Upload
marialauraarmijo7591
View
7
Download
0
Tags:
Embed Size (px)
DESCRIPTION
hbrs working paper on corporate governance
Citation preview
Electronic copy available at: http://ssrn.com/abstract=2192460
Copyright © 2012 by Andrea Hugill and Jordan Siegel
Working papers are in draft form. This working paper is distributed for purposes of comment and discussion only. It may not be reproduced without permission of the copyright holder. Copies of working papers are available from the author.
Which Does More to Determine the Quality of Corporate Governance in Emerging Economies, Firms or Countries? Andrea Hugill Jordan Siegel
Working Paper
13-055 December 21, 2012
Electronic copy available at: http://ssrn.com/abstract=2192460
Which Does More to Determine the Quality of Corporate Governance in Emerging Economies, Firms or Countries?
Andrea Hugill, Harvard Business School Jordan Siegel, Harvard Business School1
First Draft: August 1, 2011
This Version: December 21, 2012
ABSTRACT
Scholars of corporate governance have debated the relative importance of country characteristics and firm characteristics in understanding variations in the corporate governance practices of firms in emerging economies. Using panel data and a number of model specifications, we shed new light on this debate. We find that firm characteristics are as important as and often meaningfully more important than country characteristics in explaining governance ratings variance. Our findings show that firms in emerging economies over recent years had more capability to rise above home-country peer firms in corporate governance ratings than has been previously suggested.
1 Corresponding author can be reached at Morgan Hall, Harvard Business School, Boston, Massachusetts 02163, [email protected]. We thank Chris Poliquin and Chris Allen for research assistance, the CLSA staff for data assistance, as well as the Harvard Business School Division of Research for funding. All remaining errors are our own.
Electronic copy available at: http://ssrn.com/abstract=2192460
2
I. Introduction
Variation in firms' corporate governance is an important topic of debate in the
governance literature. One of the main questions is whether weak and/or incomplete public
institutions in various countries dictate the governance quality of firms located there. The most
recent scholarship on the subject has largely argued that country characteristics strongly predict
local firms’ governance (Krishnamurti, Sevic, and Sevic (2006)). Doidge, Karolyi, and Stulz
(2007) find that country variables explain 39-73% of governance variance while firms explain
only 4-22%. Moreover, they argue that firm characteristics explain almost none of the variation
in governance ratings in less-developed countries. However, several other previous papers have
argued that various firm characteristics can play an important role in determining the firm’s own
governance practices (Klapper, Laeven, and Love (2006), Sawicki (2009)). Durnev and Kim
(2005) find three firm-level variables that relate to the governance variation of firms within a
country, and show how this relationship is stronger in less investor-friendly countries.
In this paper, we offer new understanding of firm and country characteristics’
contribution to corporate governance ratings in emerging economies by using an updated panel
data set for the last decade and by including unobservable firm characteristics for comparison.
These unobservable firm fixed effects are seen by looking at fixed effects, random effects, and
nested ANOVA models. We also capture more observable firm effects by running regressions
that use a richer set of firm variables than used in previous papers. Finally, we advance
understanding by conducting multiple analyses on emerging economies and developed
economies separately. We do those latter analyses in order to explore critical differences
between the two types of countries.
3
Our main focus is on emerging economies. Much research on corporate governance in
the last 15 years has focused primarily on this unique setting. Emerging economies are often
characterized by weak governance institutions such as poorly enforced regulatory systems and
corruption. Previous literature has sought to understand the negative effect of these incomplete
institutions on corporate governance quality. One of the most fundamental questions being
asked by scholars is if the country’s institutions are entirely prescriptive of local firms’
governance practices. We repose this question with special attention to the additional research
question of why some firms establish high quality corporate governance even in economies with
very weak and/or incomplete governance institutions. By providing insight into the important
characteristics that can be used to understand firm governance practices in emerging economies,
this paper seeks to explain how some firms in emerging economies have managed to move
independently from their home country institutions in recent years.
Corporate governance fuels growth by providing investors an assurance of a return on
their investment (Shleifer and Vishny (1997)). Investors may be more willing to offer valuable
financing or pay a higher equity price for firms with better governance (Chen, Chen and Wei
(2009)). This financing could be critical to growing a firm’s value. Indeed, Black, Jang, and
Kim (2006) report that higher corporate governance ratings are causally related to higher firm
value. Corporate governance should, therefore, be especially important in emerging economies
where firms are often forced to rely on outside investors to help finance growth opportunities.
The ratings given by international organizations to firm governance practices should similarly be
important to investors looking for useful information about firms. Ratings may be especially
useful in emerging economies when other signals of firm value may be opaque.
4
Given the clear importance of corporate governance ratings for firms in emerging
economies, recent literature arguing the much greater importance of country characteristics
seems unintuitive. We argue that firms in emerging economies actually had more capacity to
rise above their home country institutions in recent years. The debate over the relative
importance of firm and country characteristics in explaining corporate governance ratings’
variance thus seems unresolved. This study is motivated by the ongoing debate in the literature.
Specifically, earlier work relies on cross-sectional analysis, uses limited firm variables, ignores
unobservable firm characteristics, and fails to account for the nested nature of firm governance
ratings.
Black and his coauthors’ quantitative approach to governance highlights the
shortcomings of cross-sectional data. Looking at the relationship between corporate governance
and share price in Russia, Black, Love, and Rachinsky (2006) show both OLS and fixed-effects
specifications are unreliable when using cross-sectional data. Studies that rely on these methods
risk that endogeneity or omitted firm-level variables are actually responsible for observed
correlations. Alternative approaches to cross-sectional work include the event study, which
Black and Khanna (2007) have used successfully to evaluate the effect of corporate governance
reforms in India. Black also employs an instrumental variables approach, which successfully
shows that corporate governance measures result in higher firm values (Black, Jang, and Kim
(2006)). We improve on previous cross-sectional studies by using a panel of data across multiple
years and multiple countries. We use three corporate governance ratings datasets ranging from
4-11 years, each covering years in the range of 2000-2010.
We also improve on previous work by expanding the original set of observable firm
characteristics analyzed. This original set of firm characteristics includes five observable
5
variables looking at a small portion of what identifies firms in emerging markets. These
variables include sales growth and cash/assets. We identify 17 additional observable firm
variables that we predict should be prescriptive of firm corporate governance choices. These
additional firm variables capture additional and highly relevant firm characteristics such as
income growth, R&D intensity, and foreign sales. We show how models that include the full set
of firm characteristics consistently explain far more of the variance than models that simply use
the original, limited set of firm variables.
In addition to expanding the set of observable firm characteristics, we also aim to capture
unobservable firm characteristics by looking at firm fixed effects. Country fixed effects have
been used in previous studies on this subject, but to our knowledge, no paper has also included
firm fixed effects. These firm fixed effects are intended to detect unobservable processes
happening inside firms that have not been captured by the observable firm variables. In addition
to firm fixed effects, we also explore unobservable firm choice using ANOVA and random
effects specifications.
Lastly, our methodology improves on that of previous studies by accounting for the
nested nature of the data. As firms are necessarily embedded within countries, it is impossible to
run analyses on firms alone and not simultaneously capture some of the country effect. Year
effects can also be mistakenly attributed to country or firm effects if not analyzed separately. To
isolate firm and country effects, we run our models successively. First, we look at year effects,
then country and year effects. Subtracting the year effects gives us our country effects. To look
at firm effects, we subtract the year and country effects from those explained by year, country,
and firm. This difference gives us an accurate measure of what firm characteristics alone are
contributing to governance variance.
6
These methodological improvements run on three sets of panel data show that, in
emerging economies, firm-level variables are anywhere from roughly equal to significantly more
important than country-level variables in explaining variance in corporate governance ratings.
For developed economies, in contrast, we see that country-level variables always explain more
variance. Therefore, we see two distinct patterns for these two different types of countries.
These patterns are consistent across all our models and throughout use of three different datasets.
Our corporate governance ratings data come from two main sources. First, we use
complete panel data provided to us by the independent investment research firm, Credit Lyonnais
Securities Asia (CLSA), which tracked corporate governance behavior of firms in emerging
economies from 2000-2010. Data from CLSA have been used by several different studies
including Chen, Chen, and Wei (2009), Doidge, Karolyi, and Stulz (2007), Khanna, Kogan, and
Palepu (2006), Durnev and Kim (2005), and Klapper and Love (2004). Secondly, we use data
from the Global Reporting Initiative (GRI), which issued market index-benchmarked as well as
industry-benchmarked corporate governance quotients for most of the last 10 years. This second
dataset is much larger and encompasses a greater number of countries and observations. As
well, the GRI dataset is dominated by developed economies, a feature we leverage to look at the
differences between emerging and developed economies.
In addition to these two main datasets, we also looked at panel data from FTSE’s
Corporate Governance Ratings Index in order to compare our results with those in previous
papers that used this data. Ratings data were provided to us by FTSE for four years beginning in
2005 until the index was discontinued in 2008. A single year of this data was used in the
previous corporate governance studies that found greater importance for country variables
including Doidge, Karolyi, and Stulz (2007). Instead of using a single year of this data, we use
7
four years of data from FTSE’s Corporate Governance Ratings Index. We run our full set of
models including OLS with observable variables, fixed effects, random effects, and ANOVA.
The results from using the FTSE data confirm the same trend we find in the CLSA and GRI
datasets, thereby strengthening the conclusion that our results are not merely due to data
selection.
Across our main two data sets from CLSA and GRI we see that, in emerging economies,
unobservable plus observable firm characteristics explain 37.3-50.3% of the corporate
governance ratings’ variance, and country characteristics explain roughly 11-28.5% of the
variance.2 We were only able to look at developed economies alone in the GRI dataset, as there
were too few developed country observations in the CLSA data. The results here for developed
economies strongly contrast with those from emerging economies. Observable and unobservable
firm characteristics explain only 15.3-19.1% of governance ratings variance in developed
economies while country characteristics explain 45.9-57.3%.3 Therefore, in emerging
economies, firm variables explain roughly the same amount and often more of the governance
variance than do country variables. In developed economies, in contrast, country variables
explain significantly more of the corporate governance ratings than do firm variables.
Our results provide evidence that many emerging economy firms distinguished
themselves above and beyond their home country peers in corporate governance ratings during
the last decade. This rise was due primarily to firm-level characteristics, the most important of
2 This range comes from the regressions that involve both observable and unobserved firm and country characteristics in the form of fixed effects (OLS), random effect regressions (xtmixed), and nested ANOVA regressions. Firm effects contributed the least in the random effects model using the CLSA corporate governance score as the dependent variable. Firm effects explained the most variance in the random effects model using the GRI Industry-Based Corporate Governance Quotient as the dependent variable. We excluded results from the regressions using only observable characteristics without fixed effects because they explained far less of the variance overall. 3 Country effects explained the most variance in the ANOVA model, using the GRI Industry-Based Corporate Governance Quotient as the dependent variable. Country characteristics explained the least variance in the random effects regression using the GRI Index-Based Corporate Governance Quotient as the dependent variable.
8
which are unobservable. The fact that firm characteristics, and especially fixed effects, played a
substantially greater role in emerging economies suggests that there is something happening
inside these firms that allowed them to differentiate themselves from their home institutions and
peer firms. These findings are important for both investors and firms in emerging economies.
Investors will be able to observe corporate governance variance within countries and identify
valuable investment opportunities. Also, firms should enjoy a sense of agency in their prospects
for growth, unhampered by an environment with weak and incomplete governance institutions or
low financial market development.
The remainder of this paper is organized as follows. Section II provides the institutional
background and describes the previous literature related to our study. In Section III, we describe
our data and present our methodology. Section IV provides the results of our analysis, Section V
presents robustness checks, and Section VI concludes.
II. Background
Investor protection, provided at the national level by the government, is important in
determining the quality of firm-level governance on the ground. Understanding country-level
institutions such as the legal protections for minority shareholders has been greatly advanced by
work looking at the legal origins of countries. Legal and colonial origins can be important
determinants of present day legal institutions (Acemoglu, Johnson, and Robinson (2001)), which
largely determine current governance practices in the country (La Porta, Lopez-de-Silanes,
Shleifer, and Vishny (LLSV hereafter) (1998)). Weak legal protection for minority shareholders
is strongly related to less developed capital markets (LLSV (1997), (1999)). Thus, firm
corporate governance measures are important for accessing capital (LLSV (2000), (2002)),
9
especially at a lower cost (Black, Jang, and Kim (2006)). In spite of growing pressure to
converge to similar institutions, country-level differences may persist due to path dependence
(Bebchuk and Roe (1999)).
Previous work has sought to answer whether country characteristics better explain
corporate governance policies than do firm characteristics. This work has provided insight into
whether firms in countries with weak legal institutions can rise above their home-country peers
to adopt strong corporate governance policies. Recent work has found a strong role for country
characteristics in explaining firm governance behavior (Doidge, Karolyi, Stulz (2007),
Krishnamurti, Sevic, and Sevic (2006)). Doidge, Karolyi, and Stulz (2007), using a cross-section
of the CLSA data, find that country variables explain 39-73% of ratings variance, while firm
characteristics only explain 4-22%. In less developed countries, they argue that firm
characteristics explain almost none of the variance because the costs of adopting good
governance outweigh the benefits. Krishnamurti, Sevic, and Sevic (2006) look at Asian firms
after the financial crisis and find a strong country effect in firm-level governance scores.
