connected firms become less
specific in making forecasts following the realigning elections. One possible explanation for this
finding is that uncertainty of connected firms’ fundamentals increases after they lose their
political ties and therefore these firms are unable to issue forecasts with high specificity. Panel E
also indicates that these results hold among a constant sample of connected firms issuing
20
We limit the analysis of forecast properties to politically connected firms to make hand-collection of forecast
property data manageable.
23
forecasts in both the pre- and post-election periods.
Our evidence so far suggests that relative to non-connected firms, politically connected
firms issue less frequent management earnings forecasts. Prior studies have shown that expanded
voluntary disclosure is associated with positive economic outcomes such as higher liquidity and
lower cost of capital (Botosan, 1997; Sengupta, 1998; Francis, Khurana, and Pereira, 2005). Thus,
a natural question arises as to why politically connected firms seem not to care about the benefits
associated with more voluntary disclosure. As discussed earlier, we posit that connected firms
already enjoy preferential access to credit and hence have less incentive to respond to market
empirically by testing whether the association between management forecasts and the cost of
We focus on the cost of debt because prior studies suggest that politically connected
firms have preferential access to credit and are able to avoid paying higher interest rates (Khwaja
and Mian, 2005). We measure the cost of debt as the ratio of interest expense in year t over the
average interest bearing obligations outstanding between year t and t-1 (Francis, LaFond, Olsson,
and Schipper, 2005; Chaney et al., 2011). To the extent that our cost of debt measure includes the
cost of both bank credit and public debt, the cost of debt would be lower for non-connected firms
with a higher level of management forecasts because voluntary disclosure will help reduce the
cost of public debt. We do not expect the level of management forecasts to affect the cost of debt
for connected firms because these firms enjoy preferential access to credits with or without
24
management forecasts and they have a lesser need for raising public debt.
Table 6 reports results of this analysis. We regress the cost of debt on PC, Freq, their
interaction term, and the set of control variables as in equation (2), as well as the interest
coverage ratio (IntCov) and natural logarithm of operating cycle (Opcycle), both of which have
been shown to affect interest rates. We report the results without the control variables in column
(1) and with the controls in column (2). The coefficient on PC is significantly negative only in
column (1). Importantly, we find that the coefficient on Freq is significantly negative at the 0.10
level or better in both columns, consistent with notion that non-connected firms with expanded
disclosure enjoy a lower cost of debt. The interaction term PC×Freq is insignificant at the
conventional levels in both columns. The sum of the coefficient on Freq and the coefficient on
the interaction term is also insignificantly different from zero in both columns, suggesting that,
for the sample of connected firms, issuing more earnings forecasts does not lead to a reduction in
the cost of debt. This result helps explain why connected firms have less incentive to issue more
frequent earnings forecasts. While non-connected firms can lower their cost of debt through
more information disclosure, creditors provide connected firms with relatively cheap capital
Conference calls are another effective channel through which managers communicate
their private information on firm performance (Bushee, Matsumoto, and Miller, 2003). In this
section, we repeat our analyses using conference calls as an alternative measure of voluntary
21
Our additional analyses (untabulated) also reveal that while connected firms tend to issue fewer management
forecasts, these forecasts, if issued, are more likely to reduce information uncertainty, measured by the average
changes in analyst forecast errors following the issuance of management earnings forecasts issued in the year. This
result is inconsistent with the argument that politically connected firms may deteriorate information environment
through intentionally-biased management forecasts. Increased disclosure has a greater effect on information
environment of politically connected firms potentially because their information environments are more opaque
prior to the disclosure.
25
disclosure. We obtain information on conference calls from Capital IQ and rerun our analysis in
Tables 3 and 4 by replacing the frequency of management forecasts with that of conference calls.
Our sample for the analysis of conference calls is identical to the sample in Table 1,
except that we now include 6,576 observations from Japan (69 connected observations
representing 27 unique firms and 6,507 non-connected observations representing 2,586 unique
firms). 22 In untabulated univariate analysis, we find that the average frequency of conference
calls for connected firms in a year (0.039) is significantly lower than that for non-connected
firms (0.173). Table 7, Panel A reports the regression results for the full sample and the results
conditional on capital market development. Panel B reports the results conditional on corruption.
