gia2016-inw-financial reporting around the...
TRANSCRIPT
1
Financial Reporting Differences Around the World: What Matters?
Helena Isidro*
Instituto Universitário de Lisboa (ISCTE-IUL), Lisbon, Portugal
Dhananjay Nanda†
School of Business Administration, University of Miami
Peter Wysocki†
School of Business Administration, University of Miami
July 2016
Abstract
The international financial reporting literature identifies a multitude of country attributes that each appear to
explain financial reporting differences around the world. We first show that a single underlying factor
explains across-country variation in 6 reporting quality measures used in the international literature. We then
examine 72 country attributes and show that they are highly correlated and that 4 underlying factors explain
most of the variation in these attributes across countries. Furthermore, individual country attributes provide
essentially no incremental explanatory power for international reporting diversity over these 4 factors, which
collectively explain over 70% of the variation in reporting differences. Our findings highlight the very high
causal density of country attributes and thus the difficulty in attributing international reporting diversity to
specific institutions and policies. We conclude with a discussion of possible future directions for research on
financial reporting around the world.
Keywords: IFRS, International accounting, Complementarities, Correlation, Factor analysis,
Financial reporting, Accounting quality, Multiple testing
* Helena Isidro: ISCTE-IUL Instituto Universitário de Lisboa, Avenida das Forças Armadas, 1649-026, Lisboa, Portugal. Tel:+351
217 903 480. Email: [email protected]
† Dhananjay Nanda: School of Business Administration, University of Miami, Coral Gables, FL 33146. Email:
† Peter Wysocki: School of Business Administration, University of Miami, Coral Gables, FL 33146. Email:
2
1.1.1.1. Introduction
The international accounting literature has associated a multitude of country-level attributes
with cross-country variation in several financial reporting outcomes around the world. These
country attributes include geographic features (e.g. country latitude), legal institutions (e.g. legal
origin), religious affiliation (e.g. percentage catholic, religiosity), cultural development (e.g.
masculinity, societal trust) and economic outcomes (e.g. per capita GDP, market capitalization,
stock market participation).1 Can all of these attributes individually affect country-level reporting
outcomes? Can existing empirical methods and data be used to distinguish between the many
proposed and competing ‘theories’ of the determinants of international reporting diversity? What
truly determines or influences the quality of reported financial numbers across countries?
This study takes a comprehensive look at the existing panapoly of country-level attributes
proposed in prior empirical studies that appear to be individually associated with variation in
several financial reporting quality measures around the world.2 First, we define and summarize the
empirical measures of country-level financial reporting quality previously used in the literature. We
show that, while there are numerous measures, they collectively seem to capture a single underlying
(latent) construct of financial reporting quality. We then survey more than a hundred empirical
studies published in the last two decades comparing country attributes and identify 72 different
variables that have been used to measure differences in economic, cultural, institutional and societal
1 The extant literature often uses the term “institutions” (see, for example, Leuz and Wysocki [2016]) to describe
possible country determinants of financial reporting and disclosure. We use the broader term “country attributes”
thoroughout our paper to capture a broader set of empirically-observable country features including institutions, but
also features such as exogenous physical geography and endogenous economic outcomes. 2 Examples of these variables include the quality of reporting standards (e.g. Christensen et al. [2015], Core et al.
[2014], Daske et al. [2013], Armstrong et al. [2010], Barth et al. [2008], Daske et al. [2008]), enforcement regulation
(e.g. Christensen et al. [2013], Brown et al. [2014], legal rules (e.g. Gupta et al. [2008], Hail and Leuz [2006]), investor
protection (e.g. DeFond et al. [2007], Leuz et al. [2003]); economic development (e.g. Chen et al. [2015]); political
institutions (e.g. Li et al. [2016], Batta et al. [2015], Boutchkova et al. [2012], Bushman and Piotroski [2006], Riahi-
Belkaoui [2004a]), and social values (Pevnzer et al. [2015], Nanda and Wysocki [2015]; McGuire et al. [2012],
Kanagaretnam et al. [2011]).
3
development across countries. Using data from 35 countries, we first empirically explore which, if
any, of the variables are incrementally associated with financial reporting quality variation beyond
an underlying core set of latent factors derived from the remaining variables.3 Based on this
analysis, we conclude that there are fundamental barriers for researchers to empirically isolate the
effect of individual country attributes in explaining reporting and disclosure diversity around the
world because (a) the country-level measures are highly correlated suggesting strong and arguably
inseparable interdependencies among country-level institutions and features, and (b) the relatively
few country observations in the world limits empiricists’ ability to statistically isolate the effects of
individual country attributes. While accounting researchers have suspected the existence of these
problems, this is the first study to systematically document the fundamental inferential problems
facing comparative country analyses that arise from: (i) the plethora of candidate “theories” and
associated country variables that potentially explain global reporting and disclosure outcomes, (ii)
the static or generally slow moving nature of country attributes, and (iii) as listed above, the high
correlation among these possible explanatory variables (i.e. high causal density) and the limited
number of country observations in the world. In summary, we lack of degrees of freedom in current
cross-country studies with too many correlated candidate theories/attributes/variables and there are
too few empirical observations to parse the possible competing effects. We further highlight that
this is not really a problem of incorrect empirical tools (i.e., multiple regression vs structural
modelling vs qualitive comparative analysis), but an issue of statistical inference based on too little
data used to identify an effect from too many interwined country-level attributes.
It should be noted that the ‘zoo’ of possible global reporting and disclosure determinants
and outcomes increases each time a seemingly ‘new’ country variable is reported to be significantly
3 Restricting the sample to 35 countries is necessitated because we require data for all financial reporting characteristics
and for most of the previously identified country attributes. Our sample largely overlaps with the countries used in
many prior international accounting studies.
4
associated with a financial reporting diversity measure across countries, or a ‘new’ financial
reporting quality variable is found to co-move with previously identified country attributes.
Unfortunately, each new variable is either empirically tested in isolation, or at best, in combination
with a few other country variables, making it difficult to evaluate whether a proposed variable
incrementally explains reporting outcomes or is simply associated with the set of previously
correlated attributes. To help guide future discoveries of “seemingly” new country-level attributes
and their relation with “seemingly” new international reporting diversity measures, we outline a
benchmark correlation analysis where any newly-proposed variable should provide incremental
explanatory power for reporting outcomes beyond 4 core empirically derived latent factors that
capture the joint explanatory power of previously-documented country attributes for reporting
outcomes around the world. We show that the 4 latent factors collectively explain a substantial
amount of the observed cross-country variation in financial reporting outcomes. Further, our
empirical analyses suggests that few, if any, existing individual country attributes provide any
significant explanatory power for reporting outcomes beyond these 4 underlying factors.
An additional issue that we highlight in our examination is that inferences should be based
on a multiple testing framework rather than on a single test perspective since many international
research studies attempt to explain the same international cross-section of reporting and disclosure
differences. A multiple testing framework corrects for the bias towards false discoveries that result
from simultaneously testing multiple hypotheses using the same data. The upshot being: to evaluate
the significance of a new country attribute in explaining financial reporting diversity, a researcher
must account for tests that document correlations between financial reporting outcomes and other
country-level atrributes.4
4 Similar problems exist in the empirical asset pricing literature in finance where a multitude of studies have proposed
and tested hundreds of isolated “return predicting signals” without acknowledging the existence and possible
5
We make two key contributions highlighting both the high correlation among reporting
quality measures as well as the high causal density of country attributes that have been used in past
research to explain the cross-country variation in these measures. We first show that commonly
used country-level reporting quality measures are highly correlated and a single factor explains
most of the cross-country variation in these measures. The six measures that we choose are 1)
abnormal returns around earnings announcement (Nguyen and Truong [2013]), 2) abnormal volume
around earnings announcement (Nguyen and Truong [2013]), 3) return synchronicity (Nguyen and
Truong [2013]), 4) reporting transparency (Leuz et. al. [2003]), 5) disclosure quality (Bushman, et.
al. [2004]), and 6) asymmetric timeliness (Bushman and Piotroski [2006]). A single principal
component underlying the six variables explains ninety percent of the cross-country variation in our
35-country sample.
Our second contribution is that we systematically tabulate the very high correlations among
many country-level variables used in international studies. These high correlations are indicative of
strong and significant comovement among country attributes. On average each variable has a
correlation of more than 0.5 with five or six other measures, and a correlation of more than 0.7 with
three or four of them. Further, the ability of country-level measures to explain other country-level
measures is quite high. For example, block premium, regulatory quality, legal origin, and latitude
(a proxy for economic development from La Porta et al. [1999]) explain 74% of private control of
self-dealing, a measure extensively used to represent the level of shareholder protection in a
country. Similarly, public enforcement of securities regulation, latitude, religious fractionalization,
and power distance explain about 65% of number of analysts in a country, a measure often used to
assess a country’s public firms’ informational environment quality (Byard et al. [2011]). To
examine whether the high correlations between these variables capture a set of underlying latent
correlations among these signals. Harvey et al. [2016], Green et al. [2014] and Green et al. [2013] highlight these
problems in the empirical asset pricing literature and provide evidence on the robustness of the prior findings.
6
factors in an economically meaningful way, we perform a factor analysis on all 72 variables. Our
factor analysis attempts to resolve the dimensionality problem that arises in international studies
because the number of possible explanatory variables far exceeds the number of observations
(countries) typically investigated. Based on their statistical significance (eigenvalues) we retain four
latent factors from our factor analysis. The four factors that we retain explain about 60% of the
variation in country attributes and the first two factors alone explain more than a third of the
variation.
Next, we analyze the relation between our identified latent factors based on country
attributes and the single “reporting quality” latent factor derived from six financial reporting
measures. We assess the four latent factors’ ability to explain variation in the latent financial
reporting quality factor across 35 countries. We find that the four factors representing country-level
characteristics explain a significant portion of the international variation in financial reporting
outcomes. A regression of the reporting quality factor on the four latent country-level factors has an
R2 of almost 70%. In particular, the first two latent factors explain more than half of the variation in
reporting outcomes across countries. Additionally, most of the 72 country-level variables in the
literature are significantly associated with variation in the financial reporting factor when tested
individually and in isolation. However, the explanatory power of each of these variables essentially
evaporates when one controls for the 4 latent factors, computed while excluding the variable
examined, that aggregate other country-level characteristics along with a multiple testing
adjustment to test statistics.5 Our findings highlight the difficulty in inferring empirically whether
differences in any single country characteristic affects observed differences in financial reporting
outcomes across countries. Collectively, our empirical evidence suggests that financial reporting
5 As described in section 3, we use both the Bonferroni and Holm adjustments to account for multiple testing issues.
7
outcomes are likely influenced by an inextricably intertwined set of interrelated country-level
features.
In the concluding section of this study, we discuss how this study’s findings can help push
the boundaries of future international research examining the association between country-level
features and economic outcomes. Overall, our analyses show that the paucity of country-level
observations coupled with the very large number of previously-identified and highly-related
country attributes sets a high bar for researchers to document novel associations between individual
country attributes and reporting and other outcomes. However, on the upside, our evidence also
shows that country attributes collectively (as captured by our four factors) explain a substantial
portion of the cross-country variation in reporting quality outcomes. Thus, we first suggest that
researchers who are investigating a ‘new’ country attribute should benchmark the attribute against
the four empirical factors presented in this paper. This benchmarking exercise can help provide
insights into both the possible incremental explanatory power (for various accounting or other
economic outcomes) of the ‘new’ country attribute and also how the attribute ‘fits’ with other
previously-identified groups of country attributes and institutions. Second, we suggest that future
research should more explicitly acknowledge the interdependencies among the many attributes and
institutions that exist in countries and regions. So, rather than focusing on an isolated country
attribute, studies would be better served by acknowledging the portfolios of correlated country
attributes and policies that work together as a ‘package’. Third, the evidence in this paper highlights
the opportunities to better understand how and why portfolios of country attributes and institutions
work together to influence financial reporting and other economic outcomes.
The remainder of our paper is organized as follows. The next section describes the data, and
our examined country characteristics and financial reporting variables. Section 3 discusses our
empirical methods and results. Section 4 explores the potential of alternative empirical methods’
8
ability to overcome the limitations of traditional regression analyses. Section 5 presents our out-of-
sample case analyses to highlight the generalizability of our findings. Section 6 presents our
conclusions and suggestions regarding future research directions.
2. Financial Reporting and Country Attributes
2.1. Financial reporting measures
The international financial reporting literature has used a variety of measures to examine
reporting quality diffrences across countries. Consistent with this literature, we use a
multidimensional approach to study firm financial reporting outcomes at the country level. We
examine the following six measures at the country-level that reflect financial reporting attributes of
public firms domiciled in those countries. Reporting transparency and disclosure quality are
accounting-based measures that reflect firms’ reporting and disclosure choices. Specifically,
reporting transparency is the negative of the opacity score of Leuz [2010], which is an update to
the Leuz et al. [2003] earnings management score and only includes data from reported financial
statements. Disclosure quality is the disclosure index created by the Center for Financial Analysis
and Research based on the inclusion or exclusion of 90 items in firms’ annual reports (reported in
Bushman et al. [2004]). Abnormal return, abnormal volume and return synchronicity are market-
based measures that capture investors’ reaction to the release of periodic financial reporting
information (reported in Nguyen and Truong [2013]). Abnormal return and abnormal volume are
market adjusted returns and trading volume measured over the three-day window around firms’
earnings announcements, respectively. These variables reflect how investors respond to the public
dissemination of accounting earnings information. Similarly, Return synchronicity is the weighted
average R-squared of regressions of firm returns around earnings announcements on market returns,
9
multiplied by minus one. It measures the extent that earnings information is impounded in stock
prices over the long-term. Asymmetric timeliness is reported in Bushman and Piotroski [2006] and
measures the relation between accounting information and market information. It is intended to
reflect managers’ accounting choices in recognizing bad news in earnings as seen by market
participants. We redefine the measures so that higher values indicate higher reporting quality and
standardized them to a zero mean and a standard deviation of one. Appendix 1 - Panel A provides a
brief description of the reporting quality variables, data sources, and examples of studies that use
these measures in an international setting.
2.2. Country attributes
We collect a comprehensive set of variables used in the financial reporting literature over
the last two decades, that represent multiple country characteristics for a sample of 35 countries
(see Table 1 for the list of countries). We collect the relevant variables from publicly available
sources for periods between 1995 and 2012 (see Appendix 1 – Panel B for definitions of data
sources of the country variables).6 We select variables that have been extensively used to explain
country-level differences in geographic, institutional, cultural and economic features, and that are
publicly available. These variables represent geographic features (e.g. latitude), economic
development (e.g. GDP per capita), capital market development (e.g. market capitalization to
GDP), legal institutions (e.g. law and order), political systems (e.g. democracy), tax mechanisms
(e.g. assessment of tax evasion), securities regulation (e.g. securities regulation disclosure
requirements), sources of financing (e.g. bank money in private sector to GDP), reporting and
auditing enforcement (e.g. enforcement of accounting standards), investor and creditor rights (e.g.
anti-director rights), foreign investment (e.g. foreign institutional holdings), analyst activity (e.g.
6 In most cases, the data sources provide only time invariant values for certain country variables. In cases where time
series data are available, we use the average time-series value of the variable.
10
number of analysts), audit quality (e.g. Big4 market share), dissemination of information (e.g.
media), cultural values (e.g. individualism), social attitudes (e.g. trust), religious practices (e.g.
catholic), and linguistic properties (e.g. English proficiency). To deal with differences in scale and
missing values we standardize all variables and replace missing observations with the sample
mean.7 Appendix 1 - Panel A describes the 72 country-level measures included in our study and
their corresponding data sources. We also cite studies that use each country-level variable to
explain international variation in financial reporting quality measures. To choose the 35 countries
we considered countries that are economically and socially important and for which data is
available from public sources for the country-level variables.8
3. Empirical analysis
3.1. Determining the dimensionality of international financial reporting quality
The summary statistics for the individual financial reporting measures (unstandardized) are
reported in Panel A of Table 1. We observe substantial cross-country variation in financial
reporting measures, but we also note that these measures are highly correlated and this suggests that
they may represent common constructs (Panel B). To identify the common constructs we perform a
principal factor analysis. Factor analysis uses the correlation patterns among the observed financial
reporting variables to identify unobserved latent factors. We conduct the factor analysis using the
principal components method and the squared multiple correlation between the variable and all
other variables for the prior communality estimates.9 We then perform a linear transformation on
7 Results are unaffected by this choice, albeit weaker due to the smaller sample size.
8 One could plausibly increase the number of countries in the sample and increase the degrees of freedom available to
test associations between country attributes and financial reporting outcomes by relaxing our data availability
constraint. However, for many countries some of the variables are either unavailable from realiable sources, or
incorrectly measured and this would increase measurement error. Tabulated data for the publicly-available country
variables are available from the authors upon request. 9As the proportion of common variance among variables is not known in advance, factor analysis requires an initial
estimate. We set the squared multiple correlation as the prior communality. This criterion is widely used in the
literature, but for robustness we also repeat the factor analysis using two alternatives: i) the maximum absolute
correlation, and (ii) a random correlation.We obtain a similar solution with one latent reporting factor.
11
the factor solution applying varimax rotation. Table 2 presents the results. The factor analysis
suggests one single underlying financial reporting factor. The single factor explains about 89% of
the total cross-country variation in reporting measures, and all variables have high loadings on the
single factor (see Panel A of Table 2). The generally used rule of selecting factors with eigenvalues
of one or higher (Kaiser [1960]) and the reliability score of 0.858 (Cronbach [1951]) confirms the
one factor solution.10
Panel B of Table 2 presents the standardized scores of the financial reporting
latent factor for all the countries. Figure 1 graphically displays the factor scores. Ireland, United
Kingdom and United States have the highest reporting quality factor scores whereas Taiwan,
Pakistan and Greece have the lowest factor scores.
3.2. The correlations and potential interdependencies among country attributes
To demonstrate the high interdependency between the country attributes, we begin with a
regression analysis of country attributes on other country attributes.11
In Table 3 we report the
adjusted R2 of 72 separate regressions for each country-level variable regressed on 4 other country-
level variables that yield the highest adjusted R2. The purpose of this exercise is to assess the
maximum ability of various country-level variables to explain other country-level variables. In each
regression we remove any variable that is mechanically related to both the dependent and the
independent variables or that represents a similar construct so that the resulting adjusted R2s are not
mechanically inflated.12
The adjusted R2s reported in Table 3 are generally very high ranging from
10
The common rule of thumb is that the reliability measure should be at least 0.5 with many researchers suggesting a
minimum value of 0.7. 11
We also perform a correlation analysis for the country attributes. In Appendix 2 - Panel A we list all 72 previously
identified variables and the set of other variables most correlated with each of these proxies for country characteristics.
We tabulate the list of variables that are very highly correlated (absolute correlation of 0.7 or higher), highly correlated
(absolute correlation between 0.5 and 0.7), and moderately correlated (absolute correlation between 0.3 and 0.5). We
note that all measures have correlations of 0.3 or higher with at least one other variable proxying for country
charcteristics. On average each measure exhibits correlations of 0.3 or higher with 10 other variables (see Appendix 2
Panel B). Moreover, many measures have very high correlations with other measures. On average each measure has
correlations of 0.7 or higher with 3 or 4 other variables, and correlations of 0.5 or higher with 5 or 6 other variables. 12
An example of a mechanically related variable is secrecy, which is the combination of three other variables
(uncertainty avoidance plus power distance minus individualism). An example of two measures that represent the same
institutional characteristic is political stability and low political risk.
12
0.33 for the audit spending variable to 0.97 for control of corruption. The high joint explanatory
power is in line with the high pair-wise correlations between variables tabulated in Appendix 2 and
it also confirms our observed interdependencies between country-level attributes. These
interdependencies between country attributes make it difficult to causally isolate, or even
incrementally associate, the effect of any single country-level attribute on observed financial
reporting outcomes (or, more generally, other economic outcomes). Although isolating individual
effects is difficult to achieve because of the degree and complexity of the interactions between
country variables, we argue that the relations and potential complementarities between constructs
should be recognized in empirical examinations that relate economic outcomes to country-level
characteristics in international settings. Another practical implication is the dimensionality problem
(see, for example, Leuz and Wysocki [2016]). The fact that there are more country-level variables
than country observations with reliable and available data leads to estimation problems. Moreover,
using time-series variation in the data is unlikely to resolve the empirical problem because many
country-level features tend change very slowly, if at all, or co-move with other country attributes.