Regressions of the components of firm level corporate governance, they argue, reveal that the
country fixed effects are the only consistently significant variables, with a few exceptions. In
general, this work argues that firms are limited in their flexibility to affect their own governance,
separate from their country-level institutions (Klapper and Love (2004)).
Other work has argued for a link between country institutions and various firm
performance and governance measures. Much of this work shows the importance of a country’s
institutions in determining firm choices and outcomes. While widely recognized that firms
underprice at IPO’s, Engelen and Essen (2010) show that the quality of a country’s legal system,
as measured by investor protection is a strong predictor of underpricing, where firm-level and
10
issue-specific variables were previously thought to be the only relevant characteristics.
Tunneling as well is closely related to the institutions of the country in which the firm is based.
How much and which kind of tunneling occurs in firms differs across countries with different
legal rules (Atanasov, Black, Ciccotello, and Gyoshev (2010)). Finally, outcomes from the
Asian Financial Crisis (1997-1998) also reveal a strong role for country-level corporate
governance institutions. During and after the crisis, legal protection for minority shareholders
best predicted exchange rate depreciation and stock market declines, not national wealth
measured by GDP (Johnson, Boone, Breach, and Friedman (2000)).
A more limited set of work has argued the relative importance of firm characteristics in
determining firm corporate governance choices (Durnev and Kim (2005), Klapper, Laeven, and
Love (2006)). Corporate governance quality and firm performance, here, is viewed as an
endogenous firm choice (Himmelberg, Hubbard, and Palia (1999)). As firms respond to external
incentives and environmental change, they adopt various corporate governance practices to
match their goals. Previous work that employs this perspective has explored governance
practices’ relationship with one or several specific firm characteristics. A firm’s ownership
structure, for example, can play a strong role in determining the firm’s corporate governance
practices, specifically the expropriation of minority shareholders (Lemmon and Lins (2003)).
Variation in the protection of minority shareholders exists within emerging economies with weak
country-level institutions. Looking at the Asian Financial Crisis from 1997-1998, firms fared
better than their peer firms if they had employed higher quality corporate governance, even in
countries with weak legal protection of minority shareholders. Dividends issued by firms can
also exhibit a relationship with firm corporate governance. Sawicki (2009) shows that in Asian
countries before, during, and after the financial crisis, dividends were strong predictors of a
11
firm’s governance, and that this relationship was incremental to the country-level variable of the
legal regime.
Most often, this literature has explored the question of whether firm governance choices
are important in determining that firm’s financial performance. This literature thus seeks to
answer whether corporate governance is smart business. The answer provided by several papers
is that, yes, strong corporate governance improves a firm’s financial outcomes (Mitton (2002),
Bae and Goyal (2010)). One such paper by Bae and Goyal (2010) shows that, when South Korea
officially liberalized their equity market, firm-level variation in governance was strongly
associated with greater stock price increase, foreign ownership, and higher rates of capital
accumulation. Bae and Goyal’s work goes on to assert the greater importance of firm
characteristics than country characteristics in explaining these outcomes. In a wide sample of 28
countries across 1990-2003, firm and industry characteristics explain more of the variation in
earnings quality, as measured by accruals quality, persistence, predictability, smoothness, value
relevance, timeliness, and conservatism, than do country characteristics (Bae and Goyal (2010)).
In addition to work recognizing the variation in within-country corporate governance and
the work on the importance of governance for financial performance, few papers have addressed
the question at the center of this study, the importance of country institutions in determining firm
governance directly. One such paper by Klapper, Laeven, and Love (2006) acknowledges that
some variation in governance measures of cumulative voting and proxy by mail can be explained
by country fixed effects. However, Klapper et al. (2006) argue that a larger portion of firm
governance is explained by firm characteristics, specifically whether or not the firm has a second
large shareholder and the use of equity as a source of external financing. Firm investment
opportunities, external financing, and ownership structure, have also been shown to predict firm
12
governance more strongly than country characteristics (Durnev and Kim (2005)). These firm
characteristics have strong, positive correlations with each governance category that composes
the CLSA corporate governance ratings. Interestingly, Durnev and Kim (2005) also find that the
relationship between firm characteristics and corporate governance is stronger in countries with
less legal protection of investors. As emerging economies are often cited to have weaker legal
protections, their work supports our decision to examine emerging and developed economies
separately.
The ability of firms to distinguish themselves from their home country institutions and
peer firms has been taken up by other literature as well. Research on cross-listing, for example,
has noted that different types of firms from the same jurisdiction are more or less likely to cross-
list on US exchanges if they have higher growth prospects and are willing to sacrifice some
control for finance (Coffee (2002)). These firms are listing on US stock exchanges precisely in
order to exhibit their value and distinguish themselves from firms in their home country (Blass
and Yafeh (2001)). By cross-listing, firms enhance their reputations, and attract outside
financing for up to two years after they cross-list (Siegel (2005)). This work thus shows that
firms are willing to incur costs in order to improve their reputations and attract financing,
especially in countries with weak legal protection for shareholders (Reese and Weisbach (2002)).
III. Data
We implement our analysis using two main data sets. The first data set comes from the
Corporate Lyonnais Securities Asia (CLSA), an independent research firm that tracked a number
of corporate governance measures for emerging economy firms during the last decade. The
second dataset is from the Global Reporting Initiative (GRI), which gave industry and index-
13
related scores from 2003-2009. We also ran our analysis on a third dataset that was used in
previous papers, intended as a robustness check on our approach and unique results. This dataset
was FTSE’s Corporate Governance Ratings Index scores from 2005-2008. We did not include
the S&P data used in previous studies, as S&P did not continue to give ratings beyond a single
year for more than very few firms and thus our panel data approach would have been limited to
one year.
The CLSA corporate governance data was shared with investors annually in the
company's “CG Watch” reports. These reports highlighted emerging economy firms who had
exceptional governance (“CG Stars”) or firms which had fallen in their scores since the previous
year. We were given access to the complete CLSA historical ratings by the company. This was
composed of 10 years of data from 2000-2010. Each firm’s corporate governance score is
composed of ratings on 57 different sub-measures (plus or minus a few depending on the year).
These 57 sub-measures fall into the categories of discipline, transparency, independence,
accountability, responsibility, fairness, and social awareness. In the final year of the CG Watch
reports, CLSA also included a measure for environmental friendliness, “Clean and Green”. Over
475 firms were ranked along these metrics and given a final corporate governance score,
computed as the average of all the smaller measure scores. Over the time span of the data, these
firms compose 4,448 observations, 91% of which are from emerging economies.
Over the ten years that CLSA tracked corporate governance for emerging economy firms,
the methods by which the rankings were gathered changed only slightly. Each year, the points
awarded to each firm were determined by its answers to a lengthy survey conducted by CLSA.
Initially, each survey question was answered simply yes or no; a single point was awarded for
each yes and a zero for each no. Later, three more options were added: largely (0.75 points),
14
somewhat (0.5 points), and marginally (0.25 points). Points for each category were then
combined and weighted to produce the firm’s final score. The exact weighting of each category
also changed over the years. In 2000, the first year the scores were computed, discipline
accounted for 10 percent of the score while transparency, independence, accountability,
responsibility, fairness, and social awareness each accounted for 15 percent. In 2007, when the
Clean and Green category was introduced, responsibility was absorbed into another category;
each of the remaining categories accounted for 15 percent of the final score while Clean and
Green represented 10 percent. The exact questions also changed over the years, increasing in
number from 53 to 87; several were dropped and replaced with others. An example of a typical
survey question is this one from the Transparency category: “Does the company publish its full-
year results within three months of the end of the financial year?” The summary statistics for
several of these corporate governance measures appear in Table 1 Panel A, and the correlations
between the variables appear in Panel B.
Our second data set comes from the Global Reporting Initiative (GRI), which was issued
by Risk Metrics, Inc. During 2003-2009, GRI ranked the corporate governance performance of
over 2,200 companies worldwide, including all companies in the S&P 500, Russell 3000,
MSCI’s Europe, Asia and Far East and the S&P/TSX Composite, FTSE All-World Developed,
and FTSE All-Share indices. The final governance quotient for each company is computed using
ratings on 63 different issues in four categories: board of directors, audit, antitakeover, and
compensation/ownership. These 63 scores are combined into a single score for each firm, which
is then compared to the scores of other companies in the same index to produce the firm’s index
corporate governance quotient. The score is also compared to those of companies in the same
industry to produce the firm’s industry corporate governance quotient. Both measures are used
15
in the analysis presented here. The source data for the raw company scores in the GRI rankings
comes from public disclosures (SEC EDGAR filings for U.S. companies), press releases, and
corporate websites. It is compiled by GRI analysts. The summary statistics for these variables
appear in Table 1, Panel C and the correlations between these variables appear in Panel D.
Although there are similarities in the processes by which firm corporate governance
scores are assembled in the CLSA and GRI datasets, the methods are different enough to ensure
that our results are confirming a trend and not merely repeating results on similar data. The first
major difference between the CLSA corporate governance score and the GRI corporate
governance quotients is that the GRI scores are all relative. Thus, a score of 40 means that that
firms’ corporate governance performance is better than 40% of its peers. For the Index Score,
firms are compared to a relevant market index such as the S&P 500, Mid-Cap 400, Small-Cap
600, Russell 3000, or the CGQ Universe. For the Industry Score, firms are compared to an
industry peer group based on the S&P GICS (Global Industry Classification System) of 24
industry groups. The CLSA scores are not computed relative to any market index or peer
industry group.
A second, major way the two scores differ is in the design of the questions. The GRI
dataset did not initially include emerging economies and only started to do so in 2003. Prior to
that, the corporate governance quotients were computed only for US companies. As such, the
questions relate to issues that dominate US corporate governance concerns such as the charter
and bylaws. The CLSA questions focus instead on issues relevant to emerging economies such
as transparency and corruption. This can be seen by comparing the categories of questions. For
the CLSA data, the categories are discipline, transparency, independence, accountability,
16
responsibility, fairness, social awareness, and clean and green. For the GRI scores, the
categories are board of directors, audit, antitakeover, and compensation/ownership.
In addition to our two main datasets, we also explored trends in data from FTSE. Our
main intention with including this data was to provide robustness to our main results by using
data from previous studies that found different conclusions (Doidge, Karolyi, and Stulz (2007)).
FTSE calculated a corporate governance index for firms around the world from 2005-2008 called
the FTSE ISS Corporate Governance Index (CGI) Series. This index was composed of countries
from their Developed CGI, Europe CGI, Euro CGI, Japan CGI, UK CGI, and the US CGI.
Scores for the index were calculated several times a month for all companies. We used the
average from an entire year’s worth of scores. This yielded one unique score for each company
for each year. The FTSE data was heavily dominated by developed economy firms. In fact, only
6.2% of the observations come from emerging economies. These economies are few.
Specifically the emerging economy observations come from Hong Kong, Singapore, and
Thailand. The developed economies, on the other hand, are well represented. 93.8% of the
FTSE firms are located in developed economies. Summary statistics and correlations for this
data can be found in Appendix 7.
One possible limitation in the CLSA, GRI, and FTSE data sets is their Western
orientation. Both ratings are issued in the West and as such, are guided by Western norms
regarding good governance. It is therefore possible that non-Western-normed governance efforts
by firms and countries in our datasets are being initiated, but we fail to capture these changes.
Many of the emerging economies we examine in our dataset are in Asia. Firms in these
countries with low governance scores could appear less keen to improve governance, when, in
fact, their governance efforts are simply targeting points not focused on in the West. Indeed,
17
looking at Appendix 1 for country statistics, we see that Western and common law countries
often receive some of the highest scores. This is true for the United Kingdom, New Zealand, and
Australia. However, it is important to point out that many firms from non-Western countries
have relatively high scores. Some firms from Brazil, Thailand, and Mexico have among the
highest scores in the CLSA dataset. Reciprocally, in the GRI dataset, firms from developed
Western economies such as Greece, Portugal, and Luxembourg have some of the lowest
governance scores. It is thus important to note that our results go against the possible prediction
that only firms in Western and common law origin countries can receive top ratings. As
described above, firms from emerging, non-Western economies can have the best scores and
firms from developed, Western economies can have the worst scores.
Throughout much of our analysis, we differentiate between emerging and developed
economies because of the unique trends we uncovered for the two types of markets. To
operationalize the category of emerging economies, we referred to OECD membership by 1990.
Any country that was a member by this year was classified as a developed economy; countries
that were not members of the OECD by 1990 were classified as emerging economies. The GRI
dataset also included several small, island nations such as Bermuda and the Cayman Islands.