Consistent with our findings in Tables 3 and 4, we find that connected firms hold less frequent
conference calls, and the difference is primarily driven by firms in countries with more
developed capital markets and less corruption. Thus, our findings are robust to using conference
Sensitivity Tests
In this section, we conduct several robustness checks of our results in Table 3 by examining
political connections, adding additional control variables, using alternative samples, and adopting
alternative schemes of clustering standard errors. Panel A of Table 8 reports the tests for
alternative definitions of political connections and additional control variables. Panel B of Table
8 reports the tests with alternative samples and adjustment to standard errors. We summarize the
22
We include Japan in this analysis because there is no mandatory requirement for conference calls in Japan. We
find that results are qualitatively identical even if we exclude Japanese firms (untabulated).
26
Examining different types of political connections. As in our analysis in Table 5, we investigate
whether the impact of political connections on voluntary disclosure varies across different
connection types. Column (1) of Table 8, Panel A shows that, consistent with Faccio (2006), the
effect of political connections on management earnings forecasts is greater among firms whose
political ties are more objectively measured. The difference of the effect of GovParli versus
Considering state ownership. Following Faccio (2006), we capture political connections as the
personal ties between high-level politicians (i.e., a head of state, a minister, or a member of
parliament) and corporate insiders (i.e., large shareholders or top directors). State ownership is
not considered as political ties in this study. Nevertheless, state ownership may constitute
another form of political connections and can be associated with both political connections and a
firm’s voluntary disclosure practice. We collect the information on state ownership based on
Claessens, Djankov, and Lang (2000) and Faccio and Lang (2002) and re-define PC as a dummy
variable, PCplus, equal to one if a firm is politically connected as in Faccio (2006) or has state-
controlled ownership. We rerun our analysis in Table 3 and find consistent results as reported in
column (2) of Table 8, Panel A. In column (3), we include state ownership as an additional
control variable in our regression analysis, and find that our results are qualitatively identical to
those reported in Table 3. 23 In addition, we find that the coefficient on PC is more negative than
the coefficient on State (significant at the 0.01 level), which suggests that the effect of high-level
political connections on voluntary disclosure is greater than the effect of state ownership. In sum,
23
By “qualitatively identical to those reported in Table 3,” we mean that the coefficient on PC in the model as in
column (2) of Table 3 is negative and significant at the 0.10 level or better.
27
Controlling for additional variables. While we control for an extensive set of variables in
equation (2), our results could still be subject to correlated omitted variables bias. Prior studies
find that connected firms have lower financial reporting quality and greater analyst forecast
errors (Chen et al., 2010; Chaney et al., 2011). Thus, we add to the model two additional
variables: (1) a proxy for earnings quality (EarnQuality), calculated as the standard deviation of
five-year performance-matched discretionary current accruals, and (2) a proxy for analyst
forecast accuracy (FError), measured by the absolute value of the difference between the last
consensus analyst forecast prior to the earnings announcement and actual earnings, deflated by
stock price at the beginning of the year. We also include an indicator variable for family
ownership as in Chaney et al. (2011). We then re-estimate equation (2) after adding these three
additional control variables. The number of observations is significantly smaller in these tests
due to data restrictions in calculating additional control variables. Column (4) of Table 8, Panel
A finds that our results are qualitatively identical to those reported in Table 3. Thus, our findings
are robust to including additional controls for earnings quality, analyst forecast accuracy, and
family ownership.
Using a matched sample. The sample distribution in Table 1 shows a much larger number of
control firms than the number of connected firms. While using the full sample helps ensure
adequate sample sizes in our partitioning analyses and additional tests, we test the robustness of
country, and by-industry matched control sample following Chen et al. (2010).24 Column (1) of
Table 8, Panel B finds that our results are qualitatively identical to those reported in Table 3,
24
Among 547 connected firm-years in our sample, we are able to find matched non-connected firms in the same
year, country, and two-digit SIC industry for 520 connected firm-years. For 20 connected firm-years without a
match, we use year-, country- and one-digit-SIC-industry-matching instead.
28
even though the number of observations is substantially reduced. Thus, our findings are robust to
after the end of the reporting period but before the release of the final earnings numbers. These
earnings forecasts are termed as “preliminary earnings announcements” and motivations for
issuing such forecasts are potentially different from those of other forecasts issued earlier in the
year (Hirst et al., 2008). We assess the sensitivity of our results by excluding management
forecasts issued between the fiscal yearend and the date of earnings announcement. Column (2)
of Table 8, Panel B finds that our results are qualitatively identical to those reported in Table 3.