3.3. Factor analysis of country-level variables
As a first step toward better understanding and addressing the aforementioned
dimensionality problem, we describe patterns in the country-level variables and investigate whether
the observed patterns are explained by a much reduced and empirically-tractable set of (latent)
factors that capture common variation in the country variables. As with the financial reporting
quality metrics, we analyse common variation in the country attributes using factor analysis. Factor
analysis takes into account the correlation patterns among the country attributtes to identify
unobserved latent factors. Factor analysis also helps address the high dimensionality problem as it
significantly reduces the number of possible country-level variables that explain variation in
another economic outome (such as financial reporting). We apply factor analysis using the principal
13
components method and extract four orthogonal factors using varimax rotation.13
The four-factor
outcome represents a balance between (i) explaining a large proportion of the variation, (ii)
retaining factors with substantial incremental explanatory power, and (iii) finding a parsimonious
solution.14 Table 4 summarizes the results of our factor analysis. The four factors explain about
58% of the total variation, with the first two factors alone explaining a significant portion (80%) of
that variation (Table 4 - Panel A).15
We also note that adding a fifth factor adds little in terms of
variation explained (less than 5%) and that the incremental explanatory power of a factor declines
in the number of factors. Table 4 - Panel B presents the loadings of each country-level variable on
the latent factors. To facilitate interpretation we only present factor loadings higher than 0.4. A
large number of country-level variables load highly in the first factor that accounts for 31% of the
total variation. This first factor comprises a mix of measures related to a country’s legal and
governance systems (e.g. regulatory quality and rule of law), economic welfare (e.g. Gdpc and
bank money in private sector to GDP), legal rights (e.g. creditor rights), and also social attributes
representing more informal institutions (e.g. trust and ethnic fractionalization). This heterogeneous
13
The squared multiple correlations cannot be used for the prior communality estimates because there are more
variables than country observations (i.e. our correlation matrix is singular). For simplification, we set prior
communality estimates equal to one, but we repeat the factor analysis assuming the largest absolute correlation or a
pseudo-random number uniformly distributed between 0 and 1. We also use the iterated principal factor method with
the largest absolute correlation for the prior communality estimates. In all cases the four factor solution continues to
explain more than half of the variation in the country variables. To simplify the factor structure and to obtain an
interpretable final solution, we apply an orthogonal rotation (varimax) of the factors. Other rotation methods result in
similar factor loadings and country scores. Specifically, we performed a quartimax rotation (orthogonal) and a promax
rotation (oblique). 14
In selecting the number of factors, we also consider commonly used criteria such as the Kaiser (1960) rule - selecting
factors with eigenvalues larger than one, and the Cattell (1966) screen test – selecting factors above the point of
inflection in a plot of eigenvalues. 15
The correlations that form the basis of the factor estimates can potentially suffer with the presence of two discrete
variables (legal origin and class action lawsuit). An alternative to analyse the relation between discrete and continuous
variables is the non-parametric Spearman rank correlation, i.e. essentially defining all variables as discrete, or the
polyserial correlation which is the inferred latent correlation between a continuous variable and a ordered categorical
variable (Drasgow [1986]). We compare the Pearson, Spearman and polyserial correlations of the two variables with all
other variables and conclude that they are similar. The correlations (Pearson) between the average Pearson, Spearman
and polyserial correlations are: (i) for legal origin - Pearson / Polyserial = 0.988, Pearson / Polyserial = 0.954,
Spearman/Polyserial = 0.947; and (ii) for class action suit - Pearson / Polyserial = 0.945, Pearson / Polyserial = 0.874,
Spearman/Polyserial = 0.858. Another alternative is to exclude the two discrete variables from the factor analysis. The
four latent factors based on only 70 variables continue to explain 58% of the cross-country variation.
14
mix suggests that economic, legal, political, and societal features are interconnected and cannot be
examined in isolation. The Cronbach’s measure of reliability presented at the bottom of Panel B is
very close to one (0.97) indicating that the set of measures are highly correlated with the latent
variable, and that they represent a cohesive factor.16 Therefore, the effect of one stand-alone
country-level variable on reporting outcomes potentially reflects variation in other variables or in
the group as a whole.
The second factor extracted from our factor analysis captures correlated variables such as
creditor and investor rights (e.g. creditor rights and private control of self-dealing index), securities
regulation (e.g. securities regulation liability standards, and securities regulation disclosure
requirements), capital market size (e.g. market cap. to GDP), and legal origin. Social characteristics
such as English proficiency and uncertainty avoidance are also associated with capital market
development. The Cronbach’s measure of reliability is also close to one (0.931). Factor 3 has high
loadings in measures related to the political process (legislative competition and number of veto
players), and some financial and tax reporting system characteristics (book tax independence, tax
compliance, and enforcement of accounting standards). The reliability measure for factor 3 meets
the recommended cut-off value of 0.7. Factor 4 captures the openness or closeness of society
particularly in relation to external investment. This factor is characterized by high negative loadings
on measures representing US institutional holdings, US cross-listing, audit spending and English
proficiency, and high positive loadings on long-term orientation, Buddhist, and bank money in
private sector to GDP. These characteristics are typical of closed economies.
Table 4 - Panel C presents the standardized factor scores for each country. For consistency
across factors we revert the scores of factor 4 so that it represents openness rather than closeness of
16
For the purpose of computing the measure of reliability, we assign variables only to the factor where they load the
highest. This avoids assigning arbitrary weights to variable loadings. But we note that the majority of the variables load
highly in only one factor.
15
society. Pakistan and the Philippines have the lowest factor 1 score (general institutional
development) whereas Finland and Sweden have the highest score. Figure 1 displays factor 1 scores
for the countries in our sample. In Appendix 3 we report the country factor scores for an extended
list of 47 countries. The US has the highest rank in terms of political systems and reporting
enforcement. It has the fifth most developed capital market and ranks 17th in the first factor
representing general institutional development.
3.4. Relation between country-level variables and financial reporting
In this section, we examine the robustness of the empirical associations between individual
country-level variables and the financial reporting factor. Specifically, we estimate the statistical
significance and incremental explanatory power of individual variables in explaining variation in
financial reporting quality (q) when (i) regressions include only an individual country-level
explanatory variable, and (ii) regressions include both an individual country variable and four latent
country factors. We start by comparing the estimated t-statistics for a country-level explanatory
variable (vi) using menthods (i) and (ii) above. Table 5 compares the t-statistics for each individual
country variable vi obtained from estimating two regression models: (i) a reporting quality (q)
regression using only an individual country-level explanatory variable vi (unidimensional model q
= a + c*vi), and (ii) a reporting quality (q) regression using both an individual country-level
explanatory variable vi reporting quality (q) in combination with four factors extracted from a factor
analysis using all country-level measures excluding the identified individual variable,
(multidimensional model q = a+b1*f1+b2*f2+b3*f3+b4*f4+c*vi). By comparing the t-statistics
obtained in these two models we examine two important aspects. First, whether a certain country-
level variable is associated with variation in financial reporting (a research design typically applied
in previous international studies). Second, after considering the country features that co-move with
a given variable, whether that country variable still significantly explains variation in observed
16
financial reporting outcomes. The main result shown in Table 5 is the large drop in the number of
statistically significant variables in the second model. The first column in Table 5 shows that
individually the majority of country-level variables (45 of the 72 variables) are statistically
significantly associated with the variation in financial reporting quality and in the direction
typically identified in the extant literature. In contrast, the second column for the multidimensional
model shows only eight significant measures. This result highlights the interdependency between
the country-level measures and is in line with the high correlations observed in Table 3.
We next apply a multiple testing approach to the significance levels and estimate t-statistics
that control for family-wise error rate based on either the Bonferroni adjustment or the Holm
adjustment. The Bonferroni adjustment controls for the probablility of a single false discovery by
equally penalizing each p-value by the number of tests conducted in the same sample (72). The
Holm correction uses a sequential approach that is based on the rank order of the p-values and thus
is less restrictive than the Bonferroni adjustment. For both types of multiple testing adjustments we
observe that only 15 of the country-level variables are statistically significant in the univariate
model (column 1 of Table 5), and only 1 variable is significant in the multivariate model (column 2
of Table 5). 17
Overall, this striking empirical evidence suggests that essentially none of the
individual country attributes are found to be statistically significant incremental determinants of
international reporting after one accounts for 4 latent factors that distill the explanatory power of
other known and documented country attributes.
In Table 6 we perform a similar analysis to that described for Table 5 but examine the
explained variation. We start by comparing the variation in the financial reporting factor that is
17
To assess the possibility that the only seemingly significant variable (Individualism) is a false positive, we perform
Monte Carlo simulations (10,000 iterations) for the multiple regression model that includes Indvidualism as an
explanatory variable and obtain parameter estimates for Individualism under several different data generation processes.
For many of the simulations, we fail to reject the null hypothesis that the estimated coefficient on Individualism is
statistically different from zero. Thus, this simulation evidence suggests that even the apparent significance of
Individualism may be a statistical fluke.
17
explained by the 4 latent country-level factors with the variation explained by each of the 72
individual country-level variables. If the country latent factors reasonably capture the multifaceted
structures that affect financial reporting, then we expect these factors to explain a substantial
portion of cross-country variation in financial reporting. Further, if an individual country variable
conveys explanatory power in excess of that explained by the identified country latent factors, then
the single variable likely represents a distinct country-level feature that separately explains
reporting quality outcomes (q). We compare the explanatory power of the following three models.
Model 1 is the unidimensional model above, i.e. a regression of the individual variable vi on the
reporting quality factor q (q = a + c*vi). Model 2 is a regression of four factors extracted from
factor analysis using all country-level variables while excluding the identified individual variable
on the reporting quality factor q (q = a+b1*f1+b2*f2+b3*f3+b4*f4). We label Model 3 as the
multidimensional model above, i.e. a regression of four factors extracted from factor analysis using
all country-level variables while excluding the identified individual variable, and the individual
variable vi on the reporting quality factor q (q = a+b1*f1+b2*f2+b3*f3 +b4*f4+c*vi). Two patterns
emerge from Table 6. First, we observe a considerable increase in the explanatory power when
moving from model 1 to model 2. The average adjusted R2 increases from 16% to 72%. Second, the
explanatory power is unchanged as we move from model 2 to model 3. The four factors used in
model 2 capture a significant part of the cross-country variation in financial reporting. Hence, most
of the individual variables considered in isolation add little, if any, explanatory power to the model
because their effect is captured by correlated country attributes. Moreover, we find that only one
variable (individualism) has significant additional explanatory power in a multiple testing context.
18
We now turn to the four original latent factors as reported in Table 4. We examine the
association between these latent factors and the financial reporting factor.18
We assign the countries
to high, medium and low groups for each country factor and compare the mean and median values
of the financial reporting factor across the groups. The univariate results reported in Panel A of
Table 7 show that financial reporting quality improves as we move from the low quality group to
the high quality group which supports the premise of a positive association between the quality of
country-level characteristics and firms’ reporting outcomes. Regression results are presented in
Panel B of Table 7. All four country factors are positively associated with financial reporting
outcomes. In particular, factor 1 representing several characteristics associated with economic and
social development and country factor 2 with high loadings of variables related to capital market
development are significantly associated with reporting quality at the 1% level, in all model
specifications. The explanatory power of the country factors is quite high. In general, adjusted R2’s
are higher than 50% and the adjusted R2 for models including all four country factors are higher
than 70%. Hence, the four country factors that capture the interrelatedness of various country-level
features explain a substantial and significant portion of the international variation in financial
reporting outcomes.19
4. Can other empirical methods crack the country comparison problem?
In this section, we examine other candidate empirical methods that potentially address the
limitations of correlation and regression analyses in isolating which country attributes affect
financial reporting diversity around the world. As discussed in the prior sections, the four
18
The regression analyses preformed in this study are based on strictly linear models with no interactive effects. Prior
research suggests possible interactive effects between individual institutional variables (see, for example, Leuz et al.
[2003]). While non-linear specifications and interactive effects may result in greater explanatory power for reporting
quality, the use of non-linear and interactive approaches would not affect the main conclusions of this study.
Specifically, the high correlations among individual country variables make it difficult, if not impossible, to separate
and ascribe the effects of individual country variables (or interactions among individual country variables). 19
As a robustness test we repeated the analysis excluding the US. Our conclusions regarding the country factors,
reporting quality factor, and regressions results do not change.
19
fundamental problems facing comparative country analyses arises from: (i) the limited number of
country observations, (ii) the static or generally slow moving nature of country attributes, (iii) the
plethora of candidate “theories” and associated country variables that potentially explain reporting
outcomes, and (iv) the high correlation among these possible explanatory variables (i.e., high causal
density). In summary, the lack of degrees of freedom in multiple regression analyses arising from
too many candidate theories/attributes/variables and too few empirical observations make it
difficult to separate the competing effects.
4.1 Alternate approaches to constructing factors
Our previous analysis took a purely statistical approach by using factor analysis to identify
the latent factors from 72 country attributes, which results in factors that are difficult to label or
define since several seemingly unrelated variables load on the same factor. Further, in the interest
of parsimony we restrict the number of factors used in our empirical analysis to only four.
However, there are potentially more factors that could explain reporting quality variation across
countries if a researcher exercises her judgment in identifying distinct measures ex-ante based on
intuition. In this section we modify our approach by explicitly and ex-ante categorizing the 72
country-level variables based on distinct common categories used in the international literature:
Economic, Sociological, Geopolitical, Regulatory, and Legal categories. We then examine whether
this categorization better explains cross-country variation in attributes as well as incrementally
explains cross-country reporting quality differences. Our categorization is described in Table 8 -
Panel A.
We further analyze the within-category correlations among variables and find these to be
substantial and significant (bottom of Table 8 – Panel A). Because of the high within-category
correlations, and for parsimony, we perform factor analyses within each category. The independent
within-category factor analyses results in 10 factors with significant eigenvalues; two Economic
20
factors, four Sociological factors, a Geopolitical factor, two Regulatory factors and a Legal factor.
Collectively, these ten factors explain 58% of the total cross-country variation in attributes.
We then determine the power of the category factors individually and
collectively in explaining cross-country reporting quality variation using regression analysis
(Columns 1-6 of Table 8 - Panel B). We find that all categories except geopolitical load
significantly in these regressions, although the explanatory power of each category varies. The two
economics factors explain almost half of the variation in reporting quality, the four sociological
factors 57%, the two regulatory factors 55% and the legal factor (the single variable legal origin)
25%. In a regression of the reporting quality factor on all ten category factors (column 6 of Table 8)
the adjusted R2 is 78%, which is marginally higher than the 72% adjusted R2 obtained using only
four factors described in section 3. From these results, we infer that an ex-ante classification of
country attributes in an effort to easily label them or enhance their explanatory power is limited in
its ability to overcome the limited observations and high causal density problems in cross-country
research. We confirm this intuition by examining the correlations among all the category factors
(Table 8 - Panel C). These high correlations attest to our conclusions that (i) the category-based
factors do explain much of the cross-country variation in reporting quality, but (ii) they are highly
correlated and thus suffer from the same issues that affect cross-country studies that relate
individual attributes to financial reporting and economic outcomes.
4.2 Alternate empirical methodologies
Several scholars have proposed using alternative empirical methods to help test hypotheses
regarding which variables are associated with or even cause observed reporting outcomes around
21
the world. A recent survey by Gow, Larcker and Riess [2016] suggests that structural equation
modelling (SEM) may provide a practical path forward for empirical accounting research using
observational data. In addition, Wysocki [2011] suggests structural equation modelling as a
possible tool to help identify the associations between reporting quality and other economic
outcomes around the world. One important advantage of SEM over standard regression analysis is
that it allows for financial reporting quality to be both an outcome itself and a possible causal
determinant of other economic outcomes (such as financial market development or cost of capital).
However, a major limitation impeding the valid use of structural equation modelling for
international comparative analyses is the lack of formal and well-defined theories to guide the
correct implementation of pathway models for country-level data. While this problem is not
insurmountable, it suggests much work for future accounting researchers.
However, the application of SEM methods for international comparative analyses is still
stymied by exactly the same empirical issues facing correlation and regression; it is an issue of too
many competing and highly correlated country attributes and variables and too few country
observations. So, while one could potentially develop a formal testable theory, apply SEM methods,
and find evidence that a variable (say, investor protection) appears to explain reporting quality, the
problem remains that there are dozens of other highly correlated variables (with associated stories,
ad hoc models or even full blown structural models) that would have indistinguishable and
confounding empirical effects. The lack of observations and the high correlation among the
candidate explanatory theories/variables implies that SEM faces the same problems that are
highlighted in sections 2 and 3 of our study.
Scholars in accounting (see, for example, Leuz and Wysocki [2016]) and in other fields
recommend Qualitative Comparative Analysis (QCA) and Fuzzy Set Analysis as empirical tools that
potentially help identify the correlation structure of observational data and to qualitatively test the
22
consistency of the data with deterministic hypotheses (see, for example Schneider and Wagemann,
[2012], Grant et al. [2010], Vaisey [2007], and Ragin [2000]). QCA is an empirical tool that
evaluates the relation between observed dichotomous outcome variable(s) and all possible Boolean
combinations of other (explanatory) variable sets. QCA examines which set combinations or
configurations are most likely to be present when an outcome variable of interest has a “high”
value. Fuzzy Set analyses extend the QCA methodology beyond dichotomous variables to include
continuous variables (that range from 0 to 1). In an on-line appendix (On-Line Appendix 1), we
provide a simple case example showing the use of QCA/Fuzzy Set analysis on the dataset of 35
country observations used in sections 2-4 of this study.
However, like SEM methods, QCA/Fuzzy Set analyses also suffer from the same empirical
issues facing correlation and regression analyses; namely, too many competing and highly
correlated country attributes and variables and too few country observations. Again, one could
outline a formal testable theory, apply QCA/Fuzzy Set methods, and seem to find qualitative
evidence of a pattern of 5 or 6 country variables that “appear together” when reporting quality is
high. However, the problem with using QCA/Fuzzy Set methods for international comparative
analyses is that there are about 35 country observations, but there are 72+ highly correlated
variables that would have indistinguishable and inseparable empirical effects.
5. Out of sample tests: The N+1th
explanatory variable
An important issue affecting international studies is the difficulty in isolating the impact of
an individual institutional variable on accounting outcomes (Leuz and Wysocki [2016]). Many
studies proposing a relation between a country-level variable and firms’ reporting outcomes do not
assess whether the proposed feature has explanatory power over and above the existing set of
relations with other previously documented country-level characteristics, and how that feature is
related to the broad set of known country characteristics. We shed some light on this issue by
23
performing the following analysis. We select a variable that is not included in the set of our 72
institutional measures and examine its incremental explanatory power on financial reporting quality
measures over and above the four institutional factors derived from the factor analysis. As an
example, we use the economic freedom index recently proposed by Chen et al. [2015] to explain
firms’ investment decisions. Chen et al. [2015] find that greater economic freedom enhances
investment efficiency, which translates into higher earnings and book value multiples. The authors
also document that the effect of economic freedom is distinct from the effect of GDP, legal origin,
law enforcement, investor protection, and quality of a country’s accounting system. In Table 8 -
Panel A we present the correlations between our four country factors and the economic freedom
index reported in Chen et al. [2015]. The correlations with factor 1 and factor 2 are large and
statistically significant. In Table 8 - Panel B, we present regression results of the relation between
economic freedom, the four country factors, and reporting quality. The results of model (1), when
our country factors are excluded, are suggestive of a positive association between economic
freedom and financial reporting quality. However, when we include the multidimensionality of
country attributes represented by the four latent factors in the models, the statistical association
between economic freedom and financial reporting quality is no longer statistically significant.
Further, the explanatory power is substantially larger for models with the four country factors than
for the model with only economic freedom as an independent variable. These results indicate that
the country-level construct that economic freedom represents is already embedded in other country
features. Although the fact that the four factors capture a large number of country-level features
makes it difficult for any new measure to have incremental effects, the empirical results suggest
that the general patterns and complementarities between country’s institutions, endowments and
other features should be recognized when assessing the role of country-level characteristics on
economic outcomes.
24
We also test whether the high explanatory power of the country factors is exclusive of the
reporting outcomes represented in our financial reporting quality factor. We select an alternative
measure of reporting incentives from Chen et al. [2015]: market value of equity to earnings
(MV/E). This measure captures both the properties of information as prepared by managers and the
use of financial information by market participants. Country factor 2 representing institutional
features linked to capital markets is strongly associated with MV/E and explains 40% of the
variation in that variable (see Table 8 - Panel C). The results confirm the idea that country-level
institutional dimensions collectively affect firms’ reporting and economic outcomes. Moreover, the
effect of a given country-level variable is difficult to assess because of the many complementarities
and interactions between many co-existing features.
To further investigate this point, we examine other reporting outcomes. First, we estimate
the model that explains cross-country differences of the quality of accounting standards measured
as the difference between domestic standards and IFRS. The measure is developed by Bae et al.
(2008) and counts the differences between domestic standards and IFRS across 21 key accounting
items. In contrast to MV/E, this measure reflects only the quality of the information prepared by
managers. Second, we estimate the model that explains cross-country differences in analyst forecast
errors measured as the average absolute analyst forecast error for forecasts of earnings of current
year, one year and two years ahead (reported in Dhaliwal et al. [2012]). Our estimation results are
presented in Table 10 - Panel A. Higher “distance to” IFRS standards is mostly (negatively)
associated with institutions related to capital market development represented in factor 2. Analyst
forecast errors are negatively associated with factor 2 and with factor 3, which captures financial
reporting enforcement, taxation and political systems.20
We also test the relation between the two
outcome variables and the reporting quality factor and find a significant association. Overall, our
20
For the analyst forecast error variable we re-estimate the factor analysis excluding “number of analysts” given the
potential mechanical relation between the two variables
25
results suggest that the multidimensional set of country-level attributes affect a diverse set of
reporting outcomes and that there appear to be complementarities in reporting outcomes.