These countries are commonly understood as tax havens and have no OECD membership in
1990, so they were all classified as emerging. GRI results that exclude the tax havens can be
found in Appendix 6. There was a number of competing emerging economies lists published by
other analyst groups. Specifically, we considered lists published by FTSE, S&P, Internet
Securities, Inc., and Dow Jones. We also considered the list of countries commonly called “The
Next Eleven/BRIC” countries, but rejected the list as it is determined not only by economic
growth, but also by increasing political importance. This explains why Iran is a member of the
18
Next Eleven, but on no other emerging economies lists. In the end, we chose the OECD
membership definition for its ability to classify all countries in our dataset as either emerging or
developed. It is also the most moderate of the lists and avoids many of the outliers presented in
other lists.
a) Empirical Design
We estimate the sources of corporate governance ratings variation using a combination of
ordinary least squares, random effects, and nested ANOVA models. For our OLS models we run
two sets of regressions. The first set of OLS models looks at observable firm and country
characteristics to analyze to what extent specific variables explain corporate governance ratings.
The second set of OLS models adds firm and country fixed effects to look at the contribution
from the unobservable characteristics of firms and countries. Our random effects models use
xtmixed specifications. These models simultaneously explore observable and unobservable
firms and country characteristics’ contribution to explained variance. Random effects models
also take into account the hierarchical nature of the data, where firms are nested inside countries,
and thus allow us to run a single model instead of sequential models. Our third set of models
relies on nested ANOVA specifications. Here we look again at the importance of firms and
countries, and run the models sequentially as we did with our OLS models. This sequential
structure is what defined “nested ANOVA” rather than simply ANOVA. The dependent variable
for these models, as in all models, is the corporate governance score for each firm in our sample.
Because the CLSA dataset has one corporate governance score, the cgscore, and the GRI dataset
has two corporate governance scores, the index-based corporate governance quotient and the
industry-based corporate governance quotient, we run three sets of regressions.
In every regression except for one random effects model we include year variables to
control for fixed effects that vary over time. The independent variables in the regressions are
19
different combinations of sets of variables. The country variables are those used in previous
studies such as Doidge, Karolyi, Stulz (2007). Specifically, we used Antidirector x Legal, which
interacts the country's Revised Antidirector Rights Index with the Rule of Law in the country,
GDP per capita, and Stock Market Cap/GDP, which divides the country's entire stock market
capitalization by the GDP. Our firm variables include the original set of firm characteristics used
in previous studies: Sales Growth, which is measured on a two year basis, Financial
Dependence, which uses EBITDA to measure dependence on external financing, Closely Held
Shares, which is the percentage of total shares that are closely held, Log(Assets), and
Cash/Assets. In addition to these original variables, we include 17 additional firm characteristics
in order to capture any effects missing in the original, and somewhat sparse, set of variables.
The full list of the 17 additional firm characteristics can be found in the variable descriptions and
correlations for each data set.
In the OLS and nested ANOVA models we look at the additional Adjusted-R² to
determine how much ratings variance is explained by the variables in that model. Because our
data is necessarily hierarchical, with firms nested inside countries, we cannot simply look at the
R² without taking this into account. Thus, we show the contribution of firm and country
characteristics by adding each set of variables successively and looking at how much additional
rating variance is explained in each model. This means that we first look at the Adjusted-R²
from year fixed effects alone. Next, we keep year fixed effects but add country variables and
look at the additional Adjusted-R² from country variables. Our regression equation at this stage
can be written out as:
20
The dependent variable is always the relevant corporate governance score for the company i in
year t. The listed variables in equation (1), attached to coefficients β1, β2, and β3, are the
observable country characteristics. Year fixed effects are included in every model. The
vector γ represents the coefficients for each year effect. To evaluate how much firm variables
explain of ratings variance we next keep year fixed effects and the country variables, but add in
firm variables and analyze the additional Adjusted-R². When we include the original set of
observable firm characteristics our regression equation becomes:
Regression equation (2) preserves the elements of regression equation (1), but adds in several
observable firm-level variables. The next regression includes not only the firm-level variables
listed above, but also the 17 additional firm characteristics not included in previous studies.
Following our observable characteristics models, we look at unobservable characteristics.
Country and firm fixed effects models involve dummy variables for every country or firm, as the
model dictates. Equation (3) captures the fixed effects regressions where we include both
country and firm fixed effects. Vector α represents all stable characteristics of countries and ρ
represents all stable characteristics of firms, or their fixed effects. We include firm-country time
21
trends in the all fixed effects models as well, captured in γ. β1 and β2 represent the unique slopes
for each country and firm.
The additional R² from each combination of variables is calculated using the Adjusted-R²
to adjust for the number of observations. For the set of regressions that look just at observable
characteristics, the additional variance explained by country characteristics is the Adjusted-R² for
Model 2 minus the Adjusted-R² for Model 1, which includes only time trends. The additional
variance explained by the original, limited set of observable firm characteristics is the Adjusted-
R² for that Model 3 minus the Adjusted-R² for Model 2. For the full set of firm characteristics
including the additional 17 firm variables, we take the Adjusted-R² for Model 4 and again
subtract the Adjusted-R² for Model 2. For the fixed effects regressions, we proceed similarly.
The additional variance explained by unobservable country characteristics is the Adjusted-R² for
Model 5, which has country and year fixed effects, minus the Adjusted-R² for Model 1, which
has only year fixed effects. Looking at the unobservable plus the observable country
characteristics, we subtract the Adjusted-R² for Model 6, which has country and year fixed
effects as well as observable country variables, from that of Model 1, to remove the variance
explained by time trends again. The additional variance explained by unobservable firm
characteristics is the Adjusted-R² for Model 7, which has firm, country, and year fixed effects,
minus the Adjusted-R² for Model 5, which has country and year fixed effects only. And lastly,
looking at the additional variance explained from unobservable and observable firm
characteristics, we take the Adjusted-R² in Model 8 and subtract the Adjusted-R² in Model 6,
22
which also includes both unobservable and observable country characteristics. Thus, our method
is to add on first time, then country, and then firm variables in various combinations to observe
the additional corporate governance ratings variance explained by each set of variables.
We also run random effects models and nested ANOVA regressions to support our initial
results. The random effects model allows us to run a single regression that accounts for the
hierarchical nature of the data and shows the variance contribution of firm and country
individually. Our random effects models use the xtmixed command and look like equation (3),
but instead of α and ρ being the fixed parameters, they are random variables. This allows for
variation within and between companies and countries. We run two random effects model, one
with and one without year fixed effects, Models 9 and 10, respectively in each set of regressions.
Finally, we run a series of models using nested ANOVA, or analysis of variance, specifications.
Nested ANOVA models again account for the hierarchical nature of the data and look at the
additional Adjusted-R² contributed by firm and country effects. ANOVA also allows us to treat
the country and company effects as categorical variables and compare the contribution of each to
overall ratings variance. Similar to the OLS models, our ANOVA models are run successively
to look at the additional Adjusted-R² contributed by each category such as country or firm. As in
our OLS models, we first run a model with only time trends, then a model with time and country
effects, and finally a third model with time, country, and firm effects.
IV. Empirical Results
Our results consistently show that, in emerging economies, firms are anywhere from
equal in importance to significantly more important than countries in explaining corporate
23
governance ratings variance. This finding is consistent regardless of which dataset and
econometric method we use.
Over the three dependent variables (CLSA cgscore, GRI index-based cg score, and GRI
industry-based cg score) in emerging economies, we see that in our fixed effects specifications
firm characteristics explain 37.9-43.8% of the ratings’ variance while country characteristics
explain only 14.4-19.4%. Our random effects, xtmixed, models show again that firm
characteristics are as important if not more important than country characteristics for corporate
governance ratings variation. Here, firms explain 37.3-50.3% of ratings variance while firms
explain 11-28.5%. The nested ANOVA results once more confirm the trend for emerging
economies. In these models additional Adjusted-R² from firms ranges from 40.5-43.6%, while
the additional Adjusted-R² from countries is between 11.5-16.2%.
We also look simply at the observable firm and country effects, evaluated in Model 4 of
our regressions. This model includes the observable country characteristics, the original set of
firm characteristics, as well as our 17 additional firm characteristics. The results here show weak
contributions from firms to corporate governance variance in emerging economies. Thus, the
results from Model 4 are incongruous with the rest of the results we find in this paper. We take
this as evidence that something unobservable is happening inside firms that allows them to affect
their corporate governance rating; this unobservable process is often not captured in specific,
observable firm variables used. This interpretation of the results is bolstered by the overall low
amounts of variance explained by firms and countries when just using observed specific
variables. Firm characteristics explain -1.3-6.0% of variance while country characteristics
explain 5.2-9.5%. Models that employ fixed effects, random effects, and ANOVA specifications
24
are therefore necessary to properly measure this unobservable contribution by firms to corporate
governance variance.
We determined that emerging and developed economies should be evaluated separately
when we initially ran our two main datasets. We found unique trends in the CLSA and GRI
datasets. In the CLSA data, firms contributed roughly equal amounts to governance ratings’
variance and often more. In the GRI data, in contrast, countries contributed more to the
variation. We soon realized that the reason for these differing results was the composition of
each dataset. The CLSA data was almost entirely composed of emerging economies while a
majority of the GRI observations came from developed economies. Once we understood the
difference and we separated the data into developed and emerging economies using a consistent
definition we found strong and reliable results in both datasets. These results were also
confirmed with the addition of the FTSE data, which followed the same pattern.
In developed economies we find that country-level variables consistently explain
substantially more of the ratings’ variance than firm-level variables. These results come solely
from the GRI dataset, as the CLSA data contained too few developed economy observations for
reliable results. The results are confirmed by the FTSE dataset, however. The sample size for
the regressions ranged from 53-207 when we looked just at CLSA developed economies.
Therefore, we focused on the GRI dataset for developed economies and saw that, for the
observable and unobservable characteristics using fixed effects models, countries explain 56.0-
57.2% of the ratings’ variance while firms explain only 15.4-17.1%. In our random effects
models, we see the result that country random effects explain 48.1-46.0% of the ratings variance
while firm random effects explain roughly 18.7-19.1%. The nested ANOVA results for GRI
25
developed markets are similar. Countries explain 55.9-57.3% of the ratings variance and firms
explain roughly 15.3-15.5%.
The results from the CLSA emerging markets data are found in Table II. The data used
to calculate the results in this table include almost all of the original data in the CLSA dataset.
We excluded developed economies in order to cleanly evaluate only emerging economies. The
observations excluded range from 58-207 observations, depending on the model. Considering
that the overall dataset includes over 4,000 observations, excluding these developed economies
does not bias our sample in any direction. In Models 1-4 we build in the different year effects,
the country variables, the limited firm variables, and then the expanded firm variables to see the
contribution of each to the Adjusted-R². The results from Model 2 show that by adding country
variables we can explain an additional 5.2% of the CLSA corporate governance ratings variance.
Adding the original set of limited firm variables does not add any explanation of variance in
Model 3, but when we include the expanded set of firm variables in Model 4, we see that firm
variables in total explain 6.0% of variance on top of what the country variables explain.
In Models 5-8 we look at fixed effects in addition to specific firm and country variables.
Model 6 shows that, including country variables as well as country fixed effects explains 14.4%
of the ratings variance over what year fixed effects explains alone. However, looking at the full
set of firm variables and firm fixed effects, we see that firm characteristics account for 41.0% of
the ratings variance. The strong results and high Adjusted-R² in the firm fixed effects models (7-
8) suggest that much of what is important and happening at the firm level in emerging economies
is unobservable.
The random effects models echo the results found in previous models, and confirm again
that firms are more important in emerging economies. In the model with year fixed effects,
26
Model 10, we see that company random effects explain 37.3% of the ratings variance while
country random effects explain 26.4%. In all of these models we look only at the Adjusted-R²,
which ensures that the rise in R² is not simply due to the larger set of variables in the company
fixed effects and random effects regressions.
In Panel B we confirm the results found with OLS fixed and random effects models using
nested ANOVA specifications again on the CLSA emerging economies data. These results were
consistent with those reported in Panel A of the same table. We see that, when using nested
ANOVA the amount of ratings variance in emerging economy firm governance explained by
firm characteristics is greater than that explained by the country characteristics. Specifically,
according to the nested ANOVA models, 41.4% of ratings variance is explained by the firm
effects while only 11.5% is explained by country effects. Overall, the picture from Table II is
consistent: Firm characteristics explain significantly more of CLSA corporate governance ratings
variance as country characteristics once we include unobservable firm characteristics.
The results from the GRI dataset both corroborate what was found when looking at the
CLSA data, and add new understanding to why previous studies perhaps found such different
results. The OLS and xtmixed results from the GRI emerging economies data are listed in
Tables III. Panel A uses the index-based corporate governance quotient compiled by GRI while
Panel B uses their industry-based corporate governance quotient.