Addressing Capital IQ coverage issues. Radhakrishnan et al. (2012) suggest that Capital IQ
expands its coverage over time and the coverage is likely to be more complete after 2004. To
address the potential concern of incomplete coverage in our sample period, we first re-estimate
equation (2) by limiting our sample to firms with analyst following, assuming that firms that are
followed by analysts are likely to be covered by Capital IQ even in earlier years. The results
reported in column (3) in Table 8, Panel B are qualitatively identical to those reported in Table 3.
In untabulated analysis, we also find that our results are robust to limiting the sample period to
2004. Thus, our findings are robust to addressing the potential coverage issues with Capital IQ.
Dropping influential countries. Table 1 indicates that the UK, Malaysia, Indonesia, Thailand, and
Singapore are the five countries with the largest number of connected firms in our sample. To
assess the sensitivity of our results to the influential countries, we drop these five countries all
together. The results reported in column (4) of Table 8, Panel B are qualitatively identical to those
29
reported in Table 3. In untabulated analysis, we also drop these five countries one at a time and
find consistent results. Thus, our findings are robust to excluding influential countries.
Using alternative clustering schemes. Following Chaney et al. (2011), we cluster the standard
errors in all our regressions at the country-industry level to control for common but unobservable
characteristics shared by observations within the same country-industry group. 25 To assess the
sensitivity of our results, we employ the following alternative clustering schemes: (1) two-way
clustering by firm and year, (2) two-way clustering by firm and country-year. As shown in
columns (5) and (6) of Table 8, Panel B, our results are qualitatively identical to those reported in
Table 3. Thus, our findings are robust to alternative standard error clustering schemes.
Performing analysis by year. Our sample period spans over 2002-2004. We conduct Fama-
MacBeth regressions and find robust results as reported in column (7) of Table 8, Panel B. In
untabulated analysis we also rerun our analysis by year and find qualitatively identical results for
each year as those reported in Table 3. Thus, our findings are robust across all sample years.
VII. CONCLUSIONS
This study examines the effect of political connections on managers’ voluntary disclosure
choices. We find that politically connected firms are associated with a lower level of
management forecasts, and this relation is more pronounced in countries with stronger capital
market development. Using worldwide presidential and legislative elections that result in a
power turnover, we find that connected firms, especially those in countries with more developed
capital markets, increase the frequency of management forecasts subsequent to these elections.
25
This choice is also to ensure that we have a sufficient number of clusters to consistently estimate the standard errors.
We note that our sample includes only three years and 24 countries, so clustering at the year or country level produces
fewer than 40 clusters. As Petersen (2009) indicates, standard errors based on fewer than approximately 40 clusters
suffer from a small sample bias.
30
Our additional analyses reveal that while non-connected firms can lower their cost of debt
through more information disclosure, connected firms enjoy relatively cheap credit regardless of
their information disclosure. We also find that our inferences are robust to using conference calls
Our study is the first to examine whether and how political connections affect voluntary
disclosure. While prior studies document mixed evidence regarding politically connected firms’
preference for information transparency, we help assess this relation in a different setting and
provide further evidence supporting the notion that politically connected firms prefer information
opacity. Our results also suggest that relatively low financing need and weak capital market
incentives, rather than greater political rent seeking and the desire to mask political favors, shape
31
REFERENCES
Adhikari, A., C. Derashid, and H. Zhang. 2006. Public policy, political connections, and
effective tax rates: Longitudinal evidence from Malaysia. Journal of Accounting and
Public Policy 25 (5): 574–595.
Ayer, B., J. Jiang, and P.E. Yeung. 2006. Discretionary accruals and earnings management: An
analysis of pseudo earnings targets. The Accounting Review 81 (3): 617–652.
Beck, T., G. Clarke, A. Groff, P. Keefer, and P. Walsh. 2001. New tools in comparative political
economy: The database of political institutions. World Bank Economic Review 15 (1):
165–176.
Botosan, C. 1997. Disclosure level and the cost of equity capital. The Accounting Review 72 (3):
323–349.
Bushee, B., D. Matsumoto, and G. Miller. 2003. Open versus closed conference calls: the
determinants and effects of broadening access to disclosure. Journal of Accounting and
Economics 34 (1-3): 149–180.
Bushman, R., J. Piotroski, and A. Smith. 2004. What determines corporate transparency? Journal
of Accounting and Research 42 (2): 207–252.