Several studies have used the properties of the accounting information to explain economic
outcomes such as capital market development, international cross-listing, corporate control,
ownership structure (e.g. La Porta et al. [1998], Dyck and Zingales [2002], Covrig et al. [2007], and
Fernandes and Ferreira [2008]). As an illustration, we test whether the reporting quality factor and
the country-level factors explain market capitalization. We obtain new country-level factors from
factor analysis excluding all variables related to capital market development (e.g. market
capitalization to GDP, listed firms to population, IPO’s to GDP, securities regulation). We report
the estimation results for the first two factors in Table 10 Panel B. We find that reporting quality is
positively related with capital market development and explains about 10% of the variation. The
two country-level factors are also significantly associated with capital market development.
Moreover, the reporting quality and capital market development association is statistically
insignificant when the country-level factors are included as additional regressors.
6. Conclusions and possible future research directions
The extant body of evidence in the empirical international accounting literature suggests
that a multitude of country-level characteristics and attributes each seem to be individually
associated with reporting outcomes around the world. This study empirically examines the multi-
dimensionality of and likely complementarities between country-level variables and how they
jointly relate to international variation in reporting and disclosure outcomes. We address the
incomplete picture in the existing accounting literature where the effects of ‘new’ country-level
variables on accounting outcomes are generally established without acknowledging or controlling
for the effects of numerous other known and documented country attributes.
26
First, we provide evidence that a single latent financial reporting factor appears to explain
most of the cross-country variation in six commonly used empirical measures of financial reporting
quality. We then investigate a comprehensive set of 72 country-level variables proposed in the
extant literature to explain international variation in economic outcomes, particularly accounting
quality. We find that country-level variables such as the quality of laws, governance and political
systems, financial and auditing enforcement mechanisms, capital market development, social
values, and culture and religious beliefs are highly correlated with each other and have high
combined explanatory power for each other. The strong co-movement among these variables
creates two problems for empirical researchers examining the relation between economic outcomes
and country-level features. First, because the country-level variables are highly correlated, it is
difficult to attribute an outcome to any one specific variable. Second, because there are fewer
country observations than there are measures of country attributes, it would appear to be difficult, if
not impossible, to control for and isolate the individual effects of all possible country attributes in
empirical studies.
In order to provide a framework and a feasible empirical methodology to better understand
the correlations between country-level variables and their relation to accounting quality, we
perform a factor analysis and identify four underlying factors that explain the common variation in
individual country-level variables across 35 countries. Moreover, after accounting for these four
factors in regression analyses, individual country variables have little, if any, ability to explain
variation in financial reporting outcomes around the world.21
Our findings suggest that accounting
outcomes, like other economic outcomes, are determined by a multidimensional and intertwined set
21
For clarity and for correct interpretation of this methodology, it is worth repeating that when evaluating the
incremental explanatory power of a given individual country variable (for example, investor protection), the four
factors are first estimated exluding the country variable of interest (for example, investor protection). Then, nested
regressions are estimated with accounting quality as the dependent variable and the four factors plus the given
individual country variable are used as explanatory variables.
27
of country-level conditions and their interactions. Hence, in order to determine the potential effect
of any ‘new’ country-level variable on observed financial reporting outcomes, one must explicitly
acknowledge and account for their association with the previously empirically identified set of
country-level attributes. Further, we highlight that the four-latent factors derived from country
attributes collectively explain a large portion of the differences in financial reporting outcomes
around the world.
In summary, there is a severe lack of degrees of freedom in current cross-country studies
with a plethora of correlated candidate theories/attributes/variables and too few empirical
observations to reasonably separate the confounding effects. We further highlight that this is not
really a problem of incorrect empirical tools (i.e., multiple regression vs structural modelling vs
qualitative comparative analysis), but an issue of too little data generated from too many
interrelated effects. We use 35 country observations in our empirical analyses to create and test a
dataset with as broad a set as possible of previously identified county attributes. It should be noted
that some international studies undertake analyses using data based on firm- or industry-level
observations to increase the sample size to hundreds, if not ten of thousands, of observations.
However, almost all comparative international empirical studies test “theories” and apply treatment
effects based on one or a few country-level variables. Therefore, using firm or industry observations
does not address or solve the fundamental problem identified in this study because the question of
interest is based on country-level (treatment) effects and we systematically document that a given
country variable is highly correlated with dozens of other country attributes. Thus, any claimed
explanatory power of a single country variable is very likely to be subsumed by other highly
correlated country attributes because there are only so many country observations. While the use of
times-series observations and changes analyses may help isolate and possibly separate the effects of
various competing country-level attributes or policies (see, for example, Christensen et al. [2013],
28
and Barth and Israeli [2013]), existing studies only account for a few examples of other competing
attributes and variables. However, as highlighted in this study, there are dozens of existing
competing theories/effects and at least 72 (and growing) documented country variables that are not
only highly correlated, but co-evolving over time.
6.1 Discussion of possible future directions for international research
A key insight of this study is that we formally document the high degree of correlation
among individual country attributes and thus the difficulty in separating the effects of individual
attributes. These findings have both retrospective and prospective implications for international
accounting research. Accounting researchers, doctoral students and policymakers can use the
evidence and data presented in this study to help interpret prior empirical studies on the
determinants and outcomes of financial reporting and disclosure around world. For example, many
existing international accounting studies apply conditional tests and regressions to test a particular
hypothesis on the determinants or outcomes of high quality financial reporting. For example,
research examining the association between variable X and reporting outcome Y may find that the
link is most pronounced when country attribute Z1 is high and this empirical evidence is then used
to argue that X “causes” Y.22 However, the evidence in this study suggests that many country
attributes (say, Z2, Z3, etc.) are often very highly correlated with country attribute Z1. Thus, it is
difficult to claim that the Z1-mediated effect of X on Y conclusively supports a particular
hypothesis when so many other plausible competing hypotheses with associated country variables
(Z2, Z3, etc.) provide equally compelling and empirically indistinguishable evidence that supports
22
Examples of this approach include Lang et al. [2004] who study the effect of analyst following mediated by variation
in investor protection across countries, or Pevzner et al. [2015] that investigates how societal trust affects market
reaction to earnings announcements as mediated by investor protection or disclosure requirements across countries.
29
alternate mechanisms for the documented determinants and outcomes of firms' financial reporting
policies.
On the upside, our findings can help guide future research exploring the determinants and
outcomes of international financial reporting. First, researchers who are investigating a ‘new’
country attribute can benchmark this attribute against the four empirical factors presented in this
paper. This benchmarking exercise can help provide insights into both the possible incremental
explanatory power (for various accounting or economic outcomes) of the ‘new’ country attribute
and also how the attribute ‘fits’ with other previously-identified groups of country attributes and
institutions.
Second, we suggest that future research should more explicitly acknowledge the
interdependencies among the many attributes and institutions that exist in countries and regions.
For example, rather than attempting to document that an isolated country attribute is associated with
variation in financial reporting quality across countries, studies would be better served by
acknowledging the portfolios of correlated country attributes and policies that work together as a
‘package’. For example, a study might model the effect of auditors on financial reporting outcomes
and then determine if this effect is more or less pronounced in countries that have a portfolio of key
institutions (for example, the correlated institutions that load strongly on Factor 1 in Table 4 of this
paper including strong rule of law, low political risk, high regulatory quality, low risk of
repudiation of contracts and expropriation by governments, high political stability, strong protection
of property rights, high judicial independence and efficiency). In other words, support for a
particular hypothesized link between auditors and financial reporting outcomes could be buttressed
if there is also supporting empirical evidence that a portfolio of strong institutions in a country
supports, or possibly substitutes for, the impact of auditors. While this portfolio approach may not
generate the type of simple policy prescriptions that some audiences may desire (i.e., countries only
30
need to change a single policy or ‘fix’ a single institution), it more reasonably reflects the realities
of how many institutions, attributes and features co-exist and interact in the real world.
Furthermore, this portfolio approach does not blindly emphasize one institutional mechanism while
ignoring other important and potentially complementary mechanisms affecting financial reporting
and other economic outcomes.
Third, the evidence in this paper highlights the opportunities to better understand how and
why portfolios of country attributes and institutions work together to influence financial reporting
and other economic outcomes. While advancing our understanding of these complementary effects
is very important, empirical researchers will still face the problem of the limited number of country
observations relative to the number of correlated country attributes. One possible step to addressing
the limited number of country observations available for international comparative analyses is to
also take advantage of within-country variation in provincial/state attributes and institutions. For
example, many countries exhibit large regional variation in laws, culture, language, history and
physical endowments within their borders.
Fourth, the recent accounting literature has correctly highlighted the need to find better-
identified empirical settings (such as natural experiments) to help isolate the stand-alone effect of a
particular economic variable or mechanism (see, for example, Leuz and Wysocki [2016]).
However, given that country attributes and institutions display very high correlations and appear to
work as systems, it may be a misguided strategy to search for “well-identified” research
experiments solely focused on single country attributes or stand-alone mechanisms. While a
researcher may be able to find an exogenous shock, the strong links among a country’s attributes
and institutions mean that: (i) any shock is unlikely to only impact a single economic mechanism
and thus would violate the exclusion restrictions necessary to draw any causal inferences about
stand-alone economic attributes or mechanisms, and (ii) even if a shock only directly affects a
31
single country attribute, the complementary links between country attributes and institutions
suggests a subsequent system-wide rebalancing of numerous attributes and institutions that would
make it difficult to isolate the role of any one institution or mechanism (also confounded by the
many variables in the system compared relatively few event observations).
Finally, we see future opportunities for international accounting researchers to better
capitalize on the unique features and advantages of empirical settings focused on single countries or
regions. While the prior international empirical accounting literature has produced a litany of
‘replication’ studies based on data from specific (non-U.S.) countries, these studies often
mechanically replicated earlier U.S. studies.23
These mechanical replications generally generated
little interest because they: (i) failed to motivate and explain why or how the findings should be
similar or different from those previously found in the U.S., and (ii) failed to capitalize on the
unique features and events in a country that could provide important experimental insights
unavailable in a U.S. research setting. Therefore, we recommend that international accounting
researchers strive to identify important research questions with potentially broad and generalizable
implications and then seek out powerful or unique experimental settings in specific countries or
regions that could not be addressed using the institutional environment or data available in the U.S.
In conclusion, we return to the three questions that motivated this study: Can all of these
attributes individually affect country-level reporting outcomes? Can existing empirical methods and
data be used to distinguish between the many proposed and competing ‘theories’ of the
determinants of international reporting diversity? What truly determines or influences the quality of
reported financial numbers across countries? At this time, the answers appear to be “the empirical
23
We acknowledge the important role for replication studies in empirical accounting research (as well as other
academic research fields) in documenting the validity and robustness of claimed initial discoveries. However, as
outlined above, the types of “replication” studies found in the prior international accounting literature often failed to
advance our understanding of why and how certain U.S. findings did or did not exist in non-U.S. samples.
32
evidence suggests no”, “unlikely”, and “we do not yet have a complete picture – more innovative
research is needed”.
33
References
Ali, A. and L. Hwang (2000). Country-specific factors related to financial reporting and the value
relevance of accounting data. Journal of Accounting Research 38 (1), 1–23.
Alesina, A., A. Devleeschauwer, W. Easterly, S. Kurlat, and R. Wacziarg (2003). Fractionalization.
Journal of Economic Growth, 8 (2), 155-194.
Alford, A., J. Jones, R. Leftwich, and M. Zmijewski. (1993). The relative information content of
accounting disclosures in different countries. Journal of Accounting Research 31 (Supplement):
183-223.
Armstrong, C., M. Barth, A. Jagolinzer, and E. Riedl, E. (2010). Market reaction to the adoption of
IFRS in Europe. The Accounting Review 85 (1), 31-61.
Bae, K.-H., H. Tan and M. Welker (2008). International GAAP differences: The impact on foreign
analysts. The Accounting Review 83 (3), 593–628.
Ball, R., S.P. Kothari and A. Robin (2000). The effect of international institutional factors on
properties of accounting earnings. Journal of Accounting and Economics 29 (1), 1–51.
Ball, R., A. Robin, A. and J. Wu (2000). Incentives versus standards: properties of accounting
income in four East Asian countries. Journal of Accounting and Economics 36 (1-3), 235 – 270.
Barth, M., D. Israeli (2013). Disentangling mandatory IFRS reporting and changes in enforcement.
Journal of Accounting and Economics 56 (2), 178-188.
Barth, M., W. Landsman and L. Lang (2008). International accounting standards and accounting
quality. Journal of Accounting Research 46 (3), 467-498.
Barth. M., W. Landsman and C. Williams (2012). Are IFRS-based and USGAAP-based accounting
amounts comparable? Journal of Accounting and Economics 54 (1), 68–93.
Basu, S. (1997). The conservatism principle and the asymmetric timeliness of earnings. Journal of
Accounting and Economics 24(1), 3-37.
Batta, G., R. Heredia and M. Weidenmier (2014). Political connections and accounting quality
under high expropriation risk. European Accounting Review 23 (4), 485–517.
Bhattacharya, U., H. Daouk and M. Welker (2003). The world price of earnings opacity. The
Accounting Review 78 (3), 641–678.
Byard, D., Y. Li and Y. Yu (2011). The effect of mandatory IFRS adoption on financial analysts’
information environment. Journal of Accounting Research 49 (1), 69–96.
Beck, T., A. Demirguç-Kunt and R. Levine (2001). The financial structure database. In: Demirguc-
Kunt, A., Levine, R. (Eds.), Financial Structure and Economic Growth: A Cross-Country
Comparison of Banks, Markets, and Development. MIT Press, Cambridge, MA, 17–80.
34
Beck, T., A. Demirguç-Kunt, A. and R. Levine (2003) Law, endowments, and finance. Journal of
Financial Economics 70 (2), 137–181.
Boutchkova, M., H. Doshi, A. Durnev and A. Molchanov (2012). Precarious politics and return
volatility. The Review of Financial Studies 25 (4), 1111-1154.
Bradshaw, M., B. Bushee and S. Miller. (2004). Accounting choice, home bias, and US investment
in non-US firms. Journal of Accounting Research 42 (5), 795–841.
Brochet, F., P. Naranjo and G. Yu. (2016). The capital market consequences of language barriers in
the conference calls of non-U.S. firms. The Accounting Review 91 (4), 1023-1049.
Brown, P., J. Preiato and A. Tarca (2014). Measuring country differences in enforcement of
accounting standards: an audit and enforcement proxy. Journal of Business, Finance and
Accounting, 41(1-2), 1-52.
Bushman,R. and Piotroski, J. and Smith, A. (2004). What determines corporate transparency.
Journal of Accounting Research 42 (2), 207-252.
Bushman, R. and J. Piotroski (2006). Financial reporting incentives for conservative accounting:
the influence of legal and political institutions. Journal of Accounting and Economics 42 (1-2), 107-
148.
Callen, J., O-K. Hope and D. Segal (2005), ‘Domestic and foreign earnings, stock return variability,
and the impact of investor sophistication. Journal of Accounting Research 43 (3), 377 – 412.
Cattel, R. B. (1966). The scree test for the number of factors. Multivariate Behavioral Research, I:
245-276.
Christensen, H., L. Hail, L. and C. Leuz (2013). Mandatory IFRS reporting and changes in
enforcement. Journal of Accounting and Economics 56 (2), 147-177.
Christensen, B. H. Lee, M. Walker and C. Zeng (2015). Incentives or standards: what determines
accounting quality changes around IFRS adoption? European Accounting Review 24 (1), 31-61.
Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika 22 (3):
297-334.
Chen, C-Y., P. Chen and Q. Jin (2015). Economic freedom, investment flexibility, and equity
value: a cross-country study. The Accounting Review 90 (5), 1839-1870.
Choy, S., T. Zheng (2011). Disclosure environment and earnings announcement premia.
International evidence and further U.S. evidence. University of Toronto, Working Paper.
Core, J., H. Luzi and R. Verdi (2014). Mandatory disclosure quality, inside ownership, and cost of
capital. European Accounting Review 24 (1), 1-29.
35
Covrig, V., M. DeFond and M. Hung (2007). Home bias, foreign mutual fund holdings, and the
voluntary adoption of international accounting standards. Journal of Accounting Research 45 (1),
41–70.
Dang L.T., F. Moshirian and B. Zhang (2016). Commonality in news around the world. Journal of
Financial Economics 116 (1), 82–110.
Daske, H., L. Hail, C. Leuz and R.Verdi (2008). Mandatory IFRS reporting around the world: early
evidence on the economic consequences. Journal of Accounting Research 46 (5), 1085 – 1142.
Daske, H., L. Hail, C. Leuz and R. Verdi (2013). Adopting a label: heterogeneity in the economic
consequences around IAS/IFRS adoptions. Journal of Accounting Research 51 (3), 595-547.
Dyck, A. and L. Zingales (2004). Private benefits of control: an international comparison. Journal
of Finance 59 (2), 537-600.
Dhaliwal, D., S. Radhakrishnan, A. Tsang and Y. Yang (2012). Nonfinancial disclosure and analyst
forecast accuracy: international evidence on corporate social responsibility disclosure. The
Accounting Review 87 (3), 723-759.
DeFond, M., M. Hung, and R. Trezevant (2007). Investor protection and the information content of
annual earnings announcements: International evidence. Journal of Accounting and Economics 43
(1), 37–67.
DeFond, M., M. Hung and S. Li (2011). The impact of mandatory IFRS adoption on foreign mutual
fund ownership: the role of comparability. Journal of Accounting and Economics 51 (3), 240–258.
Ding, Y., J. Thomas and S. Herve (2005). Why do national GAAP differ from IAS? The role of
culture. The International Journal of Accounting 40 (4), 325-350.
Djankov, S., R. La Porta, F. Lopez-de-Silanes, Florencio and A. Shleifer (2008). The law and
economics of self-dealing. Journal of Financial Economics 88 (3), 430–465
Doidge, C., A. Karolyi and R. Stulz (2004). Why are foreign firms listed in the U.S. worth more?
Journal of Financial Economics 71 (2), 205–38.
Drasgow, F. (1986). Polychoric and polyserial correlations in Encyclopedia of Statistical Sciences
7, ed. S. Kotz and N. L. Johnson, New York: John Wiley & Sons, 68–74.
Faccio (2006). Politically connected firms. The American Economic Review 96 (1), 369-386.
Francis, J. and D. Wang (2008). The joint effect of investor protection and big 4 audits on eamings
quality around the world. Contemporary Accounting Research 25 (1), 157-191.
Florou, A. and P. Pope (2012). Mandatory IFRS adoption and institutional investment decisions.
The Accounting Review 87 (6), 1993–2025.
36
Frost, C., E. Gordon and A. Hayes (2006). Stock exchange disclosure and market development: An
analysis of 50 international exchanges. Journal of Accounting Research 44 (3), 437-483.
Grant, D., M. Trautner, L. Downey and L.Thiebaud (2010). Bringing the polluters back in:
environmental inequality and the organization of chemical production. American Sociological
Review 75 (4), 479–504.
Green, J., J. Hand, and F. Zhang (2013). The supraview of return predictive signals. Review of
Accounting Studies 18 (3), 692-730.
Green, J., J. Hand, and F. Zhang (2014). The remarkable multidimensionality in the cross-section of
expected U.S. stock returns. Unpublished paper, Yale University. Available at
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2262374.
Gow, I., Larcker, D. and Reiss, P. (2016). Causal Inference in Accounting Research. Journal of
Accounting Research 54 (2), 477-523.
Gupta, M., I. Khurana and R. Pereira (2008). Legal enforcement, short maturity debt, and the
incentive to manage earnings. Journal of Law and Economics 51 (4), 619-639.
Hail, L. and C. Leuz (2006). International differences in the cost of equity capital: do legal
institutions and securities regulation matter? Journal of Accounting Research 44 (39), 485–531.
Hail, L. and C. Leuz (2009). Cost of capital effects and changes in growth expectations around U.S.
cross-listings. Journal of Accounting and Economics 93 (3), 428-454.
Harvey, C. Y. Liu, and H. Zhu (2016) …and the cross-section of expected returns. Review of
Financial Studies 29 (1), 5 - 68.
Haw, I-M., B. Hu, L. Hwang and W. Wu (2004). Ultimate ownership, income management, and
legal and extra-legal institutions. Journal of Accounting Research 42 (2), 423-462.
Hofstede, G., G. Hofstede and M. Minkov (2010). Cultures and organizations: Software of the
mind. 3rd edition. McGraw-Hill.