The results in Panels A and B can be evaluated as those were for the CLSA data. They
also show the same pattern of firms in emerging economies explaining as much if not more of
the corporate governance variance as do countries. First, we look at the specific firm and
country variables in Models 1-4. In Model 2 we see that observable country variables explain
9.5% of the index-based CGQ variance, and 7.3% of industry-based CGQ variance. Model 4
27
shows that by adding the full set of firm characteristics we explain -1.3% and 2.4% of additional
CGQ ratings variance for the index- and industry-based scores. In Model 8 we include firm and
country fixed effects for unobservable firm and country characteristics. Including the full set of
firm variables and firm fixed effects in Model 8 adds 37.3% in Panel A and 43.8% in Panel B to
variance explained. In our random effects model with year fixed effects, Model 10, we see that
firm effects contribute 37.8% in Panel A and 50.3% in Panel B to explaining corporate
governance variance. Countries explain much less: 28.5% in Panel A and 11% in Panel B. Thus,
Table III echoes the greater importance of firm characteristics, relative to country characteristics
found in emerging economies.
We further test these results by running nested ANOVA models on the GRI data, relying
on the same separation of the data into emerging and developed economies. The ANOVA
emerging economy results can be found in Panels C and D of Table III. For the index- and
industry-based corporate governance quotients, our ANOVA results exhibit the same pattern
witnessed in the OLS and xtmixed models. Countries explain only 16.2% and 11.5% additional
variance in Panels A and B, while firms explain an additional 43.6% and 40.5% of the variance.
Taken together with the previous tables, we see consistently that, in the GRI dataset when we
look at just the emerging economies, firm characteristics explain roughly equal to a greater
amount of variance in the CGQ than countries do.
In contrast to the emerging economy trends, the developed economy observations in the
GRI dataset suggest that country-level variables are far more important at explaining ratings
variance in that setting. To show this, we used the other portion of the GRI data that is made up
entirely of developed economies. We then re-ran the same analyses done on the emerging
economies. The results in all models using developed economy data are substantially different.
28
Unobservable plus observable country characteristics account for 46.0-57.3% of the ratings
variance in developed economies depending on the model and the use of the index- or industry-
based CGQ. Firm characteristics explain far less, ranging from 15.3-19.1% of the variance. The
results for the developed economies can be found in Table IV. OLS and xtmixed results are in
Panels A and B while ANOVA results are in Panels C and D.
We progress through the results of Table IV much as we did for Table III. First we can
look at the observable firm and country characteristics in Model 4 of Panels A and B. For the
index-based CGQ in Panel A, observable country characteristics explain 38.0% of the variance
while observable firm characteristics explain only 8.9%. Similarly, for the industry-based CGQ
in Panel B, observable country characteristics explain 41.7% of the variance while observable
firm characteristics explain 7.1%. In Model 8 where we include all of the observable and
unobservable firm and country effects, countries explain 56.0% and 57.2% of the variance while
firms explain only 17.1% and 15.4%. Random effects and ANOVA results also strongly reveal
the importance of countries in developed economies. In Model 10 of Panels A and B, we see
that countries explain 46.0% and 48.1% of the index- and industry-based CGQ’s, respectively,
while firms explain 19.1% and 18.7%. In Panels C and D, where the ANOVA results are located,
countries explain 55.9% and 57.1% of the ratings variance while firms explain 15.5% and
15.3%.
Therefore, overall, the results show that there are two distinct trends for emerging and
developed economies. In emerging economies, firms explain roughly equal to significantly more
of the corporate governance ratings variance than countries; in developed economies, countries
explain more of the ratings variance than firms. The variation between and within the two
datasets allowed us to discover, test, and then prove these two trends. Using OLS on observable
29
firm and country characteristics and then combinations of observable and unobservable
characteristics with fixed effects, random effects, and ANOVA specifications, we show that
these trends are consistent throughout.
Finally, we tested our conclusions against those found in previous studies by running our
models on one of the primary datasets used in the earlier work, FTSE's Corporate Governance
Ratings. Specifically we compare our results to literature that finds a much stronger role for
country characteristics in determining corporate governance ratings using the exact same data
that literature uses. One of these papers is Doidge, Karolyi, and Stulz (2007), which showed that
country characteristics were roughly 10 times more important than firm characteristics in
explaining governance variation. We are able to compare our results to theirs by using the same
dataset they used, but extending it over 4 years for analysis over time. Their work uses a single
year of data from FTSE. We acquired this panel dataset from FTSE's Corporate Governance
Ratings Index, which was calculated for four years from 2005-2008. Comparing the summary
statistics of our FTSE datasets confirms that our FTSE data is similar to that in the Doidge,
Karolyi, and Stulz (2007) paper in that it is almost entirely composed of developed markets, and
Japan and the UK are heavily represented. These results can be found in Appendix 7.
The results for the emerging economy data from FTSE can be found in Table V. The
Models lost due to insufficient observations are 4 and 8, those that include the full set of 22
specific firm variables. We can however, get a clear, fast, and reliable picture of this data by
looking at the random effects and ANOVA results, both of which use all 607 emerging economy
observations. In Model 10, the random effects results show that firms explain 34.3% of the
ratings variance while countries explain 15.8%. In Panel B, the ANOVA results show that firms
explain 32.2% of the ratings variance while countries explain only 16.3%.
30
The developed economy results from the FTSE data are in Table VI. Like the other data,
the relative importance of countries and firms switch for the FTSE developed economies. In
Model 2 we see that observable country characteristics explain an additional 59.4% of the
variance on top of what years explain while firms only explain -2.6% in Model 4. When we
include the unobservable characteristics, countries contribute 65.7% on top of what years
contribute. Firms only explain an additional 13.1%. Random effects and ANOVA models
echoes these results. In Model 10 of Panel A in Table 6, we see that countries explain 52.7% of
the variance while firms account for only 24.4%. In the nested ANOVA results in Panel B,
countries explain 65.5% of the variance while firms explain an additional 17.6% of the variance.
Therefore, our empirical approach, when applied to the FTSE data, yields results that do not
differ from the pattern found in the other datasets. Firms explain more of the ratings variance in
emerging economies and countries explain more variance in developed economies.
V. Robustness
We test our results using a variety of checks and they remain robust. The tests,
specifically, explore the importance of multinationals, corrupt regime relationships, county
dominance of results, the distribution of emerging and developed economy scores, and finally
the importance of industry.
In both the CLSA and the GRI datasets there are a number of multinational firms. These
firms either have independent subsidiaries in markets that enable them to be evaluated as local
firms or their headquarters are in the given country. The multinationals are traded under unique
tickers, but still they often bear the name of a multinational company and may have involvement
with other subsidiaries and/or the headquarters. To understand the importance of being a
31
multinational in an emerging economy, we matched all companies in the CLSA dataset to those
firms listed in the Directory of Corporate Affiliations (DCA). We then looked at whether these
firms are multinationals and how many subsidiaries their parent company has. In the CLSA
data, roughly 39% of the 4,448 observations are multinationals, totaling 1,747 observations. The
number of subsidiaries varied from 0 to 91; the average number of subsidiaries for multinationals
in this datasets is 1.5 with a standard deviation of 7.25. Looking at the firms confirmed our
earlier results. With or without multinationals the firm effect is larger than the country effect in
emerging economies to varying degrees.
Using our DCA matching to distinguish multinationals and single-market firms, we re-
ran our models for both sets of firms in emerging economies. The results from these models can
be found in Appendix 2. Across the board, we saw that the effect of firm characteristics is
stronger for emerging economy multinationals. For non-multinationals in emerging economies,
company characteristics are still slightly more important than country characteristics, but the
effect is smaller than for all emerging economy firms. This result fits well with the intuition that
firm characteristics are an important part of understanding corporate governance ratings for
emerging economies. Firms in multiple markets may have to comply with all relevant sets of
government regulations regarding governance, even if they have largely independent
subsidiaries. As well, it could be that corporate governance improvements are being driven by
corporate headquarters, even in markets with weak local institutions. Both of these effects could
possibly be driving multinational firms to have higher governance ratings and to exhibit greater
importance for country characteristics. Further research could serve to explain this phenomenon.
We also considered the possibility that firms in emerging economies may not be
motivated to improve their corporate governance if they benefit from close ties to a corrupt
32
regime. If this were the case, our results would be biased and would not accurately reflect the
capacity of firms to improve corporate governance in order to improve their access to capital. In
order to test this possibility, we re-ran all models where the SEC compliance variable is included
for both datasets. The presence of this compliance should distinguish firms that are able to
improve their standing through corrupt means and those that do so by improving corporate
governance measures in keeping with our model. In short, we would expect the results to be
different if the SEC compliance variable is both significant and if its inclusion or exclusion shifts
our results.
Results from this modified test show that the SEC compliance variable is never
significant for the CLSA data, and is only significant occasionally for the GRI dataset. The
unimportance of SEC compliance suggests that firms do not ignore governance because of
relationships with corrupt regimes. Firms in emerging economies are determining their
governance regardless of ties to corrupt governments. These results are corroborated further by
additional tests of the models with and without the SEC compliance variable. In all cases
including or excluding this variable does not change our results significantly. The Adjusted-R²
of the models changes by less than one one-hundredth of a point when we include and then
exclude the SEC compliance variable. Taken together, these findings add weight to our
interpretations that the results we find are driven by emerging economy firms looking to improve
their access to capital by improving their corporate governance ratings, not through relationships
with corrupt political regimes.
In dividing the data into emerging and developed economies, there was a risk that a
specific country or type of country was responsible for the different trends in the GRI dataset and
in developed economies. To test this question, we ran our models again, this time excluding
33
countries individually, then two at a time, three at a time and then four at a time. We examined
these results to see if excluding certain countries affected the relative importance of countries as
compared to firms. These results showed that excluding any combination of countries fails to
remove countries as the more important predictor of corporate governance ratings in developed
economies. Yet, certain combinations did weaken the effect. Specifically, excluding Japan and
the United Kingdom together showed the most dramatic decrease in the relative importance of
countries. When the models are run on all developed economies without the UK and Japan, the
firm effect is larger. These two countries are also the two largest sets of observations in the
developed economies dataset. Japan composes 4,145 observations of the 13,977 developed
economy observations, while the United Kingdom is another 3,022. Interestingly, their average
scores differed considerably. For the index-based score, Japan has an average score of 26.7
while the average score for the UK is up at 83.7.
We ran the models on all developed economy observations except for the UK and Japan
for both the index-based as well as the industry-based scores. These results can be found in
Appendix 3. We see that the importance of country effect generally drops somewhat and the firm
effect rises only slightly. To determine if there was a pattern where the worst firms in the UK
and Japan are rated higher than elsewhere, perhaps because of analyst biases, we looked at the
skewness and kurtosis of other countries. Looking only initially at the index-base quotient we
found that the average skewness for all developed economies is -0.01 while the average kurtosis
is 1.8. The United Kingdom has the longest left tail for its distribution at -1.7. Most other
countries hovered between -0.5 and 0. Japan was slightly positive at 0.7. The kurtosis was
somewhat starker, however. Most developed economies’ skew ranges between 1 and 3. Japan is
close at 3.2, but the United Kingdom is up at 9.0. This suggests that the distribution for the UK
34
firms could be driven by infrequent, extreme, and positive deviations from the average. In other
words, the United Kingdom is getting the highest scores of any country. Whether this is because
UK firms include some of the corporate governance stars or whether analysts are biased towards
particular UK firms is difficult to tell from this analysis. Still, the results here do not present a
refutation of our overall findings. The relative importance of country for developed economies
remains even in our restricted dataset without the UK and Japan.
In spite of the mildly weaker importance of countries in explaining governance ratings
variance in emerging economies, there was no combination of 3-5 developed countries that,
when excluded, causes the country effect to be smaller than the firm effect. This confirmed that
groups or types of countries are not driving the unique developed economy results. At most, our
findings regarding the United Kingdom and Japan suggest that governance ratings bodies appear
to giving systematically different ratings to certain developed countries. We show this in
Appendix 1, which lists the summary statistics by country, showing mean, standard deviation,
median, min, max, 25th percentile, and 75th percentile for the corporate governance scores by
country. We only included the CLSA scores and GRI index-based score as the GRI industry-
based score is similar enough to the index-score that trends can be seen by just looking at one of
the two. As should be expected in this table, we see higher ratings for developed economies.
As another robustness check, we considered whether developed economy firms are
simply at the corporate governance quality frontier while emerging economy firms range from
the lowest to the highest governance performance. The concern here was that perhaps emerging
economy firms are only able to rise above their country averages in just a few instances. If this
were the case, then developed economies firms would have higher scores than emerging
economy firms on average. To explore this possibility, we compared the means and variances
35
for all of our corporate governance scores. We find, as expected, that the mean for developed
economy firms is higher, but not significantly higher than it is for emerging economies.
Specifically, for the CLSA data, the mean for emerging economies is 54 while it is 59 for
developed economies. In the GRI data, the mean for emerging economies’ index governance
scores is 59.2, while the developed economies’ mean is 44.8. Thus, the average corporate
governance ratings of firms in emerging and developed economies do not differ substantially.