Chaney, P., M. Faccio, and D. Parsley. 2011. The quality of accounting information in politically
connected firms. Journal of Accounting and Economics 51 (1-2): 58–76.
Chen, S., X. Chen, and Q. Cheng. 2008. Do family firms provide more or less voluntary
disclosure? Journal of Accounting Research 46 (3): 499–536.
Chen, C., Y. Ding, and C. Kim. 2010. High-level politically connected firms, corruption, and
analyst forecast accuracy around the world. Journal of International Business Studies 41
(9): 1505–1524.
Claessens, S., S. Djankov, and L. Lang. 2000. The separation of ownership and control in East
Asian corporations. Journal of Financial Economics 58 (1-2): 81–112.
Claessens, S., E. Feijen, and L. Laeven. 2008. Political connections and preferential access to
finance: The role of campaign contributions. Journal of Financial Economics 88 (3): 554–
580.
Coller, M., and T. L. Yohn. 1997. Management forecasts and information asymmetry: An
examination of bid-ask spreads. Journal of Accounting Research 35 (2): 181–191.
Correia, M. 2014. Political connections and SEC enforcement. Journal of Accounting and
Economics 57 (2-3): 241–262.
32
Diamond, D., and R. Verrecchia. 1991. Disclosure, liquidity, and the cost of capital. The Journal
of Finance 46 (4): 1325–1359.
Faccio, M. 2006. Politically connected firms. American Economic Review 96 (1): 369–386.
Faccio, M., and L. Lang. 2002. The ultimate ownership of Western European corporations.
Journal of Financial Economics 65 (3): 365–395.
Faccio, M., R. W. Masulis, and J. McConnell. 2006. Political connections and corporate bailouts.
The Journal of Finance 61 (6): 2597–2635.
Fan, J., T.J. Wong, and T. Zhang. 2007. Politically connected CEOs, corporate governance, and
post-IPO performance of China’s newly partially privatized firms. Journal of Financial
Economics 84 (2): 330–357.
Fisman, R. 2001. Estimating the value of political connections. American Economic Review 91
(4): 1095–1102.
Francis, J., I. Khurana, and R. Pereira. 2005. Disclosure incentives and effects on cost of capital
around the world. The Accounting Review 80 (4): 1125–1162.
Francis, J., R. LaFond, P. Olsson, and K. Schipper. 2005. The market pricing of accruals quality.
Journal of Accounting and Economics 39 (2): 295–327.
Ghoul, S., O. Guedhami, and Y. Kim. 2016. Country-level institutions, firm value, and the role
of corporate social responsibility initiatives. Journal of International Business Studies
(forthcoming).
Guedhami, O., J. Pittman, and W. Saffar. 2014. Auditor choice in politically connected firms.
Journal of Accounting Research 52 (1): 107–162.
Heckman, J. 1979. Sample selection bias as a specification error. Econometrica 47 (1): 153–161.
Hillman, A., G. Keim, and D. Schuler. 2004. Corporate political activity: A review and research
agenda. Journal of Management 30 (6): 837–857.
Hirst, E., L. Koonce, and S. Venkataraman. 2008. Management earnings forecasts: A review and
framework. Accounting Horizons 22 (3): 315–338.
Kato, K., D. Skinner, and M. Kunimura. 2009. Management forecasts in Japan: An empirical
study of forecasts that are effectively mandates. The Accounting Review 84 (5): 1575–1606.
Kaufmann, D., A. Kraay, and M. Mastuzzi. 2007. Governance matters VI: Aggregate and
individual governance indicators, 1996–2006. Washington, DC: The World Bank.
33
Khwaja, A., and A. Mian. 2005. Do lenders favor politically connected firms? Rent-seeking in an
emerging financial market. The Quarterly Journal of Economics 120 (4): 1371–1411.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny. 1998. Law and finance. Journal of
Political Economy 106 (6): 1113–1155.
Lehn, K., and A. Poulsen. 1989. Free cash flow and stockholder gains in going private
transactions. Journal of Finance 44 (3): 771–787.
Leuz, C., and F. Oberholzer-Gee. 2006. Political relationships, global financing and corporate
transparency: Evidence from Indonesia. Journal of Financial Economics 81 (2): 411–439.