Hope, O.-K (2003). Disclosure practices, enforcement of accounting standards, and analysts’
forecast accuracy: an international study. Journal of Accounting Research 41 (2), 235–272.
Hope, O., T. Kang, W. Thomas and Y. Yoo (2008). Culture and auditor choice: a test of the secrecy
hypothesis. Journal of Accounting and Public Policy 27 (5), 357-373.
Hope, O-K., T. Kang, and J. Kim (2013). Voluntary disclosure practices by foreign firms cross-
listed in the United States. Journal of Contemporary Accounting and Economics (forthcoming).
Hope, O-K., W. Thomas and D. Vyas (2011). The cost of pride: why do firms from developing
countries bid higher? Journal of International Business Studies 42 (1), 128–151.
37
Hung, M. (2001). Accounting standards and the value relevance of financial statements. Journal of
Accounting and Economics 30 (3), 401-420.
Kaiser, H.F. (1960). The application of electronic computers to factor analysis. Educational and
Psychological Measurement 20, 141–151.
Kanagaretnam, K., C. Lim and G. Lobo (2011). Effects of national culture on earnings quality of
banks. Journal of International Business Studies 42 (6), 853-874.
Kanagaretnam, K., C. Lim and G. Lobo (2013). Effects of international institutional factors on
earnings quality of banks. Journal of Banking and Finance 39 (2), 87-106.
Karolyi, G.A., K-H. Lee and M. Van Dijk (2012). Understanding commonality in liquidity around
the world. Journal of Financial Economics 105 (1), 82–112.
Kaufmann, D., A. Kraay and Mastuzzi. Worldwide Governance Indicators 1996–2013.
Washington, DC: The World Bank.
Kim. J. and H. Shi (2012). IFRS reporting, firm-specific information flows, and institutional
environments: international evidence. Review of Accounting Studies 17 (3), 474–517.
Kim, J., J. Tsui and C. Yi (2011). The voluntary adoption of International Financial Reporting
Standards and loan contracting around. the world. Review of Accounting Studies 16 (4), 779–811.
La Porta, R., F. Lopez-de-Silanes , A. Shleifer and R. Vishny (1998). Law and Finance. Journal of
Political Economy 106 (6), 113-1155.
La Porta, R., F. Lopez-de-Silanes , A. Shleifer and R. Vishny (1999). The quality of government.
Journal of Law, Economics, and Organization 15, 222–279.
La Porta, R., F. Lopez-de-Silanes, F. and A. Shleifer, A. (2006). What works in securities laws.
Journal of Finance 61 (1), 1-32.
Land, J. and M. Lang, M. (2002) Empirical evidence on the evolution of international earnings. The
Accounting Review 77, 115–133.
Lang, M., M. Maffett and E. Owens (2010). Earnings comovement and accounting comparability:
the effects of mandatory IFRS adoption. Unpublished paper, University of North Carolina.
Available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1676937&rec=1&srcabs=1754199
Landsman, W., E. Maydew and J. Thornock (2012). The information content of annual earnings
announcements and mandatory adoption of IFRS. Journal of Accounting and Economics 53 (1-2)
34–54.
Lang, M., K. Lins and D. Miller (2003a). ADRs, analysts, and accuracy: does cross listing in the
United States improve a firm's information environment and increase market value? Journal of
Accounting Research 41 (2), 317-345.
38
Lang, M., K. Lins and D. Miller (2004). Concentrated control, analyst following, and valuation: do
analysts matter most when investors are protected least? Journal of Accounting Research 42 (3),
589–623.
Lang, M., S. Raedy and M. Yetman, M. (2003b). How representative are firms that are cross-listed
in the United States? An analysis of accounting quality. Journal of Accounting Research 41 (2) ,
363 – 386.
Leuz, C., D. Nanda and P. Wysocki (2003). Earnings management and investor protection: an
international comparison. Journal of Financial Economics 69 (3), 505–527.
Leuz, C. (2010). Different approaches to corporate reporting regulation: how jurisdictions differ
and why. Accounting and Business Research 40 (3), 229-256.
Leuz, C. and P. Wysocki (2016). The economics of disclosure and financial reporting regulation:
evidence and suggestions for future research. Journal of Accounting Research 54 (2), 525-622.
Li, Q., E. Maydew, R. Willis and L. Xu (2016). Corporate tax behavior and political uncertainty:
evidence from national elections around the world. Unpublished paper, Vanderbilt Univeristy.
Available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2498198.
McGuire, S., T. Omer and N. Sharp (2012). The impact of religion on financial reporting
irregularities. The Accounting Review, 87 (2), 645-673.
Morck, R., B. Yeung and W. Yu (2000). The information content of stock markets: why do
emerging markets have synchronous stock price movements? Journal of Financial Economics 58
(1), 215-260.
Mueller, G., H. Gernon and G. Meek (1994). Accounting and international perspective. New York:
Business One Irwin.
Nanda, D. and P. Wysocki (2015). The relation between trust and accounting quality. Unpublished
paper, University of Miami.
Nguyen, N. and C. Truong (2013). The information content of stock markets around the world: A
cultural explanation. Journal of International Financial Markets, Institutions & Money 26, 1-29.
Pevzner, M., F. Xie and X. Xin (2015). When firms talk do investors listen? The role of trust in
stock market reactions to corporate earnings announcements. Journal of Financial Economics 117
(1), 190-223.
Ragin, C. (2000). Fuzzy-set social sciences. Chicago: University of Chicago Press.
Riahi-Belkaoui, A. (2004a). Politically-connected firms: are they connected to earnings opacity?
Research in Accounting Regulation 17, 25-38.
Riahi-Belkaoui, A. (2004b). Effects of corruption on earnings opacity internationally. Advances in
International Accounting 17, 73–84.
39
Schneider, C. and C. Wagemann (2012). Set-theoretic methods for the social sciences: a guide to
qualitative comparative analysis. Cambridge: Cambridge University Press.
Siegel, J., A. Licht and S. Schwartz (2011). Egalitarianism and international investment. Journal of
Financial Economics 102 (3), 621-642.
Srinivasan, S., A. Wahid and G. Yu (2015). Admitting mistakes: home country effect on the
reliability of restatement reporting. The Accounting Review 90 (3), 1201-1240.
Stulz, R. and R. Williamson (2003). Culture, openness, and finance. Journal of Financial
Economics 70 (3), 313–349.
Vaisey, S. (2007). Culture, structure, and community: the search for belonging in 50 urban
communes. American Sociological Review 72, 851-873.
Wysocki, P. (2011) New institutional accounting and IFRS. Accounting and Business Research 41
(3), 309-328.
Yu, G. and A. Wahid (2014). Accounting standards and international portfolio holdings. The
Accounting Review 89 (5), 1895-1930.
40
Table 1 - Summary statistics of 6 financial reporting outcomes for 35 countries Panel A – Descriptive statistics by country
Country
Reporting
transparency
Disclosure
quality
Asymmetric
timeliness
Abnormal
return
Abnormal
volume
Return
synchronicity
Argentina -0.391 68 0.401 3.79 0.135 0.510
Australia -0.078 80 0.343 5.41 0.664 2.033
Austria -0.808 62 -0.093 4.33 0.213 1.603
Belgium -0.682 68 0.181 4.55 0.766 1.777
Brazil -0.658 56 0.152 3.53 0.250 1.249
Canada -0.162 75 0.377 6.03 0.960 2.582
Switzerland -0.504 80 0.303 4.97 0.993 2.192
Chile -0.358 78 0.017 3.31 -0.110 1.200
Germany -0.620 67 0.22 3.88 0.329 1.872
Denmark -0.530 75 0.244 5.64 1.357 1.817
Spain -0.792 72 0.314 4.07 0.622 1.352
Finland -0.26 83 0.11 5.07 1.254 1.757
France -0.536 78 0.04 5.56 0.919 2.022
United Kingdom -0.133 85 0.276 8.94 1.744 1.814
Greece -0.881 61 0.087 3.49 -0.010 1.015
Hong Kong -0.521 73 0.256 6.49 1.286 1.306
Indonesia -0.715 65 0.046 3.43 0.436 1.351
India -0.537 61 0.156 5.11 0.860 1.263
Ireland -0.199 81 0.495 9.34 1.914 1.999
Israel -0.329 74 0.23 3.97 1.108 0.795
Italy -0.826 66 0.135 4.05 0.994 1.474
Japan -0.802 71 0.107 4.82 1.002 1.384
Mexico -0.502 71 0.466 4.8 0.514 0.915
Malaysia -0.643 79 0.125 3.93 0.331 1.230
Netherlands -0.482 74 0.177 7.59 1.362 1.815
New Zealand -0.121 80 0.419 5.42 0.916 1.913
Pakistan -0.706 73 -0.085 2.68 -0.100 0.756
Philippines -0.552 64 0.231 4.32 0.074 1.055
Portugal -0.88 56 0.263 3.68 0.327 1.781
Singapore -0.601 79 0.13 6.11 1.194 1.332
Sweden -0.168 83 0.486 5.26 1.424 1.532
Thailand -0.506 66 0.337 3.55 0.247 1.173
Taiwan -0.639 58 0.158 3.25 0.081 0.618
United States -0.228 76 0.312 7.48 1.654 2.142
South Africa -0.307 79 0.051 3.42 0.278 1.595
Panel B - Correlations
(1) (2) (3) (4) (5) (6)
(1) Reporting transparency 1
(2) Disclosure quality 0.706* 1
(3) Asymmetric timeliness 0.539* 0.259 1
(4) Abnormal return 0.539* 0.537* 0.440* 1
(5) Abnormal volume 0.485* 0.583* 0.431* 0.868* 1
(6) Return synchronicity 0.380* 0.423* 0.223 0.565* 0.552* 1
Table 4 reports summary statistics for 6 reporting quality variables used in the literature for 35 countries.
Panel A reports mean values of reporting quality variables (unstandardized) by country. Panel B reports
Pearson correlations. Variable definitions are provided in Appendix 1. The symbol * indicates statistical
significance at the 0.05 level.
41
Table 2 – Factor analysis of 6 financial reporting measures for 35 countries
Panel A – Rotated factor loadings and statistics of the factor analysis
Factor loadings Statistics of the factor analysis
Abnormal return 0.868 Eigenvalue: 3.211
Abnormal volume 0.868 Variation explained: 0.889
Reporting transparency 0.864 Cronbach’s measure of reliability : 0.858
Disclosure quality 0.748
Return synchronicity 0.728
Asymmetric Timeliness 0.587
Panel B – Standardized scores for financial reporting factor
Country Financial Reporting
score
Country Financial Reporting
score
Ireland 2.214 Japan -0.270
United Kingdom 2.083 Belgium -0.302
United States 1.543 Italy -0.511
Sweden 1.168 Malaysia -0.532
Canada 0.986 Spain -0.543
Netherlands 0.967 Argentina -0.561
New Zealand 0.914 Germany -0.612
Finland 0.817 Chile -0.642
Australia 0.797 Thailand -0.650
Hong Kong 0.620 Philippines -0.752
Denmark 0.586 Indonesia -0.900
Singapore 0.462 Austria -1.048
Switzerland 0.422 Brazil -1.058
France 0.302 Portugal -1.140
Israel 0.215 Taiwan -1.230
Mexico -0.150 Pakistan -1.335
India -0.162 Greece -1.449
South Africa -0.249
Table 2 reports the results of factor analysis on the 6 financial reporting outcome
variables for 35 countries The factor analysis is performed using the principal
components method and the squared multiple correlation between the variable and all
other variables for the prior communality estimates. As the proportion of common
variance among variables is not known in advance, factor analysis requires an initial
estimate. We set the squared multiple correlation as the initital prior communality. We
then perform a linear transformation on the factor solution applying varimax rotation.
Panel A reports factor loadings and statistics of the factor analysis. Panel B reports
standardized factor scores the estimated single reporting quality measure for 35
countries. Variable definitions are provided in Appendix 1.
42
Table 3 - Adjusted R2 of regressions of country variable on the other most correlated country
variables
Short name Dependent variable Adjusted R2 4 independent variables providing max adj. R
2
AntiDir Anti-director rights 0.584 ProprR, PolitStab, Latitude, Big4Share
LTaxEv Assessment of tax evasion 0.809 BTaxInd, ReligFract, CCorr, LegalO
AuditSpend Audit spending 0.330 DisclReg, USlisting, UncAvoid, ClassAction
BankPriv Bank money in private s. to GDP 0.744 MarkCap, LRepContr, Masculin, Big4Share
Big4Share Big4 market share 0.427 BankPriv, ForeignInv, ReligFract, Trust
BlockPr Block premium 0.567 DisclReg, USlisting, Religness, Masculin
BTaxInd Book tax independence 0.482 Latitude, InstHoldDom, ReligFract, Law
Budhist Buddhist 0.577 PublCtr, Protestant, EnglProf, LegalO
Catholic Catholic 0.660 InstHoldDom, IndividH, Trust, LegalO
ClassAction Class action lawsuit 0.546 MarkCap, PublicEnf, LegislComp, ReligFract
CCorr Control of corruption 0.974 InstHoldDom, Law, Protestant, RegQ
CCorrL Corruption 0.864 ProprR, PolitStab, UShold, IndividH
CreditR Creditor rights 0.862 BankPriv, AntiDir, LegalO, JudEff
Democracy Democracy 0.667 LegislComp, IndividH, IndividW, EnglProf
InstHoldDom Domestic institutional holdings 0.388 Latitude, UShold, IndividH, UncAvoid
EnforAccS Enforcement of accg. standards 0.651 DisclReg, LibStand, IndividH, Big4Share
EnforAudS Enforcement of audit standards 0.843 EnforAccS, AntiDir, Gdpc, BTaxInd
EnglProf English proficiency 0.645 LRepContr, ForeignInv, IndividH, LegalO
EthFract Ethnic fractionalization 0.615 LangFract, Religness, UncAvoid, Big4Share
PrivCtrEA Ex ante private control self-dealing 0.745 CreditR, PolitStab, IntHoldFor, JudEff
PrivCtrEP Ex post private control self-dealing 0.648 BTaxInd, AuditSpend, NrAnal, LegalO
IntHoldFor Foreign institutional holdings 0.486 BankPriv, LegislComp, Latitude, ClassAction
ForeignInv Foreign investment to GDP 0.599 MarkCap, ReligFract, Protestant, CCorr
Gdpc GDPc 0.881 LRepContr, LegislComp, Latitude, JudEff
HierIndep Hierarchy vs independence 0.417 DisclReg, BTaxInd, LegislComp, LangFract
IndividH Individualism 0.786 Gdpc, Democracy, ForeignInv, EnglProf
IndividW Individualism in income 0.576 BankPriv, OwnConc, PolitConn, USlisting
InfoKnow Information and knowledge 0.689 Gdpc, PolitStab, LangFract, Masculin
IPO IPOs to GDP 0.525 ListedF, PolitStab, InstHoldDom, CCorr
JudEff Judicial efficiency 0.835 CreditR, BankPriv, Gdpc, IndividW
JudIndep Judicial independence 0.943 ProprR, UShold, Trust, LegalO
LangFract Language fractionalization 0.688 ProprR, BlockPr, PolitStab, EthFract
EnglProx Language proximity to English 0.906 UShold, LTorient, HierIndep, EnglProf
Latitude Latitude 0.815 PublicEnf, ReligFract, PowerD, NrAnal
LawO Law and order 0.933 ProprR, LRepContr, PolitStab, CCorr
LegalO Legal origin 0.795 CreditR, DisclReg, Gdpc, PowerD
43
Table 3 - Adjusted R2 of regressions of country variable on the other most correlated country
variables
(cont.)
Short name Dependent variable Adjusted R2 4 independent variables providing max Adj R
2
LegislComp Legislative competition 0.516 Gdpc, Democracy, PolitConn, HierIndep
ListedF Listed firms to population 0.642 OwnConc, PublicEnf, Religness, LegalO
LTorient Long-term orientation 0.604 LRepContr, Democracy, Religness, CCorr
LPolitR Low political risk 0.952 LRepContr, PolitStab, LTorient, LegalO
LRepContr Low repudiation contracts by gov. 0.893 ProprR, Gdpc, PolitConn, UncAvoid
LExprR Low risk expropriation by gov. 0.943 MarkCap, LRepContr, Big4Share, ClassAction
MarkCap Market cap. to GDP 0.713 BankPriv, ListedF, ForeignInv, InfoKnow
Masculin Masculinity 0.440 Gdpc, Latitude, IndividW, CCorr
Media Media 0.752 PublicEnf, Gdpc, InfoKnow, EnglProf
Muslim Muslim 0.805 Democracy, PolitStab, Latitude, UShold
NrAnal Number of analysts 0.654 PublicEnf, Latitude, ReligFract, PowerD
Veto Number of veto-players 0.719 Democracy, LegislComp, Law, CCorr
OReligion Other religion 0.524 ListedF, Law, CCorr, JudEff
OwnConc Ownership concentration 0.553 ListedF, DisclReg, Religness, ClassAction
PolitScore Political score 0.939 Democracy, Latitude, Law, CCorr
PolitStab Political stability 0.853 ProprR, AntiDir, LangFract, Big4Share
PolitConn Politically connected firms 0.573 BankPriv, Gdpc, LegislComp, IndividW
PowerD Power distance 0.730 DisclReg, Democracy, InfoKnow, LegalO
PrivCtrIndex Private control of self-dealing index 0.743 BlockPr, Latitude, RegQ, LegalO
ProprR Property rights 0.949 JudIndep, UShold, RegQ, Trust
Protestant Protestant 0.659 ForeignInv, Law, Trust, CCorr
PublCtr Public control of self-dealing 0.346 BlockPr, Democracy, Law, RegQ
PublicEnf Public enforcement sec.regulation 0.673 ListedF, Latitude, LTorient, NrAnal
RegQ Regulatory quality 0.957 DisclReg, LibStand, Protestant, CCorr
ReligFract Religion fractionalization 0.605 BTaxInd, Latitude, Law, Big4Share
Religness Religiousness 0.781 LegislComp, EthFract, LTorient, CCorr
Law Rule of law 0.968 Democracy, USlisting, Protestant, CCorr
Secrecy Secrecy 0.830 MarkCap, Gdpc, Protestant, EnglProf
DisclReg Securities reg.disclosure req. 0.931 LibStand, EthFract, RegQ, JudEff
LibStand Securities reg.liability standards 0.939 DisclReg, EthFract, RegQ, CCorr
SSecRegul Strengh of securities regulation 0.878 BlockPr, DisclReg, PublicEnf, ClassAction
TaxComp Tax compliance 0.557 PublicEnf, PolitConn, EthFract, CCorr
Trust Trust 0.656 Latitude, PolitConn, Religness, UncAvoid
UncAvoid Uncertainty avoidance 0.704 DisclReg, LegislComp, Protestant, EnglProf
USlisting US cross-listing 0.791 UShold, LTorient, Big4Share, LegalO
UShold US institutional holdings 0.805 ListedF, PolitStab, USlisting, PowerD
Table 3 reports the adjusted R2 of each regression for a given country variable regressed on 4 explanatory
country variables chosen from a list of 72 possible country variables (the variable descriptions are provided
in Appendix 1). The selected 4 explanatory variables are the variables that maximize the explanatory power
of the regression for a given depedendent variable (ie, maximal adjusted R2 using 4 country variables).
44
Table 4 – Factor analysis of 72 country variables for 35 countries (description at bottom of Table 4)
Panel A: Factor analysis statistics - Variation explained by candidate factors
Eigenvalue Variation explained Cumulative variation explained
Country factor 1 22.261 0.309 0.309
Country factor 2 10.868 0.151 0.460
Country factor 3 5.049 0.070 0.530
Country factor 4 3.699 0.051 0.582
Country factor 5 3.229 0.045 0.627
Panel B: Rotated factor loadings
Country variables Factor 1 Factor 2 Factor 3 Factor 4
Control of corruption 0.938
Rule of law 0.938
GDPc 0.928
Low political risk 0.914
Law and order 0.907
Regulatory quality 0.901
Low repudiation of contracts by gov. 0.891
Corruption 0.888
Political score 0.887
Political stability 0.883
Low risk of expropriation by gov. 0.875
Property rights 0.843
Judicial independence 0.802 0.428
Judicial efficiency 0.794
Latitude 0.749
Media 0.735
Information and knowledge 0.722
Enforcement of audit standards 0.702
Individualism 0.696 0.540
Trust 0.635
Democracy 0.585
Protestant 0.571
Assessment of tax evasion 0.554 0.625
Foreign institutional holdings 0.530
Bank money in private sector to GDP 0.530 0.452
Tax compliance 0.507 -0.484
Creditor rights 0.457 0.768
Politically connected firms -0.508
Individualism in income -0.533
Language fractionalization -0.540
Muslim -0.569
Ethnic fractionalization -0.580
Secrecy -0.694
Power distance -0.707
Religiousness -0.847
45
Table 4 – Factor analysis of country variables (cont.)
Panel B: Rotated factor loadings (cont.)