In addition to emerging and developed economies having roughly similar mean scores,
the distribution of scores in the two types of countries also suggests that firms in emerging
economy firms are capable of rising to world-class governance ratings in more than just a few
cases. To show this, we looked at the scores of two of our most well populated emerging
economies in the CLSA dataset: India and Hong Kong. Details of their summary statistics can
be found in Appendix 1. India has 641 observations while Hong Kong has 740. These countries
have roughly average country scores for emerging economies at 59.8 for Hong Kong and 52.4
for India. However the standard deviation in the scores for these countries is similar to the
standard deviation for developed economies. Hong Kong and India are around 13 while
developed economy scores (with observations greater than 7) have an average standard deviation
of 15. Although there are no India observations in the GRI data, the Hong Kong scores from the
GRI dataset are similar. The mean index-based CGQ is 39.7 while the average developed
economy score is 44.8. The standard deviation for Hong Kong, 16.8, is only slightly lower than
the developed economy standard deviation average, 20.9.
The percentile scores, as well, show that emerging economy firms range beyond their
country averages. In Hong Kong and India the 25th percentile scores are 52.9 and 43.4 while the
average 25th percentile score for developed economies is 49.7. The 75th percentile scores for
36
Hong Kong and India are, respectively, 67.8 and 61.1 while the average developed economy 75th
percentile score is 68.6. These developed economy averages again exclude those countries with
3 or fewer observations. The 25th and 75th percentiles for Hong Kong in the GRI dataset, 24.6
and 53.2, are very close to the average for all developed economies, 28.3 and 59.7, respectively.
These standard deviations and percentile scores show that the corporate governance scores for
developed and for emerging economies range between the best and the worst. Neither one has a
monopoly on the corporate governance quality frontier and emerging economy firms are
generally able to achieve the highest corporate governance scores in more than a limited number
of cases.
In addition, we evaluated the importance of industry in determining governance ratings
within each dataset. To do so we re-ran our nested ANOVA models using two- and three-digit
SIC codes as an intermediate level of analysis. This means, for our ordered analysis, we first ran
ANOVA for year effects; then country and year; then industry, country and year; and then finally
company, industry, country, and year. The results from these additional models are included in
Appendices 4 and 5. The results show that industry is somewhat important, especially for
emerging economies when we look at the three-digit SIC codes. However, when looking at our
main question of interest, the importance of companies in explaining emerging economy
corporate governance, we see that the firm effects remain dominant over country effects no
matter what industry specification is added. The firm variable explains more of the variance than
the country variable. This shows that firm effects are not capturing so much of an industry effect
that they are insignificant on their own. In developed economies, industry is less important
overall, suggesting that industries play a less important role in explaining corporate governance
ratings in developed economies. In these models the main result from previous models that do
37
not include industry effects hold as well: firms explain roughly equal amounts of variance in
emerging economies and countries explain more of the variance than firms in developed
economies.
VI. Conclusion
The results from our multiple specifications of firm and country characteristics provide
strong evidence that firm-level variables play an important role in explaining corporate
governance ratings in emerging economies. Prior work by Doidge, Karolyi, and Stulz (2007)
and others stated that country effects were dominant. However, by looking at panel data and
allowing unobservable firm characteristics to explain variation of firms' corporate governance
ratings with fixed effects, random effects, and nested ANOVA models, we show that firm effects
in emerging economies are as important, and often more important, than country effects are in
explaining ratings variance.
Previous results showing that firms are less important than countries were likely due to an
over-representation of developed economy firms in the dataset, failure to capture trends over
time with panel data, or failure to explore a wider range of observable and unobservable firm
characteristics. In order to compare our results more closely to those in the Doidge, Karolyi, and
Stulz (2007) paper, we attempted to recreate their results from just the 2001 data. Because Rule
of Law data was not listed for 2001, we used Rule of Law values from 2000, and where that did
not exist, 2001. Although our replicated results are the exact same to those in Doidge et al., the
results were nearly identical when we ran the same models they used. We then ran our new
model specifications that we used in this paper on this single year of data. We see that, just
looking at 2001, firms are still anywhere from comparable to meaningfully more important than
38
countries variables. In the OLS regressions, firms are slightly more important and in the xtmixed
regressions, firms are statistically equivalent to countries. Doidge et al found much of the same
and concluded that this result was due to lower variance among CLSA countries than in other
datasets. We theorize that the difference in conclusions could be for several reasons. First, the
other data sets Doidge et al. used were dominated by developed economies. 315 of the 711 S&P
observation are from developed markets and 1159 of the 1217 FTSE observations are from
developed economies. Second, their models do not account for the nested nature of the data by
first looking at countries and then adding in firms. Third, they only look at unobservable country
characteristics by including fixed effects and do not use firm fixed effects to capture
unobservable firm characteristics. And lastly, the results we found for 2001 differed slightly
from the trend we found over the entire decade that data was gathered. This suggests that 2001
could have been a unique year and those time trends were not accounted for using the cross-
sectional data.
The importance of firm-level characteristics in these results overall shows that firms in
emerging economies during the last decade had the ability to move separately from their home
country peer firms in their corporate governance ratings. Moving forward, this suggests that
firms in emerging economies have the capability to rise above home country institutions that
may be lacking or to distinguish themselves from their peer firms to both improve corporate
governance ratings, and hopefully attract greater levels of capital and grow. While the country in
which the firm is based is still important, there is agency beyond location for firms.
Differences between our CLSA data and the GRI data imply that there are unique
attributes to the different institutional and financial environments in emerging and developed
economies. Much attention has already been given to emerging economies; future research
39
could explore the mechanisms driving country importance in developed economies. As well,
future research could work to locate and test an exogenous shock to any of the firm and country
characteristics here to try to identify causality. Currently, our results are sufficient to show a
strong correlation and relationship, but only to hint at causality. By using panel data over 10
years, our results provide a stronger suggestion of causality, but a natural experiment and
subsequent analysis of corporate governance ratings would be better evidence. Such a study
could be undertaken at the country level, as shocks to the variables listed here across multiple
countries or regions would be unlikely. The results from work suggested here could help to
unpack exactly what is happening in the company and country fixed and random effects that
suggest a strong role for firms in corporate governance ratings in emerging economies and an
even stronger role for countries in developed economies’ ratings.
40
REFERENCES
Acemoglu, Daron, Simon Johnson, and James A. Robinson, 2001, The Colonial Origins of Comparative Development: An Empirical Investigation, American Economic Review, 9, 1369-1401.
Atanasov, Vladmir, Bernard Black, Conrad Ciccotello, and Stanley Gyoshev, 2010, How Does Law Affect Finance? An Examination of Equity Tunneling in Bulgaria, Journal of Financial Economics, 96, 155-173.
Bae, Kee-Hong, and Vidhan K. Goyal, 2010, Equity Market Liberalization and Corporate Governance, Journal of Corporate Finance, 16, 609-621.
Bebchuk, Lucian A., and Mark J. Roe, 1999, A Theory of Path Dependence in Corporate Ownership Governance, Stanford Law Review, 52, 127-170.
Black, Bernard S., Hasung Jang, and Woochan Kim, 2006, Does Corporate Governance Predict Firm Values?, Journal of Law, Economics, and Organizations, 22, 366-413.
Black, Bernard S. and Vikramaditya S. Khanna, 2007, Can Corporate Governance Reforms Increase Firms’ Market Values: Event Study Evidence from India, Journal of Empirical Legal Studies, 4.
Black, Bernard S., Inessa Love, and Andrei Rachinsky, 2006, Corporate Governance Indices and Firms’ Market Values: Time Series Evidence from Russia, Emerging Markets Review, 7, 361-379.
Blass, Asher A., and Yishay Yafeh, 2001, Vagabond Shoes Longing to Stray: Why Foreign Firms List in the United States” Journal of Banking and Finance, 25, 555-572.
Chen, Kevin C.W., Zhihong Chen, and K.C. John Wei, 2009, Legal Protection of Investors, Corporate Governance, and the Cost of Equity Capital, Journal of Corporate Finance, 15, 273-289.
Coffee, John C. Jr., 2002, Racing Towards the Top? The Impact of Cross-Listings and the Stock Market Competition on International Corporate Governance, Columbia Law Review, 102, 1757-1831.
Doidge, Craig, Andrew Karolyi, and Rene Stulz, 2007, Why do Countries Matter so Much for Corporate Governance?, Journal of Financial Economics, 86, 1-39.
Durnev, Artyom, and E. Han Kim, 2005, To Steal or not to Steal: Firm Attributes: Legal Environment, and Valuation, Journal of Finance, 60, 1461-1493.
Engelen and Essen, 2010, Vulnerability to the Financial Crisis: The Role of Firm and Country Level Governance Mechanisms in Europe, Journal of Banking and Finance, 34, 1958-1969.
Fan, Joseph P.H., and T.J. Wong, 2002, Corporate Ownership Structure and the Informativeness of Accounting Earnings in East Asia, Journal of Accounting and Economics, 33, 401-425.
Himmelberg, Charles P., R. Glenn Hubbard, and Darius Palia, 1999, Understanding the Determinants of Managerial Ownership and the Link Between Ownership and Performance, Journal of Financial Economics, 53, 353-384.
Johnson, Simon, Peter Boone, Alasdair Breach, and Eric Friedman, 2000, Corporate Governance in the Asian Financial Crisis, Journal of Financial Economics, 58, 141-186.
Klapper, Leora F., and Inessa Love, 2004, Corporate Governance, Investor Protection, and Performance in Emerging Markets, Journal of Corporate Finance, 10, 703-728.
41
Klapper,Leora F., Luc Laeven, and Inessa Love, 2006, What Drives Corporate Governance? Firm –level Evidence from Eastern Euorpe, World Bank Policy Research Working Paper Series, 3600.
Krishnamurti, C.. A. Sevic, and Z. Sevic, 2006, Legal Environment, Firm-Level Corporate Governance, and Expropriation of Minority Shareholders in Asia, Economic Change and Restructuring, 38, 85-111.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny, 1997, Legal Determinants of External Finance, Journal of Finance, 52, 1131-1150.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny, 1998, Law and Finance, Journal of Political Economy, 106, 1113-1155.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny, 1999, Corporate Ownership Around the World, Journal of Finance, 54, 471-517.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny, 2000, Investor Protection and Corporate Governance, Journal of Financial Economics, 58, 3-27.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny, 2002, Investor Protection and Corporate Valuation, Journal of Finance, 57, 1147-1170.
Lemmon, Michael L., and Karl Lins, 2003, Ownership Structure, Corporate Governance, and Firm Value: Evidence from the East Asian Financial Crisis, Journal of Finance, 58, 1445-1468.
Mitton, Todd, 2002, A Cross-Firm Analysis of the Impact of Corporate Governance on the East Asian Financial Crisis, Journal of Financial Economics, 64, 215-241.
Reese, William A. Jr., and Michael S. Weisbach, 2002, Protection of Minority Shareholder Interests, Cross-Listing in the United States, and Subsequent Equity Offerings, Journal of Financial Economics, 66, 65-104.
Sawicki, Julia, 2009, Corporate Governance and Dividend Policy in Southeast Asia Pre- and Post-Crisis, European Journal of Finance, 15, 211-230.
Shleifer, Andrei, and Robert W. Vishny, 1997, A Survey of Corporate Governance, Journal of Finance, 52, 737-783.
Siegel, Jordan I., 2005, Can Foreign Firms Bond Themselves Effectively by Renting US Securities Laws?, Journal of Financial Economics, 75, 319-359.
Variable Median Mean St Dev. Min Max Observations
Corporate Governance Score 55.100 54.032 14.973 0.000 97.640 3,973
Sales Growth 15.127 28.630 165.951 -98.302 5,432.525 3,840
Financial Dependence -1.186 -5.365 133.150 -6,367.904 360.323 2,626
Closely Held Shares 52.629 50.185 23.871 0.000 147.826 3,478
Log (Assets) 14.664 14.741 1.793 8.935 21.633 3,861
Cash/Total Assets 0.124 0.176 0.565 0.000 31.405 3,307
Antidirector x Legal 2.466 3.011 3.400 -4.003 8.529 3,496
GDP per capita 4,458.562 12,809.490 13,505.660 452.969 40,707.000 3,484
Stock Market Cap / GDP 90.006 163.006 177.236 5.057 617.046 3,484
Fixed Assets/Total Assets 0.315 0.339 0.250 0.000 0.994 3,853
SEC Compliance 0.000 0.105 0.306 0.000 1.000 3,999
Assets/Liability 1.483 2.061 2.400 0.098 61.553 3,226
Leverage 0.499 0.574 3.598 0.000 220.419 3,757
Tobins Q 0.980 1.599 3.805 0.004 124.362 3,693
Total Sales 0.000 14.371 56.816 -3,267.490 134.220 4,447
2yr Sales Growth 0.000 75.054 2,289.554 -147.182 121,066.700 4,447
PE Ratio 13.514 37.121 1,054.877 -1,840.714 65,000.000 3,827
Price-to-book Ratio 1.804 2.714 4.068 -75.405 70.667 3,847
Quick Ratio 1.014 1.555 2.301 0.010 60.717 3,269
Return on Assets 7.236 8.528 9.786 -82.753 137.573 3,850
R&D Intensity 0.000 0.008 0.027 0.000 0.603 4,447
CapitalExpenditure 5.592 11.093 48.591 0.000 853.855 3,793
Cash/Dividends 4.280 116.883 3,651.804 -602.517 187,367.100 2,877
3yr Dividends Growth 10.064 11.167 46.997 -100.000 848.373 3,173
5yr Income Growth 16.562 22.590 35.360 -75.566 654.618 2,789
5yr Sales Growth 16.760 22.245 30.922 -58.871 753.489 3,255
Short-Term Debt 119.433 1,587.177 10,842.960 0.000 292,020.600 3,885
5yr Assets Growth 14.885 20.358 24.796 -48.476 314.692 3,230
Total Debt 48.048 78.442 230.447 -7,457.181 2,301.523 3,965
Table I
The following table gives the summary statistics for the Credit Lyonnais Securities Asia (CLSA) data and its companies. This dataset
encompasses 10 years of CLSA tracking corporate governance performance of firms in emerging economies. The first variable is the CLSA
given corporate governance score. The next three variables are the three observable country characteristics used in our analysis.