Li, X., and H. Yang. 2016. Mandatory financial reporting and voluntary disclosure: The effect of
mandatory IFRS adoption on management forecasts. The Accounting Review 91 (3): 933–
953.
Miller, G., and D. Skinner. 2015. The evolving disclosure landscape: How changes in technology,
the media, and capital markets are affecting disclosure. Journal of Accounting Research 53
(2): 221–239.
Neumann, P. 1994. Bo¨se: Fast alle bestechen (Flaunting the Rules: Almost Everyone). Impulse
(4): 12–16.
Petersen, M. A. 2009. Estimating standard errors in finance panel data sets: Comparing
approaches. Review of Financial Studies 22 (1): 435–480.
Piotroski, J., T.J. Wong, and T. Zhang. 2015. Political incentives to suppress negative
information: Evidence from Chinese listed firms. Journal of Accounting Research 53 (2):
405–459.
Pownall, G., C. Wasley, and G. Waymire. 1993. The stock price effects of alternative types of
management earnings forecasts. The Accounting Review 68 (4): 896–912.
Radhakrishnan, S., A. Tsang, and Y. Yang. 2012. Management forecasts around the world.
Working paper, University of Texas at Dallas and The Chinese University of Hong Kong.
Ramanna, K. and S. Roychowdhury. 2010. Elections and discretionary accruals: Evidence from
2004. Journal of Accounting Research 48 (2): 445–475.
Schuler, D. A., K. Rehbein, and R. D. Cramer. 2002. Pursuing strategic advantage through
political means: A multivariate approach. The Academy of Management Journal 45 (4):
659–672.
Sengupta, P. 1998. Corporate disclosure quality and the cost of debt. The Accounting Review 73
(4): 459–474.
34
Shleifer, A., and R. Vishny. 1994. Politicians and firms. The Quarterly Journal of Economics
109 (4): 995–1025.
Subramanyam, K. R., 1996. The pricing of discretionary accruals. Journal of Accounting and
Economics 22(1): 249–281.
Tucker, J. W. 2010. Selection bias and econometric remedies in accounting and finance research.
Journal of Accounting Literature 29: 31–57.
Watts, R. L., and J. L. Zimmerman, 1983, Agency problems, auditing, and the theory of the firm:
Some evidence. Journal of Law & Economics 26: 613–633.
35
APPENDIX A
Variable Definition
Variables of interest
PC: A dummy variable indicating whether a firm is politically connected based on Faccio (2006). The variable
is equal to one if at least one of a firm’s large shareholders (anyone directly or indirectly
controlling at least 10% of votes) or top directors (CEO, chairman of the board, president,
vice-president, or secretary) is a minister or a head of state, a member of parliament, or is
closely related to a top official, and zero otherwise.
GovParli: A dummy variable equal to one if a firm it is connected to a minister or a head of state, or to
a member of parliament, based on Faccio (2006).
Relation: A dummy variable equal to one if a firm is indirectly connected to a top government official
through close friendship or other relations, based on Faccio (2006).
PCplus: A dummy variable equal to one if a firm is politically connected based on Faccio (2006) or a
firm’s largest shareholder is a government controlling at least 20% of the votes, and zero otherwise.
36
APPENDIX A (continued)
ICRGCorrupt: 26 The International Country Risk Guide's assessment of the corruption in government
based on La Porta et al. (1998). Higher scores indicate "high government officials are likely to
demand special payments" and "illegal payments are generally expected throughout lower levels of
government" in the form of "bribes connected with import and export licenses, exchange controls, tax
assessment, policy protection, or loans."
GermanCorrupt: The German exporters' corruption index developed by Neumann (1994). The index
ranges from zero to five, with higher scores indicating higher levels of corruption.
26
ICRGCorrupt is also used as a country-level control variable in the Heckman first-stage Probit regression.
37
APPENDIX A (continued)
Cost of debt: Interest expense in year t deflated by the average interest bearing obligations outstanding
between year t and t-1.
IntCov: Interest coverage ratio, calculated as operating income divided by interest expense.
Opcycle: The natural logarithm of operating cycle, defined as the sum of days in receivable and days in
inventory.
EarnQuality: Standard deviation of five year performance-matched discretionary current accruals, as in
Chaney et al. (2011).
FError: The absolute value of the difference between the last consensus forecast prior to earnings
announcement and actual earnings, deflated by stock price at the beginning of the year.