Country variables Factor 1 Factor 2 Factor 3 Factor 4
Private control of self-dealing index 0.855
Legal origin 0.836
Securities regulation liability standards 0.830
Securities regulation disclosure requir. 0.816
Strength of securities regulation 0.761
Ex ante private control of self-dealing 0.756
Ex post private control of self-dealing 0.745
Listed firms to population 0.647
Anti-director rights 0.643
Market cap. to GDP 0.636
English proficiency 0.574 -0.433
Public enforcement securities regulation 0.556
Language proximity to English 0.527
Religion fractionalization 0.479 0.434
Other religion 0.454
Foreign investment to GDP 0.430
IPOs to GDP 0.420
Block premium -0.540
Catholic -0.592
Uncertainty avoidance -0.756
Book tax independence 0.658
Legislative competition 0.627
Class action lawsuit 0.618
Number of analysts 0.541
Domestic institutional holdings 0.514
Number of veto-players 0.479
Enforcement of accounting standards 0.425
Hierarchy vs independence -0.489
Long-term orientation 0.737
Buddhist 0.440
Audit spending -0.536
US institutional holdings -0.591
US cross-listing -0.641
Ownership concentration
Public control of self-dealing
Masculinity
Big4 market share
Cronbach’s alpha measure of reliability 0.970 0.931 0.727 0.701
46
Table 4 – Factor analysis of country variables (cont.)
Panel C – Standardized scores for the four latent country factors
Country
factor 1
Country
factor 2
Country
factor 3
Country
factor 4
Score Rank Score Rank Score Rank Score* Rank
Finland 1.409 1 -0.300 19 -1.062 31 0.543 9
Sweden 1.296 2 -0.324 21 0.122 15 0.454 12
Switzerland 1.296 3 -0.067 18 -0.147 20 -0.756 26
Denmark 1.295 4 0.118 15 -0.420 25 1.007 7
Netherlands 1.092 5 -0.339 22 0.248 14 0.497 11
Austria 1.043 6 -1.320 32 -0.622 28 0.180 16
Germany 0.953 7 -0.961 28 0.556 11 -1.043 32
New Zealand 0.802 8 0.792 8 0.027 16 1.156 5
Ireland 0.773 9 0.538 11 -0.895 30 2.027 1
United Kingdom 0.687 10 1.337 4 0.875 6 -0.272 22
Japan 0.643 11 -0.412 23 -0.384 24 -2.363 35
Australia 0.592 12 0.972 7 1.262 3 0.450 13
Belgium 0.533 13 -0.836 27 -0.106 19 -0.987 31
France 0.473 14 -0.612 25 0.625 10 -1.427 33
Canada 0.470 15 1.022 6 2.130 2 0.114 17
Hong Kong 0.446 16 2.033 2 -1.236 33 -0.107 21
United States 0.362 17 1.030 5 2.438 1 -0.020 20
Spain 0.129 18 -0.680 26 0.256 13 -0.323 23
Italy 0.082 19 -0.972 29 0.370 12 -0.967 30
Portugal 0.059 20 -1.250 31 -0.311 22 -0.013 19
Singapore 0.019 21 2.310 1 -2.489 35 -0.420 24
Chile -0.186 22 -0.482 24 -1.092 32 1.100 6
Israel -0.197 23 0.449 13 -0.043 17 1.328 4
Taiwan -0.240 24 0.111 16 -0.060 18 -1.805 34
Greece -0.276 25 -1.403 34 -0.274 21 -0.685 25
South Africa -0.830 26 0.757 9 0.668 9 0.723 8
Argentina -0.950 27 -1.448 35 -0.365 23 1.685 2
Mexico -0.972 28 -1.371 33 -0.620 27 1.441 3
Brazil -1.024 29 -1.225 30 0.781 7 0.531 10
Malaysia -1.286 30 1.538 3 -0.434 26 -0.908 29
Thailand -1.359 31 0.478 12 -0.884 29 -0.894 28
India -1.419 32 0.609 10 1.240 4 0.272 15
Indonesia -1.804 33 -0.013 17 -1.748 34 -0.883 27
Philippines -1.856 34 -0.301 20 0.711 8 0.063 18
Pakistan -2.055 35 0.220 14 0.882 5 0.301 14
Table 4 reports the results of factor analysis on 72 country variables for 35 countries. The factor analysis is
performed using the principal components method. The squared multiple correlations cannot be used for the prior
communality estimates in this sample because there are more variables than country observations (i.e. our
correlation matrix is singular). Thus, for tractability, we set prior communality estimates equal to one, The four-
factor outcome represents a balance between (i) explaining a large proportion of the variation, (ii) retaining
factors with substantial incremental explanatory power, and (iii) finding a parsimonious solution. Panel A reports
statistics for the factor analysis. Panel B reports the variable loadings on the four latent country factors (for
clarifity only loadings higher than 0.4 are printed). Panel C reports standardized factor scores for the 35 countries.
Variable definitions are provided in appendix 1. * For consistency across factors we revert the sign of factor.
47
Table 5 – T-statistics from regressions of financial reporting factor (q) on country variable (vi) from simple
unidimensional regression (q = a + c*vi) and multidimensional regressions that exclude the variable of interest from
the factors (q = a+b1*f1+b2*f2+b3*f3 +b4*f4+c*vi) Country variable (vi) Unidim.
t-stat on vi
Multidim.
t-stat on vi
Country variable (vi) Unidim
t-stat on vi
Multidim
t-stat on vi
Anti-director rights 1.59 0.64 Market Cap. to GDP 2.25* -0.63
Assessment of tax evasion 3.60* -1.59 Masculinity -0.37 0.03
Audit spending 3.12* 2.00 Media 2.85* 1.38
Bank money in private sector to GDP 1.81 -0.64 Muslim -1.89 -0.76
Big4 market share 3.64* 0.75 Number of analysts 2.64* 2.06*
Block premium -2.40* -0.68 Number of veto-players -0.18 -1.27
Book tax independence 0.28 -0.70 Other religion 0.32 -1.39
Buddhist -0.94 0.09 Ownership concentration -3.33* -1.77
Catholic -0.97 1.00 Political score 3.44* 0.21
Class action lawsuit 0.37 -0.51 Political stability 2.88* -0.78
Control of corruption 4.78*#† -2.30* Politically connected firms -1.22 1.88
Corruption 4.42*#† -0.91 Power distance -2.99* 0.61
Creditor rights 5.23*#† -0.38 Private control of self-dealing index 2.54* 0.36
Democracy 1.67 0.48 Property rights 3.89*#† -2.08*
Domestic institutional holdings -0.66 0.08 Protestant 3.49* 1.12
Enforcement of accounting standards 3.13* 0.87 Public control of self-dealing -1.23 -2.9*
Enforcement of audit standards 5.46*#† 1.98 Public enforcement sec. regulation 1.11 0.94
English proficiency 3.97*#† 0.01 Regulatory quality 4.64*#† -1.12
Ethnic fractionalization -1.30 -0.3 Religion fractionalization 1.67 -1.76
Ex ante private control of self-dealing 1.83 0.79 Religiousness -2.4* 0.79
Ex post private control of self-dealing 2.80* -0.35 Rule of law 4.35*#† -1.84
Foreign institutional holdings 2.55* 1.51 Secrecy -7.90*#† -2.37*
Foreign investment to GDP 1.92 0.35 Securities reg. disclosure requirements 2.65* 2.38*
GDPc 4.38*#† 1.66 Securities reg. liability standards 2.69* 1.94
Hierarchy vs independence -1.76 -0.79 Strengh of securities regulation 2.25* 1.49
Individualism 5.56*#† 4.03*#† Tax compliance 1.80 1.69
Individualism in income -1.01 1.02 Trust 3.56* 1.38
Information and knowledge 2.30* 0.24 Uncertainty avoidance -4.26*#† -2.04
IPOs to GDP 2.38* 0.50 US cross-listing 0.72 2.91*
Judicial efficiency 5.02*#† -0.37 US institutional holdings 0.62 1.97
Judicial independence 5.21*#† -0.86 Nr. significant country variables 45 (15) 8 (1)
Language fractionalization -1.21 -1.04 Table 5 reports t-statistics for the coefficient “c” on an
individual country variable vj in the following regression
models for 35 countries: (i) a regression of reporting quality
factor q on country variable vj (unidimensional model: q= a +
c*vi), and (ii) a regression using four country latent factors
(from a factor analysis that excludes the individual country
variable vi) plus the individual country variable vi
(multidimensional model: q=a+b1*f1+b2*f2+b3*f3
+b4*f4+c*vi). Variable definitions are provided in Appendix
1. The symbol * indicates statistical significance at the 0.05
level, the symbol # indicates statistical significance at the 0.05
level for Bonferroni adjusted t-stats, and the symbol †
indicates statistical significance at the 0.05 level for Holm
adjusted t-stats.
Language proximity to English 3.42* -0.02
Latitude 2.48* 0.65
Law and order 3.21* -0.58
Legal origin 3.52* -0.49
Legislative competition 0.26 -1.31
Listed firms to population 2.25* -1.80
Long-term orientation -1.10 -1.92
Low political risk 3.53* 0.36
Low repudiation of contracts by govern. 3.46* 0.43
Low risk of expropriation by govern. 3.85*#† 1.11
48
Table 6 – Adjusted R2’s from nested regressions of financial reporting factor (q) on country-level variable (vi)
and 4 factors – Model (1): q = a + c*vi , Model (2): q = a+b1*f1+b2*f2+b3*f3 +b4*f4 , Model (3): q =
a+b1*f1+b2*f2+b3*f3 +b4*f4+c*vi
Explanatory country variable (vi) (1) (2) (3) (1) (2) (3)
Anti-director rights 0.04 0.72 0.71 Market Cap. to GDP 0.11 0.72 0.72
Assessment of tax evasion 0.26 0.73 0.74 Masculinity -0.03 0.71 0.70
Audit spending 0.20 0.69 0.72 Media 0.17 0.72 0.73
Bank money in private sector to GDP 0.06 0.71 0.70 Muslim 0.07 0.72 0.71
Big4 market share 0.27 0.71 0.71 Number of analysts 0.15 0.73 0.75 *
Block premium 0.12 0.73 0.72 Number of veto-players -0.03 0.72 0.72
Book tax independence -0.03 0.72 0.72 Other religion -0.03 0.71 0.72
Buddhist 0.00 0.73 0.72 Ownership concentration 0.23 0.72 0.74
Catholic 0.00 0.72 0.72 Political score 0.24 0.72 0.71
Class action lawsuit -0.03 0.72 0.71 Political stability 0.18 0.72 0.72
Control of corruption 0.39 0.72 0.76 * Politically connected firms 0.01 0.74 0.76
Corruption 0.35 0.72 0.72 Power distance 0.19 0.72 0.71
Creditor rights 0.44 0.73 0.72 Private control of self-dealing index 0.14 0.72 0.71
Democracy 0.05 0.72 0.71 Property rights 0.29 0.72 0.75 *
Domestic institutional holdings -0.02 0.71 0.70 Protestant 0.25 0.70 0.71
Enforcement of accounting standards 0.21 0.72 0.72 Public control of self-dealing 0.01 0.72 0.78 *
Enforcement of audit standards 0.46 0.72 0.74 Public enforcement sec. regulation 0.01 0.71 0.71
English proficiency 0.30 0.73 0.72 Regulatory quality 0.38 0.72 0.72
Ethnic fractionalization 0.02 0.72 0.71 Religion fractionalization 0.05 0.72 0.74
Ex ante private control of self-dealing 0.06 0.71 0.71 Religiousness 0.12 0.72 0.71
Ex post private control of self-dealing 0.17 0.72 0.71 Rule of law 0.35 0.72 0.74
Foreign institutional holdings 0.14 0.71 0.72 Secrecy 0.64 0.69 0.73 *
Foreign investment to GDP 0.07 0.73 0.72 Securities reg. disclosure requirements 0.15 0.72 0.75 *
GDPc 0.35 0.72 0.73 Securities reg. liability standards 0.16 0.72 0.74
Hierarchy vs independence 0.06 0.72 0.71 Strengh of securities regulation 0.11 0.71 0.73
Individualism 0.47 0.70 0.80 *#† Tax compliance 0.06 0.72 0.73
Individualism in income 0.00 0.73 0.73 Trust 0.26 0.71 0.72
Information and knowledge 0.11 0.72 0.71 Uncertainty avoidance 0.33 0.70 0.73
IPOs to GDP 0.12 0.72 0.71 US cross-listing -0.01 0.69 0.75 *
Judicial efficiency 0.42 0.72 0.72 US institutional holdings -0.02 0.70 0.73
Judicial independence 0.43 0.72 0.72 Average Adjusted R2 0.16 0.72 0.72
Language fractionalization 0.01 0.72 0.72 Table 6 reports adjusted R2’s for the following regression
models for 35 countries. Model 1: regression of reporting
outcome factor q on country variable vi (q = a + c*vi); model
2: regression of reporting outcome factor q on 4 country
factors (from a factor analysis that excludes the individual
country variable vj) (q=a+b1*f1+b2*f2+b3*f3+b4*f4 ); and
model 3: regression of reporting outcome factor q on four
country factors from a factor analysis that excludes the
individual country variable vi plus the individual country
variable vi on reporting outcome factor q (q =
a+b1*f1+b2*f2+b3*f3+b4*f4+c*vi). Variable definitions are
provided in appendix 1. The symbol * (#) [†]indicate a
statiscally higher adjusted R2 of model 3 (compared to model
2) at the 0.05 level (with Bonferroni adjustment) [with Holm
adjustment].
Language proximity to English 0.24 0.72 0.72
Latitude 0.13 0.72 0.71
Law and order 0.21 0.72 0.71
Legal origin 0.25 0.72 0.72
Legislative competition -0.03 0.73 0.73
Listed firms to population 0.11 0.73 0.75
Long-term orientation 0.01 0.67 0.70
Low political risk 0.25 0.72 0.71
Low repudiation of contracts by govern. 0.24 0.72 0.71
Low risk of expropriation by govern. 0.29 0.72 0.72
49
Table 7 – The relation between country latent factors and financial reporting factor
Panel A: Descriptive statistics by factor-groups
Mean Median
Low Medium High High-
Low
Low Medium High High
- Low
Country factor 1 -0.752 0.149 0.658 *** -0.701 0.258 0.817 ***
Country factor 2 -0.643 -0.080 0.789 *** -0.587 -0.028 0.797 ***
Country factor 3 0.042 -0.171 0.140 -0.210 -0.406 -0.162
Country factor 4 -0.439 0.185 0.278 ** -0.522 0.229 0.215 **
Panel B: Regression analysis of country factors on financial reporting factor
(1) (2) (3) (4)
Country factor 1 0.557*** 0.557*** 0.557*** 0.557***
(5.581) (6.268) (5.969) (6.426)
Country factor 2 0.502*** 0.502*** 0.502***
(4.173) (4.784) (5.849)
Country factor 3 0.165* 0.165*
(1.992) (2.040)
Country factor 4 0.317***
(3.109)
Adjusted R2 0.318 0.588 0.608 0.719
Table 7 reports univariate and regression results. Panel A reports mean and median values of the reporting
outcome factor by terciles of country factors (description of the single financial reporting factor can be
found in Table 2 and description of 4 estimated country factors can be found in Table 4). Panel B reports
estimation results of regressions of four latent country factors on financial reporting outcome factor.
Heteroskedasticity adjusted t-statistics are presented in parentheses. The symbols ***,**,and * indicate
statistical significance at the 0.01, 0.05 and 0.1 level, respectively, based on two tailed tests.
50
Table 8 – Categories of country variables
Panel A – Variables in each category
Economic Geopolitical Sociological Regulatory Legal
Assessment of tax evasion
Audit spending
Bank money in private
sector to GDP
Big4 market share
Block premium
Domestic institutional
holdings
Foreign institutional
holdings
Foreign investment to GDP
GDPc
Information and knowledge
IPOs to GDP
Listed firms to population
Low political risk
Market Cap. to GDP
Media
Number of analysts
Ownership concentration
Political stability
Politically connected firms
Tax compliance
US cross-listing
US institutional holdings
Democracy
Latitude
Legislative competition
Number of veto-players
Buddhist
Catholic
English proficiency
Ethnic fractionalization
Hierarchy vs independence
Individualism
Language fractionalization
Language proximity to English
Long-term orientation
Masculinity
Muslim
Other religion
Power distance
Protestant
Religion fractionalization
Religiousness
Secrecy
Trust
Uncertainty avoidance
Individualism in income
Judicial efficiency
Judicial independence
Law and order
Low repudiation of contracts by govern.
Low risk of expropriation by govern.
Political score
Property rights
Rule of law
Corruption
Anti-director rights
Book tax independence
Class action lawsuit
Control of corruption
Creditor rights
Enforcement of accounting standards
Ex ante private control of self-dealing
Ex post private control of self-dealing
Private control of self-dealing index
Public control of self-dealing
Public enforcement securities regulation
Regulatory quality
Securities Regulation Disclosure
Requirements
Securities regulation liability standards
Strength of securities regulation
Enforcement of audit standards
Legal Origin
Average Maximum Pairwise
Correlation Among
“Economic” Variables
= 0.624
Average Maximum Pairwise
Correlation Among “Geopolitical”
Variables
= 0.641
Average Maximum Pairwise
Correlation Among “Sociological”
Variables
= 0.617
Average Maximum Pairwise Correlation
Among “Regulatory” Variables
= 0.805
51
Panel B – Explaining financial reporting using category-derived country factors
Economic
(1)
Geopolitical
(2)
Sociological
(3)
Regulatory
(4)
Legal
(5)
All categories
(6)
Economic factor1 0.540*** 0.590**
(5.130) (2.498)
Economic factor2 0.407*** 0.125
(4.125) (1.426)
Geopolitical factor 0.225 -0.050
(1.338) (-0.433)
Sociological factor1 -0.250** -0.340**
(-2.095) (-2.117)
Sociological factor2 0.545*** 0.344***
(5.471) (3.410)
Sociological factor3 0.457*** 0.338**
(4.780) (2.484)
Sociological factor4 -0.054 -0.148
(-0.505) (-1.310)
Legal 0.501*** 0.019
(3.431) (0.094)
Regulatory factor1 0.629*** -0.348
(6.369) (-1.338)
Regulatory factor2 0.364*** 0.533***
(3.080) (3.718)
Adjusted R2 0.467 0.027 0.573 0.549 0.251 0.784
52
Panel C – Correlations between category-derived country factors
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
(1) Financial reporting factor 1
(2) Economic factor1 0.563* 1
(3) Economic factor2 0.425* 0 1
(4) Geopolitical factor 0.235 0.479* -0.230 1
(5) Sociological factor1 -0.261 -0.753* 0.089 -0.586* 1
(6) Sociological factor2 0.569* 0.151 0.321 0.175 0 1
(7) Sociological factor3 0.477* 0.406* 0.114 0.102 0 0 1
(8) Sociological factor4 -0.057 -0.014 0.271 -0.160 0 0 0 1
(9) Legal 0.523* -0.065 0.502* 0.027 0.307 0.590* 0.319 0.246 1
(10) Regulatory factor1 0.656* 0.839* 0.325 0.401* -0.615* 0.406* 0.415* 0.219 0.288 1
(11) Regulatory factor2 0.380* -0.337* 0.554* -0.294 0.513* 0.396* 0.018 0.246 0.729* 0 1
Table 8 reports country variables that are pre-classified into 5 categories (economic, geopolitical, sociological, legal and regulatory). Panel A describes the
variables included in each category. Panel B reports estimation results of regressions of financial reporting quality factor on categories of country factors
(derived from factor analysis of pre-classified factors). Heteroskedasticity adjusted t-statistics are presented in parentheses. The symbols ***,**,and *
indicate statistical significance at the 0.01, 0.05 and 0.1 level, respectively, based on two tailed tests. Panel C reports pairwise correlations. The symbol *
indicates statistical significant at 5% level.
