Critical Review of Research Journals

The quality of accounting information in politically connected

firms
(Paul K. Chaney,Mara Faccio,David Parsley)

Management Auditing Courses

Lecturer: Mahameru Rosy Rochmatullah

Arranged by :

Renada Puja Islam Mahavira

B200164022

UNIVERSITAS MUHAMMADIYAH SURAKARTA

2019
 FOREWORD

Praise the presence of Allah SWT , the Most Merciful, Most Merciful, because thanks
to His mercy and guidance, the author can compile and present an inn assignment with the
title "The quality of accounting information in politically connected firms".

This paper was made to fulfill the “auditing management" task of the odd semester of
2019.

The author realizes that in the preparation of this task there are still many
shortcomings and far from perfection. Therefore, the author expects criticism and suggestions
that can improve this task and can be a reference in compiling subsequent tasks.

The author also apologizes if in writing this task there are typos and errors that
confuse the reader in understanding the intent of the author.
CRITICAL REVIEW JOURNAL

INTRODUCTION

In this paper we investigate whether earnings quality varies systematically with

political connections in a broad sample of politically connected countries and companies.
Overall, our results reveal that there is a political connection associated with lower quality of
accounting earnings. We document that political connections have enhanced explanatory
power outside the country, regulations, and special ownership characteristics of the company.
Ex-ante, one can argue that because connected companies are subject to extensive control and
monitoring (including if observed by the media), political connections will, on in fact, it is
associated with better income quality. However, this is not a case. Based on the results in
previous studies, three explanations are consistent with our findings that the quality of labs in
politically connected companies is worse than the quality of income from non-connected
companies.

First, politically connected companies usually benefit from their connections 1 above
and above the payments they make, 2 insiders can hide, obscure, or at least attempt to delay
reporting benefits received in order to mislead investors to benefit at cost they (for example,
Schipper, 1989 or Leuz et al., 2003). Second, to the extent that politicians provide protection
to related companies so that low-quality accounting information is not punished, connected
companies may not be too concerned with the quality of the information they disclose, and
invest less time to accurately describe their accruals. In this case, the quality of information
will be low due to lack of attention on the part of the company manager. This represents a
better interpretation of poor quality accruals. Third, it might just be the case that companies
with poor income quality are more likely to establish political connections. In all cases,
political connections will be associated with poor quality of information, as we find it

We run two sets of tests to try to distinguish between possible explanations. First, for
the company sub-sample for which the establishment date of a connection can be determined,
we investigate whether poor accrual quality has an impact on the likelihood of the company
making a connection in a particular year. Second, we assess the need for connected
companies to make the investments needed to provide good quality accounting information.
This second test exploits the previous evidence in Francis et al. (2005) who found that, for
US companies, poor quality of income was associated with higher debt costs (as well as
higher equity costs). We argue that this result may not apply to politically connected
companies.

POLITICAL

The empirical evidence that we provide in this study comes from two primary
databases. First, we employ large-level enterprise datasets on corporate political connections
developed by Faccio (2006). Second, using basic accounting data available in Worldscope,
we make several measures of the quality of accounting earnings based on discretionary
variability in earnings. As discussed above, we also examine whether the effects of political
connections on accounting quality information depend on the characteristics of the company's
ownership structure (for example, the presence of large shareholders or family control).

A company is classified as politically connected if, at some point between 1997 and
2001, at least one of the large companies holds shares (anyone who directly or indirectly
controls at least 10% of the vote) or chief director (CEO, board chairman, president, vice
president, or secretary) is a member of parliament, minister or head of state, or is closely
related to the political party or party. This close relationship includes cases of friendship, past
politics (for example, heads of state or ministers) or company positions, foreign politicians,
and cases of famous relationships with political parties, as discussed further below.

Connections with government ministers include cases where the politician himself is a
large shareholder or director, as well as cases where a close relative of the politician (for
example, son or daughter) holds the position. For example, Arnoldo Mondadori Editore is
among our connected companies because it is controlled by Silvio Berlusconi, an Italian
Prime Minister. The Berhad Logistics Consortium is included in the sample because its
chairman, Mirzan bin Mahathir, is the son of Malaysian Prime Minister Mahathir bin
Mohamad.

THE QUALITY OF ACCOUNTING EARNINGS DATA

Accounting information users are generally interested in assessing current

performance and estimating future performance, and there is much debate about how well
various accounting measures reflect this goal. Some of the company's transactions only
require mechanical application of accounting rules while other types depend on the judgment
of the company's managers and accountants. Management's judgment regarding earnings
determination is often associated with discretionary accruals. Managers can use this
discretionary accrual option in an opportunistic way (perhaps to increase their own
compensation or hide poor performance) or they can use this policy to increase the value of
earnings information (possibly to communicate to investors the company's long-term
performance). However, discretionary accruals are often used as a measure of earnings
quality (for example, Dechow and Dichev, 2002; Francis et al., 2004)

Reported earnings are considered a key indicator of information quality (for example,
Dechow, 1994; Dechow et al., 1998). 3 Because company earnings differ from cash provided
from operations by the amount of accruals reported, standard practice is to focus on the
absolute magnitude and / or variability of accruals to assess earnings quality. Specifically,
because accruals include discretionary and non-discretionary components, and because
discretionary accruals are believed to better reflect managerial judgment, most earnings
quality research focuses on discretionary accruals.