Family: A dummy variable equal to one if a firm’s largest shareholder is a family or individual
controlling at least 20% of the votes, and zero otherwise.
State: A dummy variable equal to one if a firm’s largest shareholder is a government controlling at least
20% of the votes, and zero otherwise.
Others
Year FE: Indicator variables for years.
Industry FE: Variables indicating industry membership based on two-digit SIC codes.
Country FE: Indicator variables for countries.
38
TABLE 1
Sample Distribution and Institutional Characteristics
39
TABLE 1 (continued)
Table 1 reports the sample distribution by country and country-level institutional characteristics. Panel A
shows the distribution of politically connected and non-connected control firms across 24 countries in
2002-2004. Panel B shows the institutional characteristics of sample countries.
40
TABLE 2
Descriptive Statistics
Table 2 reports descriptive statistics of the firm-level variables across connected and control
firms. *, **, and *** indicate significance at the 0.10, 0.05, and 0.01 levels, respectively. See
Appendix A for variable definitions.
41
TABLE 3
Political Connections and Management Earnings Forecasts
(1) (2)
1st stage 2nd stage
Pred. sign Dep. Var. =PC Pred. sign Dep. Var. =Freq
Size "+" 0.137*** PC "-" -0.457***
(6.253) (-4.972)
LnAge "+" 0.242** Size "+" 0.055***
(2.340) (9.476)
FreeCash "+" 0.377 ROA "+" 0.097***
(1.404) (2.829)
Capital "+" 0.536*** MTB "+" 0.003**
(6.364) (2.280)
Herf ? -1.522* LEV "-" -0.056*
(-1.809) (-1.770)
IndustryPC "+" 0.140*** EarnVol "+" 0.040*
(5.635) (1.805)
ICRGCorrupt "+" -0.081 RetVol "+" 0.001
(-1.211) (0.121)
XborderRestri "+" 0.294 Nanalyst "+" 0.012***
ct (1.602) (7.074)
RegScore "-" -0.207** BadNews "+" -0.001
(-2.143) (-0.063)
LnGDP ? -0.057 EquityIssue "+" -0.005
(-0.407) (-0.596)
German ? -0.486* Cross "+" -0.020
(-1.874) (-0.862)
French ? 0.126 BigN "+" 0.016
(0.597) (1.498)
Scandinavian ? -0.409 IAS "+" 0.009
(-1.438) (0.508)
Closeheld "-" -0.125***
(-7.279)
Lambda ? 0.193***
(4.699)
No. of obs. 28,553 No. of obs. 28,553
Pseudo R2 0.210 Adj. R2 0.302
Year FE Yes Year FE Yes
Industry FE No Industry FE Yes
Country FE No Country FE Yes
Table 3 reports the Heckman two-stage regression results of management earnings forecasts issued by
politically connected and control firms. The sample consists of 547 connected firm-year observations and
28,006 non-connected control firm-years in 2002-2004. Robust t-statistics in parenthesis are based on
standard errors clustered at the country-industry level. *, **, and *** indicate significance at the 0.10,
0.05, and 0.01 levels, respectively. See Appendix A for variable definitions.
42
TABLE 4
Capital Market Incentive versus Political Rent Seeking
43
TABLE 4 (continued)
44
TABLE 4 (continued)
Table 4 presents the results of testing the channels of the effect of political connections on management
earnings forecast. Panel A tests the capital market incentive explanation and reports the results
conditional on the country median value of MktCap and StockmktEff. Panel B tests the political rent
seeking explanation and reports the results conditional on the country median value of ICRGCorrupt and
GermanCorrupt. Roust t-statistics in parenthesis are based on standard errors clustered at the country-
industry level. *, **, and *** indicate significance at the 0.10, 0.05, and 0.01 levels, respectively. See
Appendix A for variable definitions.