53
Table 9 – Out-of-sample tests for a ‘new’ country attribute and a ‘new’ reporting outcome
Panel A: Correlation between new country attribute (economic freedom index) and country factors
Factor 1 Factor 2 Factor 3 Factor 4
0.714*** 0.480** -0.167 0.080
Number of countries = 27
Panel B: Explaining financial reporting quality with a ‘new’ country variable
(Economic freedom index)
(1) (2) (3) (4)
Economic freedom 0.662*** -0.299 -0.205 -0.176
(5.607) (-1.219) (-0.955) (-0.997)
Country factor 1 0.871*** 0.821*** 0.785***
(3.503) (3.636) (4.078)
Country factor 2 0.708*** 0.699*** 0.652***
(3.758) (4.403) (4.878)
Country factor 3 0.322*** 0.303***
(4.289) (3.846)
Country factor 4 0.194**
(2.212)
Adjusted R2 0.416 0.656 0.759 0.792
Panel C: Explaining a ‘new’ financial reporting outcome (MV/E) with country factors
(1) (2) (3) (4) (5) (5)
Factor 1 -0.132 -0.144 -0.118 0.136 0.226
(-1.196) (-1.169) (-0.959) (0.668) (0.972)
Factor 2 -0.663*** -0.654*** -0.687*** -0.630*** -0.455*** -0.376**
(-4.187) (-4.117) (-4.863) (-4.972) (-2.932) (-2.164)
Factor 3 -0.196 -0.159 0.177
(-1.002) (-0.872) (1.135)
Factor 4 -0.355*** 0.034
(-3.341) (0.337)
Economic
freedom -0.034 -0.077
(-0.154) (-0.330)
BV/E 0.664*** 0.799***
(5.375) (3.941)
Adjusted R2 0.399 0.405 0.416 0.532 0.752 0.754
Table 9 reports univariate and regression results. Panel A reports Pearson correlation between the economic freedom index (see
appendix 1 for definition) and country factors derived from factor analysis of 72 country variables. Panel B reports estimation
results of regressions of country factors and economic freedom on the reporting outcome factor. Panel C repors estimation results of
regressions using country factors and economic freedom on the outcome variable MV/E (market-to-earnings). Heteroskedasticity
adjusted t-statistics are presented in parentheses. The symbols ***,**,and * indicate statistical significance at the 0.01, 0.05 and 0.1
level, respectively, based on two tailed tests.
54
Table 10 – Explaining other economic and financial reporting outcomes
Panel A: Outcome variables – IFRS difference and unexpected earnings
IFRS
difference
Unexpected
earnings
Factor 1 -0.168 -0.084
(-1.306) (-0.891)
Factor 2 -0.693*** -0.546***
(-5.572) (-3.830)
Factor 3 -0.015 -0.176**
(-0.162) (-2.251)
Factor 4 -0.197 -0.134
(-1.389) (-0.909)
Adjusted R2 0.486 0.246
Observations 35 29
Panel B: Outcome variable – capital market development
(1) (2)
Reporting outcome factor 0.381** -0.202
(2.655) (-1.140)
Factor 1 excluding market development 0.308**
(2.839)
Factor 2 excluding market development 0.698***
(2.926)
Adjusted R2 0.107 0.347
Observations = 35
Table 10 reports estimation results of regressions of country factors derived from factor analysis and
reporting outcome factor on economic outcomes. Panel A reports estimation results for IFRS
difference (score measuring differences between domestic standards and IFRS in 21 key accounting
items – Bae et al. 2008), and unexpected earnings (average absolute analyst forecast errors for
forecasts of the current year, one and two-years ahead - Dhaliwal et al 2012). Panel B reports
estimation results for capital market development as a proportion of GDP. Heteroskedasticity adjusted
t-statistics are presented in parentheses. The symbols ***,**,and * indicate statistical significance at
the 0.01, 0.05 and 0.1 level, respectively, based on two tailed tests.
55
Figure 1 – Country scores for financial reporting factor
Greece
Pakistan
Taiwan
Portugal
Brazil
Austria
Indonesia
Philippines
Thailand
Chile
Germany
Argentina
Spain
Malaysia
Italy
Belgium
Japan
South Africa
India
Mexico
Israel
France
Switzerland
Singapore
Denmark
Hong Kong
Australia
Finland
New Zealand
Netherlands
Canada
Sweden
United StatesUnited Kingdom
Ireland
-1.75 -1.50 -1.25 -1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 2.50
56
Figure 2 – Country scores for factor 1
PakistanPhilippines
IndonesiaIndia
ThailandMalaysia
BrazilMexico
ArgentinaSouth Africa
GreeceTaiwan
IsraelChile
SingaporePortugalItaly
SpainUnited States
Hong KongCanadaFrance
BelgiumAustralia
JapanUnited Kingdom
IrelandNew Zealand
GermanyAustria
NetherlandsDenmarkSwitzerlandSweden
Finland
-2.25 -2 -1.75 -1.5 -1.25 -1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1 1.25 1.5 1.75
57
Appendix 1 – Variable definitions and data sources
Panel A – Financial reporting variables
Variable name Brief description Source References
Abnormal return Abnormal returns over the three-day announcement
window around annual earnings announcements
Nguyen and Truong [2013] Core et al. [2015], Pevnzer et al. [2015], Wang
[2014], Landsman et al. [2012], Choy and Zhang
[2011], Armstrong et al. [2010], DeFond et al.
[2007], Lang et al. [2004]
Abnormal volume Abnormal trading volume over the three-day
announcement window around annual earnings
announcements
Nguyen and Truong [2013] Pevnzer et al. [2015], Wang [2014], Landsman et al.
[2012]
Return synchronicity Weighted average R2 of regressions of firm stock
returns around earnings announcement on the
country’s market return, multiplied by minus one
Nguyen and Truong [2013] Karolyi et al. [2012], Brochet et al. [2016], DeFond et
al. [2007]
Disclosure quality Center for Financial Analysis and Research index of
disclosure practices in the country. The index based
on the disclosure of 90 items in firms’ 1995 annual
reports
Bushman et al. [2004] Pevnzer et al. [2015], Nanda and Wysocki [2015],
Nguyen and Truong [2013], Karolyi et al. [2012],
Dhaliwal et al. [2012], Landsman et al. [2012], Kim
and Shi [2012], Choy and Zhang [2011], Hope et al.
[2008], Bushman et al. [2004], Francis et al. [2005],
Bhattacharya et al. [2003], Hope [2003], DeFond et
al. [2007], Dyck and Zingales [2004], Doidge et al.
[2004], La Porta et al. [1998]
Reporting transparency Aggregate score of the following four earnings
management metrics multiplied by minus one: 1)
earnings smoothing 2) the correlation between
accounting accruals and operating cash flows, 3)
magnitude of accruals, and 4) small loss avoidance
Leuz et al. [2003], Leuz [2010] Pevnzer et al. [2015], Srinivasan et al. [2015], Nanda
and Wysocki [2015], Choy and Zhang [2011],
DeFond et al. [2011], Daske et al. [2008], Doupnik
[2008], Francis and Wang [2008], DeFond et al.
[2007], Burgstahler et al. [2006], Lang et al. [ 2006],
Riahi-Belkaoui [2004a], Riahi-Belkaoui [2004b],
Bhattacharya et al. [2003], Land and Lang [2002],
Hung [2001]
Asymmetric timeliness Average country-level association between firms’
earnings and negative stock returns
Bushman and Piotroski [2006] Christensen et al. [2015], Nanda and Wysocki [2015],
Barth et al. [2012], Francis and Wang [2008], Barth
et al. [2008], Lang et al. [2006], Ball et al. [2003],
Ball et al. [2000], Basu [1977]
58
Appendix 1 – Variable definitions and data sources (cont.)
Panel B – Country variables
Variable name Short name Brief description Source References
Anti-director rights AntiDir Revised anti-director rights index Djankov et al. [2008],
La Porta et al. [1998]
Chen et al. [2015], Nguyen and Truong [2013], Kim and Shi [2012], Hope
et al. [2011], Leuz [2010], Francis and Wang [2008], Frost et al. [2006],
Bushman et al. [2004], Dyck and Zingales [2004], Doidge et al. [2004],
Haw et al. [2004], Hope [2003], Leuz et al. [2003], Hung [2001]
Assessment of tax
evasion
LTaxEv Score of low prevalence of tax evasion Djankov et al. [2008],
La Porta et al. [1998]
Dyck and Zingales [2004]
Audit spending AuditSpend Fees of country's ten largest accountig firms
as percentage of GDP
Mueller, Gernon and
Meek [1994]
Hope [2003], Ali and Hwang [2000]
Bank money in
private sector to GDP
BankPriv Claims of deposit money banks on private
sector to GDP
Beck, Demirguç-Kunt
and Levine, [2001]
Beck et al. [2003], Bushman and Piotroski [2006], Bushman et al. [2004]
Big4 market share Big4Share Market share of big 4 audit firms Francis and Wang
[2008]
Nanda and Wysocki [2015], Hope et al. [2008]
Block premium BlockPr Difference between price paid by the control
block and market price
Djankov et al. [2008] Dyck and Zingales [2004]
Book tax
independence
BTaxInd Required book-tax conformity multiplied by
minus one
Li et al. [2016] Ali and Hwang [2000]
Buddhist Budhist Percentage of population that are buddhist Stulz and Williamson
[2003],
CIA Worldfact book
[2003, 2010]
Pevnzer et al. [2015]
Catholic Catholic Percentage of population that are catholic Stulz and Williamson
[2003],
CIA Worldfact book
[2003, 2010]
Pevnzer et al. [2015], Siegel et al. [2011], Dyck and Zingales [2004], Beck
et al. [2003]
Class action lawsuit ClassAction Indicator if class-action lawsuit is available Leuz [2010]
Control of corruption CCorr Score of control of corruption (average 1995-
2013)
Kaufmann, Kraay and
Mastuzzi.
Worldwide
Governance Indicators
Pevnzer et al. [2015], Dhaliwal et al. [2012], Armstrong et al [2010],
Riahi-Belkaoui[2004b], Leuz et al. [2003], La Porta et al. [1998]
Corruption CCorrL Effectiveness of control of corruption La Porta et al. [1998] Dang et al. [2016], Nanda and Wysocki [2015], Dhaliwal et al. [2012],
Karoli et al [2012], Kim and Shi [2012], Kanagaretnam et al. [2011], Kim
et al. [2011], Gupta et al. [2008], Hope et al. [2008], Hope [2003], Leuz et
al. [2003], Morck et al. [2000]
Creditor rights CreditR Legal protection of creditors and borrowers World Economic
Forum
Chen et al. [2015], Kanagaretnam et al. [2011], Kim et al. [2011]
Democracy Democracy Democracy score (autocracy multiplied by
minus one)
Boutchkova et al.
[2012]
Bushman et al. [2004]
Domestic institutional
holdings
InstHoldDom Holdings by domestic instutitional investors
as a proportion of firms’ market value
Covrig, DeFond and
Hung [2007]
Florou and Pope [2012], Callen et al. [2005]
59
Economic freedom EcoFreed Economic freedom index Chen, Chen and Jin
[2015]
Riahi-Belkaoui [2004b]
Enforcement of
accounting standards
EnforAccS Score of accounting and market enforcement
(average 2002, 2005, 2008)
Brown, Prieato and
Tarca [2014]
Christensen et al. [2013], Alford et al. [2003]
Enforcement of audit
standards
EnforAudS Score of auditing enforcement (average
2002, 2005, 2008)
Brown, Prieato and
Tarca [2014]
English proficiency EnglProf Score for the speaking portion of the TOEFL
exam
Brochet, Naranjo and
Yu [2016]
Ethnic
fractionalization
EthFract One minus the Herfindhal index of ethnicity Alesina et al. [2003] Beck et al. [2003], Siegel et al. [2011]
Ex ante private
control of self-dealing
PrivCtrEA Index of ex-ante private control of self-
dealing
Djankov et al. [2008] Leuz [2010],
Ex post private
control of self-dealing
PrivCtrEP Index of ex-post private control of self-
dealing
Djankov et al. [2008] Leuz [2010]
Foreign institutional
holdings
IntHoldFor Percentage of holdings by foreign
instutitional investors
Covrig, DeFond and
Hung [2007]
Yu and Wahid [2014], Karolyi et al. [2012], DeFond et al. [2011], Florou
and Pope [2012], Callen et al. [2005]
Foreign investment to
GDP
ForeignInv Net inflows of investment to acquire 10
percent or more of voting stock in a foreign
enterprise (average 1995-2013)
World Bank
Development
Indicators
Bushman et al. [2004]
GDPc Gdpc GDP per capita (average 1995-2013) World Bank
Development
Indicators
Chen et al. [2015], Core et al. [2014], Srinivasan et al. [2015], Christensen
et al. [2013], Boutchkova et al. [2012], Karolyi et al. [2012], DeFond et al.
[2011], Kanagaretnam et al. [2011], Kim et al. [2011], Lang et al [2010],
Siegel et al. [2011], Leuz [2010], Djankov et al. [2008], Hope et al. [2008],
Bushman et al. [2004], Dyck and Zingales [2004], Doidge et al. [2004],
Riahi-Belkaoui [2004b], Bhattacharya et al. [2003], Leuz et al. [2003],
Land et al. [2002], La Porta et al. [1998]
Hierarchy vs
independence
HierIndep 100+%(should follow instructions)-
%(must be convinced first)
World Values Survey Pevzner et al. [2015]
Individualism IndividH Hosftede individualism score Hofstede [2010] Pevnzer et al. [2015], Nanda and Wysocki [2015], Nguyen and Truong
[2013], Kanagaretnam et al. [2011], Han el al. [2010], Doupnik [2008],
Hope et al. [2008], Ding et al. [2005], Hope [2003]
Individualism in
income
IndividW Index equal to 100+%(completly agree we
need large income difference)-%(completly
agree with income should be equal)
World Values Survey Pevzner et al. [2015]
Information and
knowledge
InfoKnow Score of information and knowledge based
on 9 dimensions (average 2002-2012)
Global Democracy
Rankings
Bushman et al. [2004]
IPOs to GDP IPO Ratio of the equity issued by newly listed
firms to its GDP
Djankov et al. [2008],
La Porta et al. [2006]
Leuz et al. [2003]
Judicial efficiency JudEff Score of judicial efficiency La Porta et al. [1998] Nanda and Wysocki [2015], Kanagaretnam et al. [2013], Dhaliwal et al.
[2012], Kim and Shi [2012], Kanagaretnam et al. [2011], Kim et al.
[2011], Gupta et al. [2008], Hope et al. [2008], Bushman and Piotroski
[2006], Bushman et al. [2004], Dyck and Zingales [2004], Doidge et al.
[2004], Haw et al. [2004], Hope [2003], Leuz et al. [2003]
60
Judicial independence JudIndep Score of judicial independence World Economic
Forum
Bushman and Piotroski [2006]
Language
fractionalization
LangFract One minus the Herfindhal index of language
measure
Alesina et al. [2003] Siegel et al. [2011]
Language proximity
to English
EnglProx The distance between English and the main
language based on a 5-point classification
system, multiplied by minus one
Brochet, Naranjo and
Yu [2016]
DeFond et al. [2011], Frost et al. [2006]
Latitude Latitude Geographic latitude La Porta et al. [1999] Siegel et al. [2011], Beck et al. [2003]
Law and order LawO Score of law and order La Porta et al. [1998] Chen et al. [2015], Brochet et al.[2016], Core et al. [2014], Nanda and
Wysocki [2015], Kanagaretnam et al. [2013], Nguyen and Truong [2013],
Dhaliwal et al. [2012], Kim and Shi [2012], Hope et al. [2011],
Kanagaretnam et al. [2011], Kim et al. [2011], Gupta et al. [2008], Hope et
al. [2008], Hail and Leuz [2006], Frost et al. [2006], Brown and Higgins
[2005], Riahi-Belkhaoui [2004a], Dyck and Zingales [2004], Haw et al.
[2004], Hope [2003], Leuz et al. [2003], Beck et al. [2003]
Legal origin LegalO Indicator of legal origin: Common,
Civil/French,Civil/German,Civil/Scandinavia
Stulz and Williamson
[2003],
La Porta et al. [2006]
Dang et al. [2015], Pevnzer et al. [2015], Chen et al. [2015], Siegel et al.
[2011], Armstrong et al [2010], Leuz [2010], Francis and Wang [2008],
Bushman and Piotroski [2006], Bushman et al. [2004], Doidge et al.
[2004], Dyck and Zingales [2004], Haw et al. [2004], Lang et al. [2004],
Beck et al. [2003], Bhattacharya et al. [2003], Lang et al. [2003a], Leuz et
al. [2003], Beck et al. [2001], Hung [2001], Ball et al. [2000]
Legislative
competition
LegislComp Index of the number of parties competing in
legislative elections
Beck, Demirguç-Kunt
and Levine [2003]
Beck et al. [2001]
Listed firms to
population
ListedF Listed firms per 1000 population World bank
Development
Indicators
Frost et al.[2006], Leuz et al. [2003]
Long-term orientation LTorient Hosftede long-term orientation score Hofstede [2010] Doupnik [2008]
Low political risk LPolitR Index of political risk multiplied by minus
one
Boutchkova et al.
[2012]
Bushman and Piotroski [2006]
Low repudiation of
contracts by govern.
LRepContr Score of repudiation of contracts by the
government multiplied by minus one
La Porta et al. [1998] Dang et al. [2016], Nguyen and Truong [2013], Kim and Shi [2012],
Karoli et al [2012], Kanagaretnam et al. [2011], Kim et al. [2011], Morck
et al. [2000]
Low risk of
expropriation by
govern.
LExprR Score of expropriation by the government
multiplied by minus one
La Porta et al. [1998] Dang et al. [2016], Kanagaretnam et al. [2013], Nguyen and Truong
[2013], Karoli et al [2012], Kim and Shi [2012], Kanagaretnam et al.
[2011], Kim et al. [2011], Bushman and Piotroski [2006], Bushman et al.
[2004], Morck et al. [2000]
Market Cap. to GDP MarkCap Market capitalization to GDP (average 1995-
2013)
World Bank
Development
Indicators
Brochet et al. [2016], Karolyi et al. [2012], DeFond et al. [2011], Djankov
et al. [2008], Frost et al. [2006], La Porta et al. [2006], Bushman et al.
[2004], Dyck and Zingales [2004], Leuz et al. [2003], La Porta [1997]
Masculinity Masculin Hosftede masculinity score Hofstede [2010] Pevnzer et al. [2015], Kanagaretnam et al. [2011], Han el al. [2010],
Doupnik [2008], Ding et al. [2005], Hope [2003]
Media Media Average rank of the media development
(print and television)
Bushman et al. [2004] Kanagaretnam et al. [2013], Frost et al. [2006], Dyck and Zingales [2004],
Haw et al. [2004]
61
Muslim Muslim Percentage of population that are muslim Stulz and Williamson
[2003],
CIA Worldfact book
[2003, 2010]
Pevnzer et al. [2015], Beck et al. [2003]
Number of analysts NrAnal Number of financial analysts follow firms Bae, Tan and Welker
[2008]
Pevnzer et al. [2015], Srinivasan et al. [2015], Choy and Zhang [2011],
Landsman et al. [2012], Covrig et al. [2007], DeFond et al. [2007], Frost et
al. [2006], Bushman et al. [2004], Lang et al. [2004], Hope [2003], Land
and Lang [2002]
Number of veto-
players
Veto Number of veto-players in the political
decision-making process
Beck et al. [2003] Li et al. [2016], Beck et al. [2001]
Other religion OReligion Percentage of population from religions
other than catholic, protestant, buddhist, or
muslim
Stulz and Williamson
[2003], La Porta,
Lopez-de-Silanes and
Schleifer [2006]
Pevnzer et al. [2015], Beck et al. [2003]
Ownership
concentration
OwnConc Percentage of common shares owned by top
three shareholders in the 10 largest firms
Djankov et al. [2008],
La Porta et al. [2006]
Chen et al. [2015], Boutchkova et al. [2012], Hope et al. [2008],
Burgstahler et al. [2006],Bushman and Piotroski [2006], Bushman et al.
[2004], Dyck and Zingales [2004], Leuz et al. [2003]
Political score PolitScore Index of political quality based on 8
dimensions(average 2002-2012)
Global Democracy
Rankings
Boutchkova et al. [2012]
Political stability PolitStab Score of political stability (average 1995-
2013)
Kaufmann, Kraay and
Mastuzzi
Worldwide
Governance Indicators
Boutchkova et al. [2012]
Politically connected
firms
PolitConn Percentage of firms connected to polititians Faccio [2006] Batta et al. [2014], Riahi-Belkhaoui [2004b]
Power distance PowerD Hosftede power distance score Hofstede [2010] Pevnzer et al. [2015], Nanda and Wysocki [2015], Kanagaretnam et al.
[2011], Han et al. [2010], Doupnik [2008], Hope et al. [2008], Ding et al.
[2005], Hope [2003]
Private control of self-
dealing index
PrivCtrIndex Index of ex-ante and ex-post private control
of self-dealing
Djankov et al. [2008] Pevnzer et al. [2015], Siegel et al. [2011], Leuz [2010], Bushman and
Piotroski [2006]
Property rights ProprR Score of property rights World Economic
Forum
Li et al. [2016], Bushman et al. [2004], Beck et al. [2003], La Porta et al.
[1999]
Protestant Protestant Percentage of population that is protestant Stulz and Williamson
[2003],
La Porta et al. [2006]
Pevnzer et al. [2015], Siegel et al. [2011]
Public control of self-
dealing
PublCtr Index of public enforcement of anti self-
dealing
Djankov et al. [2008] Leuz [2010]
Public enforcement
securities regulation
PublicEnf Index of public enforcement (La Porta) La Porta et al. [2006] Dhaliwal et al. [2012], Leuz [2010], Francis and Wang [2008], Bushman
and Piotroski [2006]
Regulatory quality RegQ Score of regulatory quality Kaufmann, Kraay and
Mastuzzi. Worldwide
Governance Indicators
Dang et al. [2015], Christensen et al. [2013], Leuz et al. [2013]
Religion
fractionalization
ReligFract One minus the Herfindhal index of religion
measure
Alesina et al. [2003] Siegel et al. [2011]
62
Religiousness Religness Principal component of religious attendance
and importance of religion in life
World Values Survey McGuire et al. [2012]
Rule of law Law Score of the quality of the rule of law
(average 1995-2013)
Kaufmann, Kraay and
Mastuzzi.