Antidirector x Legal captures the interaction of the Revised Antidirector Rights Index and the Rule of Law in that country. The following firm
variables include the observable firm characteristics included in previous studies (sales growth, financial dependence (EBITDA based), closely
held shares (as a percent of total shares), log(assets), and cash to assets ratio). The remaining variables described below are additional
observable firm characteristics used to capture the complex interaction between firms and corporate governance in emerging economies.
Panel A: CLSA Variable Descriptions
Corporate
Governance Score
Sales
Growth
Financial
Dependence
Closely Held
Shares
log
(Assets)
Cash/ Total
Assets
Antidirector x
Legal
GDP per
capita
Stock Market
Cap / GDP
Fixed/ Total
Assets
SEC
Compliance
Assets/
Liabilities Leverage Tobins Q
Total
Sales
Corporate Governance Score 1.000
Sales Growth -0.015 1.000
Financial Dependence -0.012 0.042 1.000
Closely Held Shares -0.135*** 0.034** -0.015 1.000
Log(Assets) 0.028* -0.026 -0.032 -0.067*** 1.000
Cash/Total Assets 0.031* 0.018 -0.057*** 0.005 -0.118*** 1.000
Antidirector x Legal 0.170*** 0.028 0.018 -0.066*** 0.045*** 0.058*** 1.000
GDP per capita 0.123*** 0.037** 0.024 -0.077*** 0.081*** 0.068*** 0.923*** 1.000
Stock Market Cap / GDP 0.028 0.066*** 0.022 0.077*** -0.034* 0.048*** 0.679*** 0.724*** 1.000
Fixed Assets/Total Assets -0.081*** -0.032* 0.043** 0.140*** -0.050*** -0.128*** -0.045*** -0.067*** -0.088*** 1.000
SEC Compliance 0.096*** 0.018 0.013 -0.008 0.182*** 0.004 -0.035** -0.009 -0.091*** 0.160*** 1.000
Assets/Liabilities -0.004 0.005 -0.025 -0.020 -0.227*** 0.101*** 0.082*** 0.111*** 0.138*** -0.277*** -0.028 1.000
Leverage 0.018 0.002 -0.049** 0.012 -0.008 0.956*** 0.017 0.010 -0.010 -0.028* -0.007 -0.029 1.000
Tobins Q 0.077*** 0.035** 0.014 0.017 -0.208*** 0.288*** -0.007 -0.016 -0.008 0.015 0.033** 0.024 0.271*** 1.000
Total Sales 0.028* 0.003 0.005 -0.010 0.009 0.004 0.123*** 0.139*** 0.133*** -0.036** 0.010 0.051*** -0.007 0.011 1.000
2yr Sales Growth 0.023 0.001 0.001 -0.021 -0.002 -0.004 -0.003 -0.008 -0.013 0.004 0.031** -0.010 -0.001 -0.001 0.021
PE Ratio -0.014 -0.019 0.001 0.008 -0.028* -0.004 0.028 -0.015 -0.012 0.013 -0.007 -0.010 0.000 -0.013 0.002
Price-to-book ratio 0.075*** 0.039** 0.015 0.057*** -0.157*** 0.171*** -0.004 -0.030* 0.037** -0.098*** -0.010 -0.008 0.131*** 0.094*** 0.015
Quick Ratio 0.010 0.010 -0.008 -0.025 -0.201*** 0.118*** 0.108*** 0.131*** 0.155*** -0.228*** 0.009 0.951*** -0.025 0.034* 0.056***
Return on Assets 0.077*** 0.056*** 0.021 0.088*** -0.273*** 0.136*** -0.024 -0.034** 0.023 0.001 0.004 0.163*** 0.050*** 0.243*** 0.056***
R&D Intensity 0.018 -0.015 0.008 -0.174*** -0.133*** 0.059*** 0.048*** 0.119*** -0.013 -0.084*** 0.128*** 0.143*** -0.017 0.040** 0.058***
Capital Expenditure -0.019 0.013 0.024 0.015 -0.084*** -0.019 -0.040** -0.054*** -0.007 0.192*** 0.003 -0.026 0.001 0.004 -0.007
Cash/Dividends -0.008 -0.002 0.005 -0.002 -0.015 -0.004 -0.030 -0.023 -0.025 -0.015 -0.007 -0.004 0.000 0.009 0.006
3yr Dividends Growth 0.115*** 0.035* 0.001 -0.013 -0.006 0.055*** 0.023 0.006 -0.017 -0.036** 0.047*** 0.022 0.026 0.124*** 0.007
5yr Income Growth -0.062*** 0.150*** -0.002 0.041** -0.026 0.055*** -0.078*** -0.053*** -0.017 -0.075*** 0.082*** 0.003 0.029 0.129*** 0.032*
5yr Sales Growth -0.073*** 0.198*** 0.009 0.073*** -0.055*** 0.031 -0.004 0.007 0.059*** -0.046** 0.076*** -0.022 0.017 0.082** 0.035**
Short-Term Debt 0.058*** -0.002 0.003 -0.066*** 0.349*** -0.037** 0.034** 0.057*** -0.037** -0.115*** 0.083*** -0.108*** 0.010 -0.038** 0.003
5yr Assets Growth -0.065*** 0.144*** 0.006 0.027 -0.057*** 0.043** 0.030 0.020 0.081*** -0.106*** 0.074*** 0.056*** 0.020 0.088*** 0.071***
Total Debt -0.004 -0.030** -0.003 -0.054*** 0.152*** -0.028 -0.028 -0.038** -0.051*** -0.046*** 0.025 -0.071*** 0.011 -0.059*** -0.013
2yr Sales Growth PE Ratio
Price-to-book
ratio Quick Ratio
Return on
Assets
R&D
Intensity
Capital
Expenditure
Cash/
Dividends
3yr Dividends
Growth
5yr Income
Growth
5yr Sales
Growth
Short Term
Debt
5yr Assets
Growth
2yr Sales Growth 1.000
PE Ratio -0.001 1.000
Price-to-book ratio 0.000 0.002 1.000
Quick Ratio -0.011 -0.009 0.003 1.000
Return on Assets 0.018 -0.010 0.391*** 0.157*** 1.000
R&D Intensity -0.006 -0.003 0.034** 0.146*** 0.011 1.000
Capital Expenditure -0.003 0.000 -0.008 -0.036** -0.035** -0.015 1.000
Cash/Dividends -0.001 0.029 0.012 -0.003 0.012 0.001 -0.010 1.000
3yr Dividends Growth 0.006 -0.006 0.146*** 0.022 0.306*** -0.023 0.067*** 0.039* 1.000
5yr Income Growth 0.005 -0.015 0.152*** 0.003 0.226*** 0.005 0.181*** 0.025 0.236*** 1.000
5yr Sales Growth 0.004 -0.002 0.092*** -0.024 0.113*** 0.016 0.21*** 0.020 0.045** 0.673*** 1.000
Short-Term Debt -0.004 -0.002 -0.0312** -0.096*** -0.093*** -0.028* -0.069*** -0.005 0.015 -0.011 -0.026 1.000
5yr Assets Growth 0.005 -0.007 0.103*** 0.060*** 0.077*** 0.069*** 0.301*** 0.031 0.028 0.543*** 0.716*** -0.027 1.000
Total Debt -0.005 0.003 0.161*** -0.071*** -0.128*** -0.055*** 0.044*** 0.004 -0.055*** -0.010 0.004 0.129*** 0.038** 1.000
The following table displays the correlations among the variables in the Credi Lyonnais Securities Asia dataset. This dataset encompasses 10 years of CLSA tracking corporate governance performance of firms in emerging economies.
Correlation are marked with an * for 5% significance, ** for 1% significance, and *** for 0.1% significance.
Panel B: CLSA Variable Correlations
Total Debt
Variable Mean Median St. Dev. Min Max Observations
Index Corporate Governance Quotient 50.2 50.2 28.8 0 100 15,390
Industry Corporate Governance Quotient 50.7 50.7 28.8 0 100 15,390
2yr Sales Growth 9.8 47.7 4,035.90 -1,114.70 481,801.50 14,261
Financial Dependence -1.9 -4 25.4 -808.2 230 10,013
Closely Held Shares 29 32.7 23.5 0 158.7 13,602
Log (Assets) 8.1 8.2 1.9 1.3 15.1 14,399
Cash/Total Assets 0.1 0.1 0.1 0 1.8 12,992
Antidirector x Legal 6.1 6.2 1.8 -2.3 8.7 15,262
GDP per Capita 28,367.80 29,760.10 7,717.60 2,032.60 56,624.70 15,134
Stock Market Cap/GDP 103.2 114.8 87.1 13.2 617 15,093
Fixed Assets 0.3 0.3 0.2 0 1 14,280
SEC Compliance 0 0.1 0.3 0 1 15,267
Current Ratio 1.4 1.8 1.9 0 67.7 12,088
Leverage 0.6 0.6 2.9 0 195 13,695
Tobin's Q 0.8 1.5 15.6 0 1,129.40 13,336
Foreign Sales 38.9 42 35.4 -1,225.10 626.4 11,451
1yr Foreign Sales Growth 5.4 171.1 5,205.60 -100 378,490.00 11,004
SEC Compliance 0 0.1 0.3 0 1 15,267
PE Ratio 15.3 28.5 1,044.20 -1,833.00 123,300.00 14,037
Price-to-Book Ratio 1.7 2.1 21 -2,105.40 413.1 14,079
Quick Ratio 0.9 1.3 1.8 0 67.3 12,095
Return on Assets 4.4 4.9 29.7 -594.5 3,292.10 14,296
R&D Intensity 0 4,090.10 62,280.30 -536.4 3,734,452.80 15,267
Capital Expenditure 3.8 5.4 6.7 0 184.9 13,718
Cash/Dividends 4.9 27.5 905.8 -26,302.00 46,526.00 10,954
3yr Dividends Growth 5.8 4.8 32.6 -100 299.6 13,204
5yr Income Growth 8.7 11.8 24.6 -100 330.5 11,499
5yr Sales Growth 6.2 9.1 21.1 -100 573.3 13,804
Short-Term Debt 126.2 4,615.50 31,905.50 0 917,515.30 14,261
5yr Assets Growth 6 9.3 20.5 -61.1 1,196.60 13,772
Total Debt 53 139.1 1,515.20 -76,200.00 99,728.50 14,362
Panel C: GRI Variable Descriptions
This table displays the summary statistics for the variables in the Global Reporting Initiative (GRI) dataset. GRI tracked corporate governance behavior
of rms around the world from 2003-2009, but mostly in developed economies. It is for this reason that the mean GDP per capita differs so
dramatically from the mean GDP per capita in Table 1 where emerging economies create a lower overall average statistic. The first two variables
reported below are the two corporate governance scores awareded to firms. The next three variables are the three observable country characteristics
used in our analysis. Antidirector x Legal captures the interactions of the Revised Antidirector Rights Index and the Rule of Law in that coutnry. The
following firm variables include the observable firm characteristics included in previous studies (sales growth, financial dependence (EBITDA based),
closely held shared (as a percent of total shares), log(assets), and cash-to-assets ratio. The remaining variables described below are additional
observable firm characteristics used to capture the complex interaction between firms and corporate governance in emerging economies.