45
TABLE 5
The Effect of Political Realignment on Voluntary Disclosure
46
TABLE 5 (continued)
Panel C: Changes in Management Forecasts in the Two Years Before and After the
Elections with Power Turnover
(1) (2) (3) (4) (5)
Large Small More Less
Dep. Var. = Freq Avg effect MktCap MktCap ICRGCorrupt ICRGCorrupt
PC -0.144*** -0.243*** 0.062 0.034 -0.295***
(-2.640) (-3.646) (1.007) (0.636) (-4.330)
Post 0.006 -0.015 0.103*** -0.014 0.008
(0.467) (-1.150) (3.390) (-1.632) (0.480)
PC x Post 0.180** 0.382*** -0.187** -0.123 0.405***
(2.209) (4.074) (-2.075) (-1.640) (4.317)
Size 0.058*** 0.059*** 0.052*** 0.034*** 0.062***
(8.625) (7.719) (6.846) (6.473) (7.612)
ROA 0.096 0.087 0.228*** 0.103*** 0.092
(1.620) (1.396) (3.070) (2.841) (1.520)
MTB 0.005** 0.004* 0.018*** 0.010*** 0.005**
(2.334) (1.693) (4.520) (4.241) (2.138)
LEV 0.071 0.071 0.102 0.018 0.080
(0.786) (0.702) (1.275) (0.321) (0.793)
EarnVol 0.055 0.056 0.047 0.071 0.062
(1.467) (1.442) (1.047) (1.562) (1.617)
RetVol -0.012 -0.012 -0.011 -0.006 -0.010
(-1.047) (-0.890) (-0.922) (-0.666) (-0.720)
Nanalyst 0.014*** 0.014*** 0.012*** 0.017*** 0.013***
(6.419) (5.871) (3.175) (4.193) (5.644)
BadNews -0.020*** -0.016** -0.024* -0.003 -0.023***
(-2.801) (-2.141) (-1.684) (-0.430) (-2.716)
EquityIssue -0.001 0.008 -0.071*** -0.028** 0.004
(-0.050) (0.576) (-3.123) (-2.401) (0.249)
Cross -0.022 -0.184*** 0.091* 0.150** -0.086
(-0.474) (-2.982) (1.789) (2.573) (-1.589)
BigN 0.062*** 0.068*** 0.032 -0.006 0.075***
(3.223) (2.884) (1.651) (-0.656) (2.908)
IAS 0.116*** -0.028 0.176*** 0.041 0.131***
(3.327) (-0.926) (5.697) (1.581) (3.275)
Closeheld -0.128*** -0.183*** 0.001 -0.059** -0.161***
(-4.459) (-4.988) (0.045) (-2.188) (-4.578)
Year FE No No No No No
Industry FE Yes Yes Yes Yes Yes
Country FE Yes Yes Yes Yes Yes
No. of obs. 22,624 19,196 3,428 4,716 17,908
Adj. R2 0.361 0.363 0.375 0.330 0.327
47
TABLE 5 (continued)
Panel D: Changes in Management Forecasts in the Two Years Before and After the
Elections with Power Turnover, Considering Different Types of Political Connections
(1) (2) (3) (4) (5)
Large Small More Less
Dep. Var. = Freq Avg effect MktCap MktCap ICRGCorrupt ICRGCorrupt
GovParli -0.208*** -0.352*** 0.098 0.119 -0.387***
(-3.194) (-6.408) (1.178) (1.484) (-6.942)
Relation 0.038 0.207 -0.006 -0.067 0.385
(0.370) (1.110) (-0.068) (-1.029) (1.469)
Post 0.006 -0.015 0.103*** -0.014 0.008
(0.447) (-1.164) (3.383) (-1.647) (0.463)
GovParli x Post 0.310*** 0.522*** -0.153 -0.154 0.514***
(3.457) (6.433) (-1.379) (-1.587) (6.575)
Relation x Post -0.196* -0.218 -0.248* -0.086 -0.417
(-1.834) (-1.163) (-1.842) (-0.835) (-1.469)
Post 0.006 -0.015 0.103*** -0.014 0.008
(0.447) (-1.164) (3.383) (-1.647) (0.463)
Size 0.058*** 0.059*** 0.052*** 0.035*** 0.062***
(8.618) (7.645) (6.898) (6.690) (7.539)
ROA 0.096 0.087 0.228*** 0.104*** 0.093
(1.619) (1.398) (3.071) (2.876) (1.531)
MTB 0.005** 0.004 0.018*** 0.010*** 0.005**
(2.309) (1.625) (4.535) (4.388) (2.043)
LEV 0.072 0.071 0.106 0.020 0.080
(0.789) (0.704) (1.341) (0.361) (0.798)
EarnVol 0.055 0.056 0.047 0.074 0.061
(1.470) (1.423) (1.062) (1.610) (1.601)
RetVol -0.012 -0.011 -0.012 -0.006 -0.010
(-1.058) (-0.856) (-0.955) (-0.730) (-0.698)
Nanalyst 0.014*** 0.014*** 0.012*** 0.017*** 0.013***
(6.396) (5.846) (3.179) (4.207) (5.638)
BadNews -0.020*** -0.016** -0.025* -0.003 -0.022***
(-2.778) (-2.086) (-1.727) (-0.473) (-2.640)
EquityIssue -0.000 0.008 -0.070*** -0.027** 0.004
(-0.039) (0.583) (-3.085) (-2.351) (0.262)