Worldwide
Governance Indicators
Srinivasan et al. [2015], Nanda and Wysocki [2015], Landsman, Maydew
and Thornock [2012], Armstrong et al [2010], Byard et al. [2010], Daske
et al. [2008]
Secrecy Secrecy Uncertainty avoidance plus power distance
minus individualism
Hope et al. [2008] Nanda and Wysocki [2015]
Securities regulation
disclosure
requirements
DisclReg Index of disclosure requirements by security
laws
La Porta et al. [2006] Pevnzer et al. [2015], Core et al. [2014], Nanda and Wysocki [2015],
Kanagaretnam et al. [2011], Leuz [2010], Francis and Wang [2008], Hail
and Leuz [2006], Hope at al. [2006], Haw et al. [2004], Leuz et al. [2003]
Securities regulation
liability standards
LibStand Index of the procedural difficulty in
recovering losses
La Porta et al. [2006] Srinivasan et al. [2015], Leuz [2010], Francis and Wang [2008]
Strengh of securities
regulation
SSecRegul The strength of securities regulation
mandating and enforcing disclosures. Mean
of disclosure requirements index, liability
standard index, and public enforcement
index
Hail and Leuz [2006] Bhattacharya et al. [2003]
Tax compliance TaxComp Tax avoidance spread multiplied by minus
one
Li et al. [2016] Dyck and Zingales [2004], Haw et al. [2004]
Trust Trust Index equal to 100+%(most people can be
trusted)-%(can't be too careful)
World Values Survey Pevzner et al. [2015], Nanda and Wysocki [2015], La Porta et al. [2006]
Uncertainty avoidance UncAvoid Hosftede uncertainty avoidance score Hofstede [2010] Pevnzer et al. [2015], Brochet et al. [2016], Nanda and Wysocki [2015],
Nguyen and Truong [2013], Kanagaretnam, et al. [2011], Han el al.
[2010], Doupnik [2008], Ding et al. [2005], Hope [2003]
US cross-listing USlisting Percentage of American Depositary Receipt
(ADR) trading on a U.S. exchange
Bradshaw, Bushee
and Miller [2004]
Daske et al. [2013], Hope et al. [2013], DeFond et al. [2011], Hail and
Leuz [2009], Daske et al. [2008], DeFond et al. [2007], Lang et al. [2004] ,
Lang et al. [2003a], Lang et al. [2003b]
US institutional
holdings
UShold Total market value of shares owned by U.S.
institutions as a proportion as firms’ market
value
Bradshaw, Bushee
and Miller [2004]
Srinivasan et al. [2015], Florou and Pope [2012], Callen et al. [2005]
63
Appendix 2 – Correlations between country characteristics
Panel A: Correlations between variables
Variable name Short name Absolute correlation >= 0.7 0.5 =< Absolute correlation < 0.7 0.3 =< Absolute correlation < 0.5
GDPc Gdpc
Law, LRepContr, LawO, LExprR, CCorr,
LPolitR, RegQ, CCorrL, Media, PolitStab,
JudEff, ProprR, EnforAudS, Religness,
InfoKnow, PolitScore, JudIndep, Secrecy
IndividH, Latitude, PowerD, Trust,
LTaxEv, BankPriv, EnforAccS
IntHoldFor, CreditR, EthFract, Protestant, NrAnal, IPO,
Democracy, OwnConc, Muslim, TaxComp, LangFract,
PolitConn, IndividW, Big4Share, ListedF, LTorient,
EnglProx, BlockPr
Rule of law Law
CCorr, RegQ, ProprR, LawO, LPolitR,
LRepContr, LExprR, Gdpc, JudIndep,
CCorrL, PolitStab, JudEff, PolitScore,
Religness, LTaxEv, EnforAudS
Secrecy, Media, InfoKnow, IndividH,
Latitude, PowerD, CreditR, Trust,
BankPriv, EthFract, IntHoldFor
Muslim, Protestant, IndividW, Democracy, TaxComp,
EnforAccS, PolitConn, ListedF, IPO, OwnConc,
LangFract, NrAnal, PrivCtrEP, BlockPr, EnglProx,
Big4Share, MarkCap, ForeignInv, ReligFract
Regulatory quality RegQ
CCorr, Law, ProprR, LPolitR, LRepContr,
LawO, PolitStab, Gdpc, CCorrL, LExprR,
JudIndep, JudEff, PolitScore, LTaxEv
Religness, EnforAudS, Media, Secrecy,
InfoKnow, CreditR, BankPriv, PowerD,
IndividH, Muslim, Latitude, ListedF,
Trust
IntHoldFor, EthFract, EnforAccS, IPO, Protestant,
ForeignInv, TaxComp, IndividW, MarkCap, LangFract,
Democracy, Big4Share, PolitConn, PrivCtrEP, NrAnal,
EnglProx, ReligFract, OwnConc
Control of corruption CCorr
Law, RegQ, ProprR, LPolitR, CCorrL,
LawO, JudIndep, PolitStab, Gdpc,
LRepContr, LExprR, JudEff, PolitScore,
Religness, LTaxEv, Secrecy
EnforAudS, Media, InfoKnow, PowerD,
CreditR, IndividH, Latitude, Trust,
Protestant, BankPriv, IntHoldFor
Muslim, EthFract, ListedF, IndividW, PolitConn,
TaxComp, EnforAccS, Big4Share, LangFract, Democracy,
OwnConc, NrAnal, MarkCap, ForeignInv, IPO, EnglProx,
PrivCtrEP, ReligFract, BlockPr, UncAvoid, EnglProf
Law and order LawO
LPolitR, PolitStab, LExprR, LRepContr,
Law, CCorr, Gdpc, RegQ, CCorrL,
ProprR, PolitScore, Religness, JudEff,
JudIndep
Media, EnforAudS, IndividH, InfoKnow,
Latitude, Secrecy, Trust, LTaxEv,
BankPriv, PowerD, Muslim
LangFract, NrAnal, EnforAccS, Democracy, Protestant,
EthFract, IntHoldFor, OwnConc, IPO, CreditR, TaxComp,
Big4Share, IndividW, PolitConn, UShold, ReligFract
Judicial efficiency JudEff
CCorr, Law, JudIndep, CCorrL, ProprR,
Gdpc, RegQ, CreditR, LExprR, LawO,
Religness, LRepContr
Secrecy, LPolitR, PolitStab, EnforAudS,
LTaxEv, IndividH, PolitScore, InfoKnow,
Media, PowerD, ListedF
BankPriv, Latitude, Trust, Protestant, IndividW, EnglProx,
PolitConn, PrivCtrEP, EnforAccS, Big4Share, EthFract,
OwnConc, EnglProf, ReligFract, MarkCap, TaxComp,
LegalO, IPO, UncAvoid, Muslim, ForeignInv, Democracy,
LangFract, OReligion
Corruption CCorrL
CCorr, Law, ProprR, LPolitR, RegQ,
LRepContr, JudEff, LawO, JudIndep,
Gdpc, PolitStab, LExprR, PolitScore,
EnforAudS, Religness
IndividH, Secrecy, Media, Latitude,
PowerD, CreditR, InfoKnow, LTaxEv,
Protestant, BankPriv, Muslim,
Democracy
PolitConn, Trust, IndividW, EnforAccS, NrAnal, ListedF,
ReligFract, Big4Share, LangFract, MarkCap, IntHoldFor,
TaxComp, EthFract, IPO, OwnConc, BlockPr, EnglProx
Judicial independence JudIndep
ProprR, CCorr, Law, RegQ, JudEff,
CCorrL, LRepContr, CreditR, LExprR,
Secrecy, Gdpc, LTaxEv, LawO
LPolitR, PolitStab, PolitScore, Religness,
EnforAudS, Trust, PowerD, BankPriv,
IndividH, Protestant, InfoKnow, Media,
LegalO
Latitude, EnglProx, IntHoldFor, PrivCtrEP, ListedF,
EnforAccS, UncAvoid, OwnConc, EnglProf, IndividW,
MarkCap, ReligFract, Catholic, Democracy, TaxComp,
BlockPr, PolitConn, PrivCtrIndex, EthFract, ForeignInv,
Muslim, Big4Share
Property rights ProprR
JudIndep, Law, CCorr, RegQ, LRepContr,
CCorrL, LExprR, JudEff, PolitStab,
LPolitR, LawO, Gdpc, LTaxEv
Religness, CreditR, PolitScore,
EnforAudS, BankPriv, Secrecy, Media,
Trust, IndividH, InfoKnow
Protestant, PowerD, IntHoldFor, MarkCap, Latitude,
ListedF, IndividW, ReligFract, IPO, PrivCtrEP, EnglProx,
OwnConc, Muslim, PolitConn, EnforAccS, ForeignInv,
TaxComp, EthFract, UncAvoid, BlockPr, Democracy,
LegalO, LTorient, Big4Share, UShold, EnglProf
64
Panel A: Correlations between variables (cont)
Bank money in private
sector to GDP BankPriv
LRepContr, MarkCap, ProprR, LTaxEv,
JudIndep, RegQ, Law, CCorr, LExprR,
Gdpc, CreditR, Religness, LawO, CCorrL,
LTorient
JudEff, EnforAudS, LPolitR, PolitStab, Media,
ReligFract, ListedF, InfoKnow, IndividW, EnforAccS,
IPO, TaxComp, USlisting, BlockPr, IntHoldFor, Trust,
LibStand, OwnConc, PolitScore
Low repudiation of
contracts by gov. LRepContr
LExprR, LawO, Law, Gdpc, ProprR,
RegQ, LPolitR, CCorr, PolitStab, CCorrL,
JudIndep, EnforAudS, Religness,
PolitScore, JudEff
Media, BankPriv, InfoKnow, Secrecy,
IndividH, LTaxEv, Trust, Latitude,
PowerD, EnforAccS, CreditR
IntHoldFor, Muslim, Democracy, LTorient, TaxComp,
NrAnal, OwnConc, IPO, EthFract, Protestant, LangFract,
IndividW, PrivCtrEP, Big4Share, ReligFract, UShold,
BlockPr, ListedF
Low risk of
expropriation by gov. LExprR
LRepContr, LawO, Law, Gdpc, CCorr,
RegQ, LPolitR, ProprR, CCorrL, PolitStab,
JudIndep, EnforAudS, PolitScore, JudEff,
Religness
Media, IndividH, Secrecy, InfoKnow,
Latitude, LTaxEv, Trust, BankPriv,
PowerD, IntHoldFor, NrAnal, EnforAccS
CreditR, Democracy, Muslim, OwnConc, LangFract,
Big4Share, TaxComp, Protestant, LTorient, EthFract,
IPO, IndividW, PrivCtrEP, ReligFract
Assessment of tax
evasion LTaxEv
RegQ, ProprR, CCorr, CreditR, JudIndep,
Law
JudEff, PrivCtrEP, BankPriv, LRepContr,
CCorrL, LExprR, ListedF, Gdpc, LawO,
ReligFract, PolitStab, LegalO, MarkCap,
LPolitR, Secrecy
PrivCtrIndex, EnforAudS, LibStand, Big4Share,
OwnConc, UncAvoid, BlockPr, Religness, InfoKnow,
Catholic, Media, PowerD, Trust, IntHoldFor, IPO,
DisclReg, EnglProx, EnforAccS, ForeignInv, AntiDir,
EnglProf, Protestant, PrivCtrEA, TaxComp, SSecRegul,
PolitScore
Low political risk LPolitR
PolitStab, LawO, Law, CCorr, RegQ,
LRepContr, Gdpc, CCorrL, LExprR,
ProprR, PolitScore, Religness
JudEff, JudIndep, Media, Muslim,
Latitude, EnforAudS, IndividH,
InfoKnow, Secrecy, LangFract, EthFract,
Democracy, LTaxEv
PowerD, BankPriv, Protestant, Trust, TaxComp,
IntHoldFor, IndividW, IPO, PolitConn, NrAnal, CreditR,
EnforAccS, Big4Share, OwnConc
Political stability PolitStab
LPolitR, LawO, CCorr, Law, RegQ,
LRepContr, CCorrL, ProprR, Gdpc,
LExprR, PolitScore
Religness, JudEff, JudIndep, Media,
Muslim, LangFract, Latitude, EnforAudS,
LTaxEv, EthFract, InfoKnow
IndividH, Secrecy, BankPriv, TaxComp, Trust,
Democracy, Protestant, PowerD, IndividW, IPO,
Big4Share, IntHoldFor, PolitConn, CreditR, ForeignInv,
NrAnal, UShold, MarkCap
Foreign institutional
holdings IntHoldFor LExprR, Latitude, CCorr, Law
RegQ, PolitScore, Gdpc, LRepContr, ProprR, InfoKnow,
JudIndep, LPolitR, LawO, Trust, LTaxEv, PolitStab,
CCorrL, Secrecy, Religness, BankPriv, NrAnal, PowerD,
Media, EnforAudS, TaxComp, Protestant, IndividH,
EthFract
Information and
knowledge InfoKnow Gdpc
Religness, CCorr, Law, PowerD, LawO,
LRepContr, RegQ, LExprR, CCorrL,
JudEff, LPolitR, Latitude, EnforAudS,
PolitScore, Media, JudIndep, PolitStab,
Secrecy, Trust, ProprR
EthFract, LangFract, IntHoldFor, Protestant, IndividH,
LTaxEv, BankPriv, TaxComp, PolitConn, EnforAccS,
OwnConc
Media Media Gdpc
LRepContr, LawO, LExprR, LPolitR,
RegQ, Law, CCorr, CCorrL, PolitStab,
EnforAudS, Religness, ProprR,
PolitScore, JudEff, InfoKnow, Latitude,
JudIndep
IPO, NrAnal, IndividH, Secrecy, BankPriv, Trust,
PowerD, EnforAccS, EthFract, LTaxEv, LangFract,
Muslim, TaxComp, ListedF, Protestant, Democracy,
IntHoldFor, OwnConc, MarkCap, LTorient
65
Panel A: Correlations between variables (cont)
Protestant Protestant
Trust, Secrecy, CCorr, JudIndep,
PolitScore, CCorrL, PowerD, IndividH
Law, Latitude, Religness, ProprR, LPolitR, Gdpc, LawO,
JudEff, RegQ, InfoKnow, PolitStab, UncAvoid, Masculin,
LRepContr, EnforAudS, LExprR, Catholic, Media,
CreditR, OwnConc, IntHoldFor, Big4Share, LTaxEv
Religiousness Religness
PolitScore, CCorr, Law, Gdpc,
LawO, LRepContr, CCorrL, JudEff,
LExprR, LPolitR
RegQ, PolitStab, ProprR, InfoKnow,
JudIndep, EthFract, Media, Latitude,
Trust, EnforAudS, PowerD, Democracy,
IndividH, BankPriv
PolitConn, IndividW, Secrecy, Protestant, Muslim,
LangFract, TaxComp, OwnConc, LTorient, LTaxEv,
CreditR, ListedF, IntHoldFor, BlockPr, PublicEnf
Trust Trust
Protestant, Secrecy, JudIndep, Gdpc,
PowerD, Religness, CCorr, LawO, Law,
LExprR, LRepContr, Latitude, PolitScore,
ProprR, InfoKnow, RegQ
IndividH, CCorrL, LPolitR, JudEff, PolitStab,
EnforAudS, Media, Catholic, UncAvoid, IntHoldFor,
TaxComp, LTaxEv, OwnConc, BlockPr, CreditR,
BankPriv, Big4Share, Masculin
Creditor rights CreditR JudIndep, LegalO, JudEff, LTaxEv
ProprR, PrivCtrIndex, Secrecy,
PrivCtrEP, ListedF, UncAvoid, CCorr,
CCorrL, EnforAudS, RegQ, Law,
EnglProf, EnglProx, MarkCap,
PrivCtrEA, BankPriv, AntiDir,
LRepContr
LibStand, ReligFract, DisclReg, LExprR, Catholic, Gdpc,
BlockPr, EnforAccS, ForeignInv, IndividH, Religness,
LawO, LPolitR, Trust, Big4Share, OReligion, PolitStab,
PowerD, SSecRegul, OwnConc, Protestant, TaxComp,
IPO
Market cap. to GDP MarkCap
ListedF, BankPriv, ForeignInv, CreditR,
LibStand, LTaxEv, IPO, DisclReg
ProprR, PrivCtrIndex, SSecRegul, UncAvoid, RegQ,
PrivCtrEA, JudEff, BlockPr, JudIndep, LegalO,
ReligFract, CCorrL, PrivCtrEP, CCorr, AntiDir,
OReligion, EnforAudS, EnglProx, Big4Share, Catholic,
Law, Media, PolitStab, EnglProf
Listed firms to
population ListedF
MarkCap, CreditR, ForeignInv, LTaxEv,
OReligion, PrivCtrEP, PrivCtrIndex,
RegQ, JudEff
IPO, LegalO, EnforAudS, SSecRegul, CCorr, JudIndep,
ProprR, LibStand, Law, BankPriv, DisclReg, PrivCtrEA,
CCorrL, PublicEnf, UncAvoid, EnglProf, AntiDir,
EnforAccS, Religness, Media, Gdpc, Catholic, BlockPr,
LRepContr
Anti-director rights AntiDir
PrivCtrIndex, PrivCtrEA, LegalO,
UncAvoid, CreditR
LibStand, DisclReg, PrivCtrEP, Latitude, SSecRegul,
MarkCap, ListedF, LTaxEv, Catholic
Ex ante private control
of self-dealing PrivCtrEA PrivCtrIndex
LegalO, AntiDir, CreditR, LibStand,
DisclReg, PrivCtrEP, Latitude
UncAvoid, PolitConn, MarkCap, SSecRegul, ListedF,
Catholic, ForeignInv, PublicEnf, LTaxEv, BlockPr,
EnglProf, Democracy
Ex post private control
of self-dealing PrivCtrEP PrivCtrIndex
LegalO, LibStand, DisclReg, CreditR,
LTaxEv, SSecRegul, ListedF, PrivCtrEA
BlockPr, JudIndep, AntiDir, JudEff, OReligion, EnglProf,
OwnConc, ProprR, Catholic, EnforAudS, Law,
UncAvoid, MarkCap, RegQ, PublicEnf, EnforAccS,
CCorr, LExprR, ForeignInv, LRepContr, EnglProx
Private control of self-
dealing index PrivCtrIndex PrivCtrEA, PrivCtrEP, LegalO
CreditR, LibStand, DisclReg, AntiDir,
SSecRegul, ListedF
UncAvoid, LTaxEv, MarkCap, Catholic, BlockPr,
ForeignInv, PublicEnf, EnglProf, OReligion, Latitude,
PolitConn, ReligFract, JudIndep, EnglProx
66
Panel A: Correlations between variables (cont)
Foreign investment to
GDP ForeignInv MarkCap, ListedF
RegQ, CreditR, PrivCtrIndex, PrivCtrEA, ProprR, CCorr,
UncAvoid, PolitStab, InstHoldDom, JudEff, LTaxEv,
PrivCtrEP, EnglProx, JudIndep, Law, IPO, EnglProf,
BTaxInd
Catholic Catholic LegalO, OReligion, BlockPr
CreditR, PrivCtrIndex, UncAvoid, Trust, USlisting,
PrivCtrEP, PrivCtrEA, LibStand, DisclReg, LTaxEv,
JudIndep, OwnConc, Protestant, ListedF, SSecRegul,
MarkCap, Muslim, AntiDir
Other religion OReligion Catholic, ListedF
PrivCtrEP, SSecRegul, IPO, PrivCtrIndex, CreditR,
MarkCap, LegalO, DisclReg, JudEff
Uncertainty avoidance UncAvoid
LegalO, CreditR, Secrecy, EnglProf,
LibStand, DisclReg, EnglProx,
SSecRegul, AntiDir
PrivCtrIndex, PrivCtrEA, LTaxEv, Catholic, Protestant,
MarkCap, JudIndep, PublicEnf, AuditSpend, Trust,
ListedF, PrivCtrEP, Big4Share, ProprR, ForeignInv,
BlockPr, JudEff, CCorr, IPO, ReligFract
Legal origin LegalO CreditR, PrivCtrIndex
PrivCtrEP, DisclReg, EnglProf, Catholic,
LibStand, UncAvoid, PrivCtrEA,
EnglProx, AntiDir, LTaxEv, SSecRegul,
JudIndep
Secrecy, ListedF, MarkCap, JudEff, BlockPr, EnforAudS,
OwnConc, ReligFract, OReligion, PublicEnf, ProprR
Latitude Latitude PolitScore
PowerD, Gdpc, IndividH, CCorrL,
Democracy, LawO, Law, LPolitR, CCorr,
Religness, PolitConn, LExprR, Secrecy,
PolitStab, LRepContr, InfoKnow, Trust,
NrAnal, RegQ, IntHoldFor, Media,
PrivCtrEA, EnforAudS
JudIndep, Protestant, JudEff, ProprR, IndividW, AntiDir,
PublicEnf, Muslim, LangFract, PrivCtrIndex, EthFract,
LegislComp, SSecRegul, EnforAccS
Democracy Democracy PolitScore, Muslim
Latitude, IndividH, PolitConn,
LegislComp, Religness, IndividW,
LPolitR, CCorrL, PowerD
Law, LawO, LRepContr, EthFract, LExprR, Gdpc,
PolitStab, RegQ, EnforAudS, CCorr, Secrecy, JudIndep,
LangFract, ProprR, PublicEnf, Media, JudEff, HierIndep,
PrivCtrEA
Political score PolitScore
Latitude, Democracy, Law,
LPolitR, CCorr, Religness,
CCorrL, IndividH, LawO, Gdpc,
RegQ, PolitStab, LExprR,
LRepContr, PowerD
Muslim, ProprR, JudIndep, Secrecy,
PolitConn, JudEff, EnforAudS, Media,
IndividW, Protestant, Trust, InfoKnow,
EthFract
IntHoldFor, LangFract, NrAnal, PublicEnf, EnforAccS,
TaxComp, OwnConc, Big4Share, Budhist, BankPriv,
LTaxEv
Legislative