Index Corp Gov
Quotient
Industry Corp
Gov Quotient
2yr Sales
Growth
Financial
Dependence
Closely Held
Shares
Log
(Assets)
Cash/ Total
Assets
Antidirector
x Legal
GDP per
capita
Stock Market
Cap/ GDP Fixed Assets
SEC
Compliance
Current
Ratio Leverage
Tobin's
Q
Foreign
Sales
Index Corporate Governance Quotient 1.000
Industry Corporate Governance Quotient 0.933*** 1.000
2yr Sales Growth -0.004 -0.003 1.000
Financial Dependence 0.004 0.006 0.001 1.000
Closely Held Shares -0.283*** -0.312*** 0.014 0.0216** 1.000
Log (Assets) -0.048*** -0.030*** 0.006 -0.066*** -0.115*** 1.000
Cash/Total Assets -0.011 -0.015* -0.006 -0.059*** 0.060*** -0.273*** 1.000
Antidirector x Legal 0.387*** 0.396*** -0.003 0.000 0.183*** -0.204*** 0.030*** 1.000
GDP per capita -0.237*** -0.254*** 0.006 0.013 -0.071*** 0.025*** 0.096*** 0.252*** 1.000
Stock Market Cap/GDP 0.100*** 0.099*** 0.014 0.000 0.123*** -0.063*** 0.064*** 0.357*** 0.107*** 1.000
Fixed Assets -0.011 -0.013 -0.006 0.100*** 0.025*** -0.066*** -0.037*** 0.042*** -0.029*** 0.014* 1.000
SEC Compliance 0.124*** 0.184*** -0.003 0.017* -0.176*** 0.212*** 0.0179** 0.038*** -0.111*** 0.0178** 0.066*** 1.000
Current Ratio -0.031*** -0.007 -0.001 -0.009 0.022** -0.215*** 0.446*** 0.046*** 0.053*** 0.091*** -0.169*** 0.037*** 1.000
Leverage 0.037*** 0.034*** -0.002 -0.007 -0.014 -0.010 -0.0192** 0.0211** -0.013 -0.010 -0.026*** -0.011 -0.033*** 1.000
Tobin's Q 0.029*** 0.030*** -0.003 0.001 -0.005 -0.055*** 0.017* 0.027*** -0.003 0.008 -0.016* -0.005 0.005 0.897*** 1.000
Foreign Sales 0.145*** 0.161*** -0.001 0.013 -0.057*** -0.006 0.055*** 0.064*** -0.108*** 0.077*** -0.087*** 0.188*** 0.070*** 0.010 0.019* 1.000
1yr Foreign Sales Growth -0.002 -0.003 0 -0.005 -0.001 -0.008 0.000 0.003 -0.014 0.011 -0.010 -0.006 -0.002 -0.002 -0.001 0.021**
SEC Compliance 0.141*** 0.198*** -0.003 0.018* -0.185*** 0.229*** 0.016* 0.022*** -0.087*** 0.019** 0.063*** 0.922*** 0.030*** -0.009 -0.004 0.201***
PE Ratio -0.012 -0.014* 0 0.003 0.001 0.001 -0.005 -0.002 0.014 -0.001 0.005 -0.003 -0.003 -0.001 -0.001 -0.008
Price-to-Book Ratio -0.009 -0.012 0 0.001 0.014 -0.011 0.032*** -0.005 -0.009 0.007 -0.009 0.015* -0.003 0.003 0.009 0.001
Quick Ratio -0.028*** -0.011 -0.001 -0.005 0.033*** -0.204*** 0.476*** 0.033*** 0.036*** 0.092*** -0.144*** 0.043*** 0.972*** -0.033*** 0.002 0.058***
Return on Assets 0.015* 0.008 -0.001 0.005 0.043*** 0.007 -0.001 0.007 -0.014 0.053*** 0.006 -0.005 0.013 -0.012 0.003 0.014
R&D Intensity 0.032*** 0.032*** 0 0.003 -0.022** -0.090*** 0.157*** 0.031*** -0.006 -0.007 -0.045*** 0.014* 0.115*** 0.347*** 0.399*** -0.002
Capital Expenditure 0.027*** 0.028*** -0.003 0.062*** 0.013 -0.069*** -0.126*** 0.026*** -0.073*** -0.002 0.462*** 0.077*** -0.055*** -0.019** -0.003 0.031***
Cash/Dividends -0.016 -0.013 0 0.007 -0.006 0.011 -0.001 -0.018* 0.016* -0.010 0.006 -0.006 -0.009 0.002 -0.004 -0.008
3yr Dividends Growth -0.064*** -0.060*** 0.002 -0.002 -0.013 0.071*** 0.056*** 0.044*** 0.109*** 0.085*** -0.026*** 0.022** 0.032*** -0.033*** -0.024***-0.026***
5yr Income Growth 0.026*** 0.022** 0.242*** -0.0242** -0.012 -0.049*** 0.030*** 0.005 -0.046*** 0.032*** -0.043*** 0.020** -0.014 -0.007 -0.004 0.054***
5yr Sales Growth 0.039*** 0.046*** 0.123*** -0.005 0.025*** -0.013 0.040*** 0.054*** -0.081*** 0.107*** 0.030*** 0.034*** 0.011 -0.018** -0.013 0.027***
Short-Term Debt 0.055*** 0.070*** -0.001 -0.090*** -0.078*** 0.378*** -0.056*** -0.041*** -0.040*** -0.023*** -0.142*** 0.137*** -0.100*** 0.012 -0.010 -0.058***
5yr Assets Growth 0.079*** 0.089*** 0.012 -0.016 -0.015* 0.024*** 0.014 0.067*** -0.131*** 0.051*** 0.029*** 0.050*** 0.065*** -0.019** -0.015* 0.048***
Total Debt -0.011 -0.011 -0.001 -0.011 -0.001 0.077*** -0.022*** -0.023*** -0.007 -0.015* -0.020** 0.009 -0.025*** 0.006 -0.003 -0.031***
1yr Foreign
Sales Growth
SEC
Compliance PE Ratio
Price-to-Book
Ratio Quick Ratio
Return on
Assets
R&D
Intensity
Capital
Expenditure
Cash/
Dividends
3yr Dividends
Growth
5yr Income
Growth
5yr Sales
Growth
Short-Term
Debt
5yr Assets
Growth
1yr Foreign Sales Growth 1.000
SEC Compliance -0.007 1.000
PE Ratio 0.005 -0.004 1.000
Price-to-Book Ratio 0.000 -0.006 0.000 1.000
Quick Ratio -0.001 0.036*** -0.005 0.004 1.000
Return on Assets -0.010 -0.003 -0.001 0.007 0.012 1.000
R&D Intensity 0.003 0.012 -0.001 0.004 0.123*** -0.046*** 1.000
Capital Expenditure 0.004 0.080*** -0.003 0.012 -0.036*** 0.038*** -0.025*** 1.000
Cash/Dividends -0.001 -0.007 0.000 0.004 -0.008 -0.002 0.003 0.001 1.000
3yr Dividends Growth 0.006 0.027*** 0.000 0.018** 0.037*** 0.087*** -0.015* 0.060*** 0.001 1.000
5yr Income Growth 0.007 0.028*** -0.017* 0.027*** 0.004 0.193*** -0.017* 0.133*** 0.003 0.294*** 1.000
5yr Sales Growth 0.008 0.038*** -0.003 0.018** 0.018** 0.154*** 0.009 0.174*** -0.003 0.107*** 0.535*** 1.000
Short-Term Debt -0.003 0.128*** -0.002 -0.003 -0.082*** -0.016* -0.009 -0.083*** -0.001 -0.023*** -0.002 0.011 1.000
5yr Assets Growth -0.006 0.045*** -0.005 0.014* 0.077*** 0.029*** -0.014 0.221*** -0.012 0.091*** 0.440*** 0.649*** 0.032*** 1.000
Total Debt 0.000 -0.006 0.000 0.533*** -0.021** -0.009 -0.005 -0.019** 0.015 -0.022** 0.000 -0.002 0.086*** -0.003 1.000
Panel D: GRI Variable Correlations
The following table displays the correlations among the variables in the Global Reporting Initiative (GRI) dataset. Variables below were used in the analysis of GRI data on corporate governance behavior of firms around the world from 2003-2009. As can be seen
below, the two corporate governance scores issued by GRI, the Index Corporate Governance Quotient and the Industry Corporate Governance Quotient are highly correlated but not identical. The two scores allow us to run two sets of analyses on the GRI data.
Correlations are marked with an * for 5% significance, ** for 1% significance, and *** for 0.1% significance.
Total Debt
Independent Variables *(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth -0.001 0.002 0.002
(0.003) (0.002) (0.012)
Financial Dependence -0.031 0.191 -0.083
(0.151) (0.193) (0.227)
Closely Held Shares -0.055* -0.062 -0.021
(0.026) (0.034) (0.05)
log(Assets) -0.477 -0.792 -1.210
(0.379) (0.573) (3.195)
Cash/Total Assets 2.075 -9.514 -3.641
(3.985) (5.896) (9.785)
Antidirector x Legal 2.107** 1.736** 1.355* 0.08 7.060**
(0.372) (0.458) (0.573) -0.695 (2.430)
GDP per capita -0.000** -0.000 -0.000 0.000 -0.000
(0.000) (0.000) (0.000) (0.000) (0.001)
Stock Market Cap/GDP 0.002 -0.008 0.001 0.009* -0.01
(0.003) (0.005) (0.005) (0.004) (0.013)
Expanded Firm Variables yes yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Firm FE yes yes
Observations 3,635 2,744 1,445 782 3,635 2,744 3,635 782 3763 3635
R² 0.080 0.133 0.125 0.223 0.212 0.230 0.730 0.796
Adjusted-R² 0.0782 0.130 0.116 0.190 0.206 0.222 0.606 0.631
Additional Adjusted-R² 0.052 -0.014 0.06 0.128 0.144 0.4 0.41
Country Random Effect 8.531 7.89
(1.668) (1.562)
Firm Random Effect 9.38 9.364
(0.276) (0.271)
Residual 9.921 9.225
(0.137) (0.129)
Firm 53.21 41.41
Year 7.81% 7.59%
Country 11.97 11.48
Table II
The tables below show the coefficient estimates from the CLSA corporate governance ratings for emerging economies only. In Panel A there are the OLS and xtmixed models; in Panel B
there are the nested Anova results. The regressions below explore the relative importance of countries and firms in explaining corporate governance ratings of firms in emerging
economies. The OLS models include different combinations of observable firm and country characteristics as well as observable firm and country fixed effects. We run three sets of firm
variables - the first is the original set of observable firm characteristics used in previous analyses. The second is more inclusive and adds 17 additional observable firm characteristics to
capture any firm effect not previously accounted for. The third are fixed effects to capture unobservable firm characteristics. The Anova models similarly capture unobservable firm
characteristics. Because firms are necessarily nested within countries, which are nested within years, we analyze years, then add country effects, and then add firm effects. Xtmixed
model specifications include random effects, which allows for the hierarchical nature of the country and firm data, enabling us to do a single regression and analyze the importance of
firms and countries. The results in Panel A below show that firm characteristics generally explain more corporate governance ratings variation than country characteristics. Looking at
observed characteristics, countries explain 5.2% of the ratings variance while an inclusive set of firm characteristics explains slightly more variance at 6%. Including fixed effects and
observable characteristics, countries now explain 14.4% of the variance and firms explain 41%. In the random effects, xtmixed, models the amount of ratings variance explained can be
found by first squaring both firm and country standard deviations to get variances, adding them, and then dividing firm and country variances by the total. In model 10, this shows that
countries explain 26.4% of variance and firms explain 37.3%. Correlations are marked with an * for 5% significance, ** for 1% significance, and *** for 0.1% significance. The results in
Panel B confirm those found in Panel A. Firms explain more ratings variance than countries.
Panel A: OLS and xtmixed Results in CLSA Emerging Economies Only
Panel B: Nested Anova Results for CLSA Emerging Economies Only
Source of Variation Additional R ² Additional Adjusted-R ²
Independent variables *(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth 0.003 0.007 -0.00730
(0.007) (0.03) (0.0266)
Financial Dependence 0.077 -0.592 2.022
(0.193) (0.809) (1.528)
Closely Held Shares -0.026 -0.002 0.114
(0.07) (0.08) (0.0970)
Log (Assets) 0.219 0.269 5.094
(0.992) (1.649) (6.152)
Cash/Assets 2.400 -7.326 -24.81
(9.648) (13.81) (21.75)
Antidirector x Legal -0.428 -0.141 3.385+ -3.799** 7.822
(0.650) (1.149) (1.910) (1.055) (6.416)
GDP per capita 0.001** 0.005 -0.000 0.001** -0.00199+
(0.000) (0.000) (0.001) (0.000) (0.00106)
Stock Market Cap/GDP -0.038** -0.035** -0.041** 0.031* 0.0374*
(0.006) (0.009) (0.011) (0.012) (0.0170)
Expanded Firm Variables yes yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Company FE yes yes
Observations 1,413 1,292 748 351 1,413 1,292 1,413 351 1,413 1,413
R² 0.170 0.266 0.241 0.313 0.337 0.372 0.748 0.843
Adjusted-R² 0.167 0.262 0.227 0.249 0.329 0.361 0.675 0.740
Additional Adjusted-R² 0.095 -0.035 -0.013 0.162 0.194 0.346 0.379
Country Random Effect 10.992 11.605
(4.115) (4.103)
Firm Random Effect 13.514 13.365
(0.765) (0.685)
Residual 15.566 12.635
(0.331) (0.268)
Table III
Panel A: OLS and xtmixed Results for GRI Emerging Economies Only - Index Weighted Score
These tables show the regression results on the corporate governance scores using GRI emerging economies data only. Panel A uses both OLS and random effects regression to look at the index-
weighted corporate governance quotient, Panel B again uses OLS and xtmixed but to look at the industry-weighted corporate governance quotient, Panel C looks at the index-weighted corporate
governance quotient using Anova, and Panel D looks at the industry-weighted corporate governance quotient using Anova. All four of these panels are intended to explore the relative
importance of countries and firm in explaining governance ratings in emerging economies. As well, similarly to the CLSA data, firms are nested within countries which are nested within years, so
we proceed successively, evaluating the additional contribution to ratings variance explained by each category. The results below show that firm characteristics explain anywhere from roughly
statistically equal to significantly more of the ratings variance than country characteristics do in emering economies. In models 4 in Panel A and B, the full set of observable firm characteristics
explains 2-3 percentage points more than country characteristics in model 2. Looking at unobservable characteristics with fixed effects, in addition to observable characteristics using specific
variables, we see that firms again explain more variance (37.9% in Panel A, 44.7% in Panel B) than countries explain (14.5% in Panel A, 11.8% in Panel B). The random effects results in Model 10
suggest that firm effects are anywhere from somewhat to significantly larger than country effects. In Panel A, firm random effects account for 37.8% of the variance while country random
effects account for 28.5%. In Panel B, firms account for 50.3% while countries account for 11% of variance. Correlations are marked with an * for 5% significance, ** for 1% significance, and ***
for 0.1% significance. Anova results in Panels C and D exhibit the same pattern of firms explaining substantially more variance than countries.