Cross -0.021 -0.170*** 0.083 0.133** -0.078
(-0.459) (-2.806) (1.642) (2.372) (-1.455)
BigN 0.062*** 0.068*** 0.033* -0.007 0.076***
(3.217) (2.902) (1.709) (-0.677) (2.922)
IAS 0.118*** -0.028 0.176*** 0.040 0.133***
(3.367) (-0.918) (5.663) (1.536) (3.329)
Closeheld -0.130*** -0.184*** -0.000 -0.059** -0.162***
(-4.522) (-5.027) (-0.011) (-2.211) (-4.613)
Year FE No No No No No
Industry FE Yes Yes Yes Yes Yes
Country FE Yes Yes Yes Yes Yes
No. of obs. 22,624 19,196 3,428 4,716 17,908
Adj. R2 0.361 0.363 0.376 0.332 0.328
48
TABLE 5 (continued)
Table 5 presents the analysis of changes in management earnings forecasts issued by politically connected
firms and control firms in countries that experience elections resulting in power turnovers. Panel A
describes the political systems and the elections, Panel B reports the sample distribution, Panels C and D
report the regression analysis, and Panel E presents the changes in forecast properties among connected
firms following the elections. Robust t-statistics in parenthesis are based on standard errors clustered at
the country-industry level. *, **, and *** indicate significance at the 0.10, 0.05, and 0.01 levels,
respectively. See Appendix A for variable definitions.
49
TABLE 6
Management Forecasts and Cost of Debt
Dep. Var. = Cost of debt (1) (2)
PC (β1) -0.030*** 0.042
(-5.400) (1.124)
Freq (β2) -0.013*** -0.007*
(-3.901) (-1.908)
PC x Freq (β3) 0.018 0.019
(1.078) (0.998)
Size -0.005***
(-3.150)
ROA -0.101***
(-3.955)
MTB 0.001
(1.482)
LEV -0.157***
(-10.173)
EarnVol 0.033**
(2.406)
RetVol 0.004
(1.544)
Nanalyst -0.000
(-0.993)
BadNews 0.003
(1.131)
EquityIssue 0.017**
(2.469)
Cross 0.010
(0.913)
BigN 0.002
(0.287)
IAS 0.011
(0.708)
Closeheld -0.010
(-1.272)
IntCov 0.000***
(2.789)
Opcycle -0.007**
(-2.054)
Lambda -0.023
(-1.287)
Year, industry, country FE No Yes
Test of β2+β3 p=0.763 p=0.537
No. of obs. 15,976 15,976
Adj. R2 0.002 0.080
Table 6 presents the regression results of the effect of issuing management earnings forecasts on the cost
of debt for politically connected and control firms. Robust t-statistics in parenthesis are based on standard
errors clustered at the country-industry level. *, **, and *** indicate significance at the 0.10, 0.05, and
0.01 levels, respectively. See Appendix A for variable definitions.
50
TABLE 7
Using Conference Calls as an Alternative Measure of Voluntary Disclosure
51
TABLE 7 (continued)
Table 7 presents the results of using conference calls as an alternative measure of voluntary disclosure.
Panel A reports the results for the full sample and the results conditional on capital market development.
Panel B reports the results conditional on corruption. Robust t-statistics in parenthesis are based on
standard errors clustered at the country-industry level. *, **, and *** indicate significance at the 0.10,
0.05, and 0.01 levels, respectively. See Appendix A for variable definitions.
52
TABLE 8
Sensitivity Tests
Table 8 presents the results of sensitivity tests. Panel A reports the results using alternative definitions of
political connections and additional control variables. Panel B reports the results using alternative samples and
clustering methods. Robust t-statistics in parenthesis are based on standard errors clustered at the country-
industry level except in models 5 and 6 in Panel B. *, **, and *** indicate significance at the 0.10, 0.05, and
0.01 levels, respectively. See Appendix A for variable definitions.
53