DESCRIPTIVE STATISTICS

Matching accounting data from Worldscope with data on political connections from
Faccio (2006) described in Part 1, and requiring that at least one company is connected in
each country, our final sample (using 5 years of accrual quality measures) includes 4954
companies, of which 209 are connected to politicians, from 19 countries. Examples of using a
10-year quality accrual measure include 4308 companies, of which 168 are politically
connected. Table 1 presents a snapshot of a number of summary statistics at the country level.
The overall impression is that there is wide variation in samples sampled in all country-level
measures

Comparisons across samples

Presenting associations from our two main measures of quality of accounting

information vis-ea-vis the connection variables, and the ownership structure of the
company. Ceteris paribus, higher values of this variable indicate lower income quality.
In the first set of two columns we focus on (5 years) the standard deviation of
discretionary accruals calculated from Eq. (1) In the second column set, we display
the same information for standard deviation (10 years) discretionary accruals. Each
set of three rows presents statistics based on company characteristics. For example,
there are 209 companies connected to us that can calculate the standard deviation of 5
years, and the average value for connected companies is 5.751, and 5.206 for other
non-connected companies. The difference between these averages is statistically
significant. We note the important warning that functions to us in detecting significant
differences in accrual quality between connected and unconnected companies in
Indonesiates. This simple sample comparison: as research has shown that connected
companies tend to be relatively large, and large companies tend to have better accrual
quality, without control for company size our conclusions tend to be biased. This is
probably behind the insignificant difference in using the 10-year measure.

Control variables

Before reporting our regression results, we describe a number of company and

country characteristics that we use for control in our regression analysis. Their
inclusion was motivated by previous studies that have found them to be related to the
quality of accounting information at the company or at the country level (Doidge et
al., 2007; Fan and Wong, 2002; Hribar and Nichols, 2007; Leuz et al., 2003; Leuz,
2006) . We introduce controls for measures of voting ownership held by the largest
shareholder (Control). Data ownership was taken from Claessens et al. (2000) and
Faccio and Lang (2002), the stock exchange website or its supervisory authority,
Worldscope, and Extel. Ownership related data are generally recorded at the end of
1997. Controls are built according to La Porta et al. (1999), which argues that an
investor can gain control in a company by directly owning controlling shares, or
indirectly through ownership of shares in another company. In the first case, an
investor's share of control rights will correspond to the faction of the vote that he has
the right to disclose because the shares he owns. In the second case, the investor's
share of control rights is measured by the weakest control relationship along the
spiramide. We also define Family as a dummy variable equal to 1 if the largest
shareholder is a family or individual controls at least 20% of the votes, and 0
otherwise.

Regression analysis

Present the results of OLS regression where the dependent variable is the
standard deviation variable calculated in Eq. (1) calculated for a period of 5 years
(2001-2005). In all regressions, the dependent variable is multiplied by 100. The
independent variable is the size of the connection, the ownership variable, and other
 companies and country-level attributes. In the initial cross-sectional test, we classified
companies as connected if political connections were recorded at any time between
1997 and 2001. Furthermore (in Section 4.4), we examined the subsamples of
companies that were targeted we could identify the date of connection establishment.
There we answer the question whether connected companies exhibit different income
qualities (for example, worse) (relative to their peers) before the formation of their
reconnection politics.

ROBUSTNESS TESTS

1. Excluding individual countries and differences between countries. To provide

assurance that our results are not driven by a particular country, we repeat our estimation
repeatedly (1) in Tables 3 and 4, eliminating different countries each iteration. These results
are not reported in the table for space considerations, but can be summarized as follows.
When the 5-year standard deviation of discretionary accruals is used as the dependent
variable, the coefficient on connections is always positive, ranging from 0.407 to 0.792. The
coefficient is significant in 17 of the 19 regressions, with p values of 0.052 or lower; only
when we excluded Japan or England, did the coefficients lose statistical significance (p-
values 0.146 and 0.171, respectively). When a 10-year measure was used, the coefficients on
connections ranged from 0.140 to 0.971. The coefficient is significant in 18 of 19 regressions,
with p values of 0.053 or lower; the coefficient only loses statistical significance when we
exclude Japan.