competition LegislComp Veto, Democracy
HierIndep, PolitConn, Latitude, LTorient, IndividW,
IndividH, ClassAction
Politically connected
firms PolitConn
PolitScore, IndividW, Latitude,
Democracy, Muslim
Religness, CCorrL, LegislComp, PowerD, CCorr,
PrivCtrEA, Law, JudEff, Gdpc, IndividH, EthFract,
LPolitR, ProprR, RegQ, PolitStab, BTaxInd, EnforAudS,
LawO, JudIndep, InfoKnow, PrivCtrIndex, Secrecy,
LangFract, EnglProf, EnglProx
67
Panel A: Correlations between variables (cont)
Muslim Muslim Democracy
PolitScore, LPolitR, PolitStab, RegQ,
CCorrL, PolitConn, LawO
Law, CCorr, EthFract, IndividW, LangFract, LRepContr,
Religness, LExprR, IndividH, Gdpc, ProprR, Latitude,
Media, EnforAudS, JudEff, Catholic, PowerD, JudIndep
Power distance PowerD Secrecy, PolitScore
Latitude, Gdpc, IndividH, InfoKnow,
CCorrL, JudIndep, CCorr, Law, Trust,
RegQ, Religness, LExprR, LRepContr,
JudEff, Protestant, LawO, EnforAudS,
Democracy
LPolitR, ProprR, PolitConn, PolitStab, Media, LTaxEv,
IndividW, CreditR, PublicEnf, LangFract, IntHoldFor,
EthFract, EnglProx, EnglProf, OwnConc, Muslim
Individualism in
income IndividW PolitConn, PolitScore, Democracy
Religness, Law, Muslim, Latitude, CCorr, CCorrL,
JudEff, PolitStab, RegQ, ProprR, LPolitR, JudIndep,
LangFract, LawO, BankPriv, EthFract, PowerD, Gdpc,
IndividH, LExprR, LRepContr, LegislComp, Budhist,
EnforAudS
US institutional
holdings UShold USlisting
LTorient, BlockPr, ProprR, LRepContr, PolitStab, LawO,
IPO
Domestic institutional
holdings InstHoldDom BTaxInd, ForeignInv
US cross-listing USlisting UShold LTorient, BlockPr, Catholic, BankPriv
Masculinity Masculin Protestant, Trust
Long-term orientation LTorient BankPriv
USlisting, LRepContr, Religness, UShold, LExprR,
AuditSpend, LegislComp, PublicEnf, ProprR, Gdpc,
Media
IPOs to GDP IPO MarkCap
Media, ListedF, EnforAudS, RegQ, Gdpc, BlockPr,
PolitStab, LRepContr, LPolitR, Law, ProprR, LawO,
OReligion, LibStand, SSecRegul, OwnConc, JudEff,
EnforAccS, LTaxEv, BankPriv, LExprR, CCorr, CCorrL,
DisclReg, UncAvoid, ForeignInv, CreditR, EnglProx,
UShold
Block premium BlockPr DisclReg, OwnConc, LibStand, Catholic
SSecRegul, PrivCtrEP, CreditR, IPO, PrivCtrIndex,
LTaxEv, USlisting, MarkCap, LegalO, Trust, UShold,
Law, JudIndep, UncAvoid, ProprR, BankPriv, Religness,
EnforAudS, CCorr, PrivCtrEA, CCorrL, TaxComp,
Secrecy, LRepContr, ListedF, Gdpc
Public control of self-
dealing PublCtr PrivCtrIndex
Strengh of securities
regulation SSecRegul
DisclReg, PublicEnf,
LibStand
PrivCtrEP, PrivCtrIndex, UncAvoid,
LegalO, ClassAction
BlockPr, ListedF, MarkCap, LangFract, PrivCtrEA,
OReligion, EnglProf, IPO, AntiDir, ReligFract, CreditR,
EnforAccS, Catholic, OwnConc, Latitude, LTaxEv
Securities regulation
disclosure
requirements DisclReg LibStand, SSecRegul
PrivCtrIndex, LegalO, PrivCtrEP,
PublicEnf, UncAvoid, PrivCtrEA,
BlockPr, MarkCap
CreditR, AntiDir, ListedF, OwnConc, Catholic,
LangFract, LTaxEv, EnglProf, ReligFract, EnforAccS,
IPO, OReligion, EnglProx, HierIndep, Budhist
68
Panel A: Correlations between variables (cont)
Public enforcement
securities regulation PublicEnf SSecRegul DisclReg, LibStand
ClassAction, Latitude, UncAvoid, PrivCtrIndex, ListedF,
PolitScore, PrivCtrEP, PrivCtrEA, Democracy, PowerD, LegalO,
LTorient, TaxComp, Religness
Securities regulation
liability standards LibStand DisclReg, SSecRegul
PrivCtrIndex, PrivCtrEP, LegalO,
PublicEnf, UncAvoid, PrivCtrEA,
MarkCap, BlockPr
CreditR, LTaxEv, AntiDir, ListedF, EnforAccS, Catholic, IPO,
ReligFract, LangFract, EnglProf, OwnConc, EnglProx, BankPriv,
EnforAudS
Tax compliance TaxComp
EthFract, LPolitR, PolitStab, Law, ClassAction, Religness, CCorr,
RegQ, LRepContr, Gdpc, BTaxInd, Trust, LangFract, JudEff,
LawO, LExprR, ProprR, CCorrL, Media, JudIndep, BankPriv,
InfoKnow, PolitScore, IntHoldFor, PublicEnf, CreditR, LTaxEv,
BlockPr
Book tax
independence BTaxInd
ReligFract, TaxComp, PolitConn, IndividH, InstHoldDom,
NrAnal, ForeignInv, EnglProx
Number of veto-
players Veto LegislComp ClassAction
Language
fractionalization LangFract EthFract, PolitStab, LPolitR
LawO, Muslim, InfoKnow, Religness, PolitScore, Gdpc, RegQ,
SSecRegul, CCorr, Law, TaxComp, LExprR, IndividW, DisclReg,
CCorrL, LRepContr, Media, LibStand, Latitude, Democracy,
PowerD, JudEff, PolitConn, EnforAudS
Ethnic
fractionalization EthFract
LangFract, Religness, LPolitR, PolitStab,
PolitScore, Law
TaxComp, RegQ, Muslim, InfoKnow, Gdpc, CCorr, Democracy,
LawO, JudEff, Media, LRepContr, PolitConn, IndividW, LExprR,
ProprR, Latitude, CCorrL, PowerD, JudIndep, IntHoldFor
Class action lawsuit ClassAction SSecRegul PublicEnf, TaxComp, Veto, ReligFract, LegislComp, OwnConc
Religion
fractionalization ReligFract LTaxEv
CreditR, EnglProx, Big4Share, BankPriv, ClassAction, BTaxInd,
ProprR, OwnConc, JudEff, CCorrL, MarkCap, LibStand,
EnglProf, JudIndep, Secrecy, EnforAudS, SSecRegul, LegalO,
DisclReg, CCorr, IndividH, RegQ, PrivCtrIndex, LExprR,
LRepContr, Law, LawO, UncAvoid
Buddhist Budhist EnglProf, IndividW, PolitScore, DisclReg, IndividH
English proficiency EnglProf EnglProx LegalO, Secrecy, CreditR, UncAvoid
Budhist, EnforAudS, IndividH, PrivCtrEP, Big4Share, JudIndep,
JudEff, PrivCtrIndex, SSecRegul, ReligFract, ListedF,
AuditSpend, DisclReg, LibStand, EnforAccS, LTaxEv, PowerD,
PolitConn, CCorr, PrivCtrEA, ForeignInv, MarkCap, ProprR
Language proximity to
English EnglProx EnglProf Secrecy, CreditR, LegalO, UncAvoid
JudIndep, IndividH, EnforAudS, ReligFract, JudEff, ProprR,
Big4Share, LTaxEv, CCorr, EnforAccS, LibStand, DisclReg,
RegQ, Gdpc, MarkCap, Law, ForeignInv, PowerD, PrivCtrIndex,
CCorrL, PrivCtrEP, PolitConn, AuditSpend, BTaxInd, IPO
69
Panel A: Correlations between variables (cont)
Big4 market share Big4Share
Secrecy, LTaxEv, ReligFract, JudEff, EnglProf, CCorr,
RegQ, CCorrL, IndividH, EnforAudS, LExprR, LawO,
PolitStab, HierIndep, UncAvoid, CreditR, EnglProx, Gdpc,
OwnConc, MarkCap, Protestant, ProprR, Law, PolitScore,
Trust, LPolitR, LRepContr, JudIndep
Audit spending AuditSpend UncAvoid, LTorient, EnglProf, Secrecy, EnglProx
Enforcement of audit
standards EnforAudS
Gdpc, IndividH, LExprR, CCorrL,
LRepContr, Secrecy, EnforAccS,
Law
CCorr, RegQ, JudEff, LawO, ProprR,
JudIndep, Media, PolitScore, LPolitR,
CreditR, NrAnal, Religness, PolitStab,
InfoKnow, PowerD, Latitude
BankPriv, IPO, LTaxEv, ListedF, EnglProx, Trust, EnglProf,
Democracy, OwnConc, PrivCtrEP, Protestant, Big4Share,
LegalO, ReligFract, Muslim, PolitConn, MarkCap,
IntHoldFor, BlockPr, HierIndep, LangFract, LibStand,
IndividW
Enforcement of
accounting standards EnforAccS EnforAudS
IndividH, NrAnal, Gdpc, Secrecy,
LExprR, LRepContr
RegQ, LawO, Law, LibStand, CCorrL, JudIndep, CreditR,
CCorr, JudEff, Media, ProprR, IPO, BankPriv, LPolitR,
PolitScore, PrivCtrEP, ListedF, EnglProx, DisclReg,
LTaxEv, EnglProf, InfoKnow, SSecRegul, Latitude
Ownership
concentration OwnConc BlockPr
LTaxEv, LExprR, Gdpc, Religness, HierIndep, LRepContr,
JudIndep, Secrecy, Law, EnforAudS, JudEff, ReligFract,
PrivCtrEP, LawO, ProprR, DisclReg, IPO, IndividH, Trust,
CCorr, LegalO, Catholic, LibStand, CreditR, CCorrL,
Big4Share, InfoKnow, Protestant, SSecRegul, Media, RegQ,
PolitScore, BankPriv, NrAnal, LPolitR, PowerD,
ClassAction
Individualism IndividH Secrecy, EnforAudS, PolitScore
Gdpc, CCorrL, Latitude, LExprR,
PowerD, JudEff, LawO, Law, EnforAccS,
Democracy, CCorr, LRepContr, LPolitR,
JudIndep, NrAnal, RegQ, Religness,
Protestant, ProprR
PolitStab, Trust, Media, EnglProx, Muslim, EnglProf,
InfoKnow, PolitConn, CreditR, Big4Share, OwnConc,
IndividW, BTaxInd, ReligFract, LegislComp, IntHoldFor,
Budhist
Secrecy Secrecy
IndividH, PowerD, JudIndep,
EnforAudS, CCorr, Gdpc
JudEff, Trust, CCorrL, Protestant, Law,
PolitScore, RegQ, CreditR, LExprR,
ProprR, EnglProx, EnglProf, LRepContr,
LawO, UncAvoid, LPolitR, Latitude,
EnforAccS, InfoKnow, LTaxEv
Religness, LegalO, PolitStab, Big4Share, Media, OwnConc,
NrAnal, Democracy, ReligFract, AuditSpend, IntHoldFor,
PolitConn, BlockPr
Hierarchy vs
independence HierIndep
LegislComp, OwnConc, Big4Share, EnforAudS,
Democracy, DisclReg
Number of analysts NrAnal
EnforAudS, IndividH, EnforAccS,
Latitude, LExprR
LawO, Media, Gdpc, LRepContr, PolitScore, CCorrL, Law,
LPolitR, Secrecy, CCorr, RegQ, BTaxInd, IntHoldFor,
OwnConc, PolitStab
70
Panel B: Statistics of the number of correlated variables
Absolute
correlation >= 0.7
0.5 =< Absolute
correlation < 0.7
0.3 =< Absolute
correlation < 0.5
mean 3.7 7.5 16.9
median 2 7 18
min 0 0 1
max 18 23 37
Appendix 2 reports patterns of correlatations between variables representing country characteristics. Panel A reports the most correlated variables with each one of the country variables.
Panel B reports summary statistics on the number of correlated variables for each group of correlations. Variable definitions are provided in appendix 1.
71
Appendix 3 - Standardized scores of country factors for 47 countries Country Factor 1 Factor 2 Factor 3 Factor 4
Finland 1.555 -0.215 -0.335 0.172
Switzerland 1.459 0.096 0.101 -0.344
Sweden 1.405 -0.443 0.937 -0.164
Austria 1.382 -1.199 -0.512 0.762
Norway 1.373 -0.670 1.489 -0.306
Denmark 1.319 0.109 0.681 0.269
Netherlands 1.176 -0.256 0.738 0.424
Germany 1.171 -1.140 0.741 -0.813
Ireland 0.990 1.081 -0.716 2.120
Japan 0.880 -0.541 -0.765 -2.617
Belgium 0.837 -0.729 -0.333 -0.284
New Zealand 0.792 0.986 0.582 0.684
France 0.729 -0.564 0.169 -0.920
United Kingdom 0.696 1.560 0.905 -0.424
Hong Kong 0.662 2.822 -1.827 0.419
Australia 0.624 1.197 1.192 0.258
Canada 0.517 1.243 1.933 0.203
Portugal 0.471 -1.053 -0.846 0.591
Spain 0.401 -0.586 0.087 0.122
Italy 0.335 -0.926 0.099 -0.621
United States 0.315 1.152 2.267 -0.288
Chile 0.260 -0.036 -1.716 1.452
Singapore 0.159 2.804 -1.859 0.015
Czech Republic 0.120 -0.510 -0.368 -0.409
Greece 0.104 -1.398 -0.756 -0.392
Israel 0.064 0.789 -0.207 1.456
Korea 0.046 -0.421 -0.797 -1.562
Poland 0.045 -0.800 -0.415 0.178
Taiwan -0.093 0.128 -0.562 -1.934
Argentina -0.517 -1.150 -0.783 1.786
Mexico -0.530 -1.115 -0.871 1.784
China -0.580 -0.055 -0.744 -1.828
Russian Federation -0.589 -0.519 -1.012 -0.631
Brazil -0.651 -1.190 0.374 0.949
South Africa -0.791 0.840 0.950 0.315
Turkey -0.820 -0.925 -0.419 -0.263
Colombia -0.962 -0.501 -0.391 1.211
Peru -1.053 -0.424 -0.625 1.539
Malaysia -1.077 1.856 -1.054 -0.949
Thailand -1.136 0.612 -1.075 -1.441
India -1.256 0.674 0.839 -0.026
Kenya -1.370 -0.274 1.413 0.219
Philippines -1.591 -0.138 0.248 0.319
Zimbabwe -1.595 -0.019 1.071 0.210
Indonesia -1.647 -0.196 -0.746 -1.114
Nigeria -1.781 0.003 1.795 -0.099
Pakistan -1.848 0.038 1.123 -0.024
72
Online Appendix 1 to “Financial Reporting Differences Around the World: What
Matters?”
Application of Qualitative Comparative Analysis (QCA)/Fuzzy Sets
1) Terminology and brief explanation
Outcome variable = Financial reporting factor (q) as reported in section 2.1 the paper (i.e. factor from
factor analysis applied to six financial reporting characteristics: reporting transparency, disclosure quality,
abnormal return, abnormal volume, return synchronicity, and timeliness).
Predictor sets = Country factor 1, country factor 2, country factor 3, country factor 4 (factors obtained
from factor analysis on 72 country-level attributes as explained in section 3.3 of the paper).
Sets High presence of set Low presence of set
Country factor 1 F1 f1
Country factor 2 F2 f2
Country factor 3 F3 f3
Country factor 4 F4 f4
QCA/Fuzzy Sets evaluates the relation between the financial reporting outcome factor (q) and all possible
Boolean combinations of country factor sets (e.g., F1 . F2, F1 . f2, F2 . f1, f1 . f2, F1 . F3, F1 . f3, … ;
where operator “.” represents the Boolean “and”). QCA/Fuzzy Sets examines which combinations of sets
or configurations are most likely to be present when the financial reporting outcome q is high.
Country observations are represented in terms of the level of membership in a set, for example the level
of membership of the US in country factor 1 is 0.529. In our context the level of set membership can take
any value between 0 (completely exclusive) and 1 (completely inclusive). Hence our setting is a fuzzy set.
2) Fuzzy set analysis
Table 1 – Combinations of country factors that exist when financial reporting outcome is high
Combinations
Y
consistency
N
consistency F value P-value
Country factors
F1 F2 f3 F4 0.978 0.651 11.48 0.002 High factors 1,2 and 4; Low factor 3
F1 F2 F3 f4 0.946 0.695 4.780 0.036 High factors 1,2 and 3; Low factor 4
F1 F2 F3 F4 0.986 0.593 19.98 <0.001 High factors 1,2, 3 and 4
Table 1 reports the combinations that pass the sufficiency condition “if the combination of country factors
exist then financial reporting outcome (q) also exist”. The three combinations pass the commonly used
73
sufficiency tests. First, the probability that “the combination exists when outcome q exists“ or Y
consistency is significantly greater than the probability that “the combination exists when outcome q does
not exist” or N consistency. Second, the Y consistency exceeds the recommended benchmark value of 0.8
(Ragin 2000, 2006).
Because the three significant combinations may overlap we perform a reduction of the proposed solution
(Table 2).
Table 2 – Final solution: country factors combinations leading to high financial reporting outcome
Combinations
Raw
coverage
Unique
coverage Consistency
Country factors
F1 F2 F3 0.499 0.133 0.983 High country factor 1, 2 and 3
F1 F2 F4 0.455 0.089 0.958 High country factor 1, 2 and 4
Total coverage: 0.588
Total consistency: 0.962
The final reduced solution suggests that there are two recipes of country factors that lead to high quality
financial reporting. The first is a high level of country factor 1 (associated with good institutional and
governance systems, and economic and social welfare), high level of country factor 2 (associated with
strong protection of investors’ rights and capital markets development), and high level of country factor 3
(associated with political transparency, and tax and accounting enforcement). The second is a
combination of high country characteristics represented in factors 1, 2 and 4 (openness of society to
external investment). The two recipes together explain 58.8% of the membership in the financial reporting
outcome, a percentage that can be interpreted as R2 in regression models. The overall solution
consistency of 0.962 indicates that few cases deviate from the two patterns identified in the data
(coverage would be 1 if all countries perfectly fit into one of the two recipes).
Comparison of the statistics between the two recipes suggests that having high country factors 1, 2 and 3
is probably a better recipe to achieve high quality financial reporting. The extent to which that recipe
explains financial reporting (raw coverage), the proportion of cases that the recipe alone explains (unique
coverage), and consistency is higher for the first recipe.
In summary, the Fuzzy Set analysis confirms the regression results discussed in the paper. The Fuzzy Set
analysis suggests that a solution with one single country factor is not sufficient to achieve high quality
financial reporting outcomes. Further, country factor 1 and country factor 2 are necessary but not
sufficient for the desired financial reporting outcome. We conclude that an intertwined combination of
many country characteristics needs to exist in a country for high quality financial reporting to exist.