2. Transforming dependent variables and alternative estimation methods,With

construction, the dependent variables in our regression model are positive. Because of this
deduction, interference conditions may not be distributed normally. One possible solution to
this problem is to use logistic transformation, where the new dependent variable is calculated
in this case if i meets ÀNoz i oN, and thus the cutting problem is avoided. We can therefore
use these specifications to assess the robustness of our previous results in Table 3. For spatial
reasons, the results of this resilience are not bible. When possible y to (alternatively) assume
a value of 0.001, 0.01, or 1 and re-run the regression. We find that the coefficient of the
Connected range is between 0.109 and 0.112, and is statistically significant at the
conventional level, with a p-value less than 0.10
 3. An alternative measure of earnings quality,We present the results of thirteen
different estimates. In each of the first six regressions, consider different earnings quality
measures that have been suggested by other studies. In regression (7) - (12) the dependent
variable is the median absolute value of discretionary accruals, rather than standard deviation,
during 2001-2005. In regression (13) the dependent variable is the average ratio of the
absolute value of the accrual to the absolute value of the cash flow from operations . All
models below are estimated by industry including business cycle control. The steps are
defined as follows: PADCA is the size of the portfolio adjusted for discretionary accrual
performance. This is discretionary accrual (as measured in Eq. (1), although it does not
include the terms ROA in equations (2) and (3)) for companies minus the discretionary
medianrual industry based on the decile ranking of ROA for each industry determined in the
previous year

4. Are companies with poor accrual quality more likely to establish political
connections?

As shown in the introduction, several hypotheses that might be consistent with our
general findings that connected firms have worse accrual quality. Specifically, it might just be
the case that companies with poor earnings quality are more likely to establish political
connections. If that is the case, we must find poor accrual quality (ex-ante) associated with a
higher likelihood that a connection is made in a particular year

The regression results in Table 6 show that accrual quality does not have a significant
impact on the likelihood of companies establishing connections. 13 The coefficient of the
accrual quality variable is actually negative, which indicates that, if any, companies with
(previously) worse accrual quality are less likely to establish new political connections in a
given year. The results confirm that the size of the company is an important determinant of
the possibility of establishing connections; as well as (at least in one of its specifications)
anti-director rights, corruption, operating cycles, market-to-books, leverage, and company
headquarters locations in the nation's capital.

WHY DO CONNECTED COMPANIES NOT CARE ABOUT THE CONSEQUENCES OF

POOR EARNINGS QUALITY?

We have shown that, on average, the accrual quality of connected companies is worse
than the accrual quality of unconnected companies. From an empirical point of view, a
number of studies have shown that poor accrual quality results in a number of negative
consequences at the firm level, including higher capital costs (Francis et al., 2005), or higher
likelihood of claims. Thus, the question becomes why connected companies do not seem to
care about the consequences. The only possibility is that their political ties might reduce or
even eliminate such effects. So, for example, it's possible that lenders from connected
companies give them relatively cheap capital, regardless of the opacity / quality of their
accountant information.

1. The average debt cost is realized

We use two approaches to deduce debt costs. In this subsection, we follow

Francis et al. (2005) and Liu and Wysocki (2007), and calculate the cost of debt as a
ratio of the company's interest expense in year t (in our case 2005) (WC01251) during
the average interest bearing obligations outstanding between quarter 4 years tÀ1 and
quarter 4 years t (WC03255A). The average interest bearing liabilities are calculated
using temporary data, if available. 15 When calculating the outstanding average
interest rate, we need a minimum of two observations in each given year. This gives
us the cost of debt realized in the company's local currency. To make these rates
comparable in various countries, we convert them in US dollars using covered interest
parity.

2. Issuance of bonds

In this section, we focus on the issuance of public debt and measure the cost of
debt in terms of the spread between the yield to maturity for a particular issuance and
the yield to maturity of treasury bonds with comparable maturity issued by the
Indonesian government of the country where the company is headquartered. Focusing
on yield until maturity has several advantages. First, it is a direct measure of the cost
of debt. Second, because we only focus on public debt, it is unlikely that differences
in personal information flows between lenders and borrowers will have a different
impact on the quality of accruals on the cost of debt for connected vs. unconnected
companies. However, dependence on public debt also has several limitations. First,
not all companies issue bonds. Issuer bonds actually tend to be large and mature
companies. In addition, unconnected companies are more likely to have public debt
than connected companies, which is consistent with the argument that connected
companies avoid public securities that require more transparency to external
 investors, as suggested earlier by Leuz and Oberholzer-Gee (2006). Apart from these
limitations, it is important to provide validation of previous results using different
metrics.

CONCLUSIONS

This study documents that the quality of accounting information reported

systematically is worse for companies with political connections than companies without
such connections. This conclusion is based on an analysis of accounting data for more than
4,500 companies in 19 countries. Political connections seem to be important predictors of
accounting quality even after controlling for some commonly used country-level variables
such as the overall level of corruption, or shareholder rights indicators, and for company-
specific factors (ownership structure, size, growth, leverage, market-to-book ratio, or cash
flow volatility or sales growth). While connections are associated with poor quality post-
posts, previous quality accruals do not explain the tendency of companies to establish
connections within a certain period. So, we rule out that it's as simple as that companies with
poor accrual quality ultimately build political connections.