*(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth 0.005 0.017 -0.004
(0.007) (0.028) (0.027)
Financial Dependence 0.095 -0.200 1.511
(0.171) (0.791) (1.883)
Closely Held Shares -0.067 -0.020 0.054
(0.074) (0.083) (0.097)
Log (Assets) 0.555 1.086 6.466
(1.054) (1.712) (6.079)
Cash/Total Assets 9.190 2.724 -19.61
(10.06) (15.75) (24.44)
Antidirector x Legal -0.003 0.215 2.726 -4.111** 9.664
(0.716) (1.197) (1.931) (1.177) (6.717)
GDP per capita 0.001** 0.000 0.000 0.001** -0.002*
(0.000) (0.000) (0.001) (0.000) (0.001)
Stock Market Cap/GDP -0.035** -0.034** -0.044** 0.023+ 0.033+
(0.006) (0.009) (0.011) (0.014) (0.019)
Expanded Firm Variables yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Company FE yes yes
Observations 1,413 1,292 748 351 1,413 1,292 1,413 351 1,413 1,413
R² 0.143 0.218 0.208 0.302 0.265 0.306 0.737 0.840
Adjusted-R² 0.140 0.213 0.194 0.237 0.255 0.294 0.662 0.735
Additional Adjusted-R² 0.073 -0.019 0.024 0.115 0.154 0.407 0.438
Country Random Effect 6.774 6.988
(2.593) (2.74)
Firm Random Effect 15.084 14.944
(0.803) (0.743)
Residual 15.536 13.105
(0.331) (0.279)
Additional adjusted R²
14.13%
11.54
40.53
Source of Variation
Year
Country
Firm
Additional Ordinary R²
14.49%
12.18
47.09
Year
Panel D: Nested Anova Results for GRI Emerging Economies Only - Industry Weighted Score
Source of Variation
Country
Firm
Additional R ²
17.06%
16.72
40.99
Additional Adjusted-R ²
16.71%
16.17
43.62
Panel B: OLS and xtmixed Results for GRI Emerging Economies Only - Industry Weighted Score
Panel C: Nested Anova Results for GRI Emerging Economies Only - Index Weighted Score
Independent variables *(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth -0.000** -0.016 -0.015
(0.000) (0.016) (0.019)
Financial Dependence 0.008 0.032* 0.014
(0.023) (0.016) (0.017)
Closely Held Shares -0.212** -0.210** -0.024
(0.022) (0.03) (0.051)
Log (Assets) 0.639* -0.230 2.732
(0.303) (0.459) (2.016)
Cash/Total Assets 11.82** 9.987+ -9.262
(2.827) (5.195) (8.046)
Antidirector x Legal 7.208** 6.335** 6.043** -1.445+ -12.41**
(0.285) (0.444) (0.548) (0.777) (2.413)
GDP per capita -0.002** -0.002** -0.002** 0.000 -0.004**
(0.000) (0.000) (0.000) (0.000) (0.001)
Stock Market Cap/GDP 0.175** 0.165** 0.174** -0.012 0.029
(0.01) (0.015) (0.017) (0.015) (0.028)
Expanded Firm Variables yes yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Company FE yes yes
Observations 13,977 13,779 7,473 4,468 13,977 13,779 13,977 4,468 13,977 13,977
R² 0.001 0.382 0.416 0.474 0.559 0.562 0.768 0.794
Adjusted-R² 0.001 0.381 0.414 0.470 0.558 0.561 0.712 0.728
Additional Adjusted-R² 0.38 0.033 0.089 0.557 0.56 0.155 0.171
Country Random Effect 18.081 18.114
(2.769) (2.774)
Firm Random Effect 11.680 11.672
(0.231) (0.231)
Residual 15.868 15.804
(0.106) (0.105)
Table IV
Panel A: OLS and xtmixed Results for GRI Developed Economies Only - Index Weighted Score
These tables show the results regressions on the corporate governance scores using GRI developed economies data only. Panel A uses both OLS and random effects regression to look at the
index-weighted corporate governance quotient, Panel B again uses OLS and xtmixed but to look at the industry-weighted corporate governance quotient, Panel C looks at the index-weighted
corporate governance quotient using Anova, and Panel D looks at the industry-weighted corporate governance quotient using Anova. All four of these panels are intended to explore the
relative importance of countries and firm in explaining governance ratings in developed economies. As well, similar to Tables 5 and 6, firms are nested within countries which are nested within
years, so we proceed successively, evaluating the additional contribution to ratings variance explained by each category. The results below show that country characteristics explain
substantially more governance ratings variance than firm characteristics in developed economies. In models 4 in Panel A and B countries (36.9% and 40.5%, respectively) explain roughly three
times the variance explained by firms (9.8% and 8.3%). Looking at unobservable characteristics with fixed effects in addition to observable characteristics using specific variables in Models 6
and 8, we see that countries again explain more variance (55.8% in Panel A, 57% in Panel B) than firms explain (17.1% in Panel A, 15.8% in Panel B). The random effects results in Model 10 echo
these results. Countries explain 46% and 48.1% in Panel A and B, respectively, while firms explain 19.1% and 18.7%. Correlations are marked with an * for 5% significance, ** for 1% significance,
and *** for 0.1% significance. Anova results in Panels C and D exhibit the same pattern of countries explaining more variance than firms.
*(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth -0.000** -0.017 -0.021
(0.000) (0.016) (0.018)
Financial Dependence 0.02 0.047** 0.008
(0.021) (0.014) (0.019)
Closely Held Shares -0.235** -0.205** -0.021
(0.022) (0.027) (0.053)
Log (Assets) 0.975** 0.169 4.323*
(0.295) (0.445) (1.973)
Cash/Total Assets 12.39** 8.228 -11.14
(2.779) (5.060) (7.925)
Antidirector x Legal 7.346** 6.526** 6.308** -1.258+ -11.75**
(0.284) (0.438) (0.529) (0.745) (2.328)
GDP per capita -0.002** -0.002** -0.002** -0.000 -0.003**
(0.000) (0.000) (0.000) (0.000) (0.001)
Stock Market Cap/GDP 0.190** 0.177** 0.183** -0.018 -0.005
(0.011) (0.015) (0.017) (0.015) (0.026)
Expanded Firm Variables yes yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Company FE yes yes
Observations 13,977 13,779 7,473 4,468 13,977 13,779 13,977 4,468 13,977 13,977
R ² 0.001 0.418 0.455 0.492 0.571 0.574 0.777 0.793
Adjusted-R² 0.001 0.418 0.454 0.489 0.57 0.573 0.723 0.727
Additional Adjusted-R² 0.417 0.036 0.071 0.569 0.572 0.153 0.154
Country Random Effect 18.569 18.593
(2.84) (2.843)
Firm Random Effect 11.604 11.576
(0.226) (0.226)
Residual 15.488 15.448
(0.103) (0.103)
Additional Adjusted-R ²
0.06%
57.08
15.25
Source of Variation
Year
Country
Firm
Additional R ²
0.10%
57.13
20.52
Additional Adjusted-R ²
0.07%
55.85
15.46
Panel D: Nested Anova Results for GRI Developed Economies Only - Industry Weighted Score
Source of Variation
Year
Country
Firm
Additional R ²
0.11%
55.9
20.92
Panel B: OLS and xtmixed Results for GRI Developed Economies Only - Industry Weighted Score
Panel C: Nested Anova Results for GRI Developed Economies Only - Index Weighted Score
Independent variables *(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth 0.002** -0.004
(0.001) -0.011
Financial Dependence 0.017 0.211
(0.014) -0.255
Closely Held Shares -0.001 0.012
(0.003) -0.022
Log (Assets) -0.03 0.722*
(0.05) -0.365
Cash/Total Assets 0.976* 8.568
(0.541) -6.041
Antidirector x Legal 0.363*** 0.464** 5.390 0.211
(0.138) (0.199) (3.390) -0.243
GDP per capita -0.000** -0.000** 0.003 0.000
(-0.000) (0.000) (0.003) -0.000
Stock Market Cap/GDP -0.001 -0.000 -0.016 -0.001
(0.001) (0.001) (0.021) -0.001
Expanded Firm Variables yes yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Company FE yes yes
Observations 607 607 354 38 607 607 607 38 607 607
R² 0.044 0.216 0.213 0.871 0.208 0.217 0.663 1
Adjusted-R² 0.039 0.209 0.187 0.471 0.202 0.207 0.527
Additional Adjusted-R² 0.17 -0.022 0.262 0.163 0.168 0.325
Country Random Effect 0.355 0.353
(0.167) (0.166)
Firm Random Effect 0.506 0.521
(0.042) (0.041)
Residual 0.667 0.628
(0.023) (0.021)
Firm 44.99 32.2
Year 4.35% 3.88%
Country 16.47 16.29
Table V
The tables below show the coefficient estimates from FTSE corporate governance ratings for emerging economies only. In Panel A there are the OLS and xtmixed models; in Panel B there are the nested
Anova results. These emerging economies are defined as those countries not a member of the OECD by 1990. The regressions below explore the relative importance of countries and firm in explaining
corporate governance ratings of firms in emerging economies. We run three sets of firm variables - the first is the original set of observable firm characteristics used in previous analyses. The second is
more inclusive and adds 17 additional observable firm characteristics to capture any firm effect not previously accounted for. The third are fixed effects to capture unobservable firm characteristics.
The Anova models similarly capture unobservable firm characteristics. Because firms are necessarily nested within countries, which are nested within years, we analyze years, then add country effects,
and then add firm effects. Xtmixed models specifications include random effects, which allows for the hierarchical nature of the country and firm data, enabling us to do a single regression and analyze
the importance of firms and countries. The results in Panel A below show that for emerging economies in the FTSE data, firms explain more of the ratings variance than countries do. However, these
results are limited given the small number of emerging economies in the FTSE dataset. We could not make strong conclusions based on Models 4 and 8 due to the small sample size. There are still
some conclusions we can gather, still, from the other models with larger sample size. Specifically, we can see that when we just look at unobservable firm characteristics captured in the firm fixed
effects in Model 7, firms explain 32.5% of the ratings variance. Unobservable country characteristics explain 16.3%. Our random effects model also gives us reliable results. In Model 10, firm random
effects account for 34.3% of ratings variance and country random effects account for 15.8%. Correlations are marked with an * for 5% significance, ** for 1% significance, and *** for 0.1% significance.
The results in Panel B confirm those found in Panel A. Firms explain more ratings variance than countries.
Panel A: OLS and xtmixed Results for FTSE Emerging Markets Only
Panel B: Nested Anova Results for FTSE Emerging Economies Only
Source of Variation Additional R ² Additional Adjusted-R ²
Independent variables *(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth -0.000*** 0.001 0.002
(0.000) (0.001) (0.002)
Financial Dependence -0.000*** 0.001 -0.001***
(0.000) (0.001) (0.001)
Closely Held Shares -0.006*** -0.005*** 0.001
(0.001) (0.001) (0.003)
Log (Assets) 0.036*** 0.052*** -0.109
(0.011) (0.019) (0.262)
Cash/Total Assets 0.000 0.085 -0.272
(0.004) (0.17) (0.322)
Antidirector x Legal 0.357*** 0.378*** 0.306*** 0.239*** 0.414**
(0.01) (0.017) (0.027) (0.054) (0.162)
GDP per capita -0.000*** -0.000*** -0.000*** 0.000 0.000
(0.000) (0.000) (0.000) (0.000) (0.000)
Stock Market Cap/GDP 0.008*** 0.007*** 0.005*** 0.005*** 0.007**
(0.001) (0.008) (0.001) (0.001) (0.003)
Expanded Firm Variables yes yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Company FE yes yes
Observations 9,129 9,112 4,971 1,701 9,121 9,112 9,112 1,701 9,121 9,121
R² 0.001 0.595 0.605 0.576 0.656 0.659 0.881 0.86
Adjusted-R² 0.001 0.595 0.604 0.569 0.655 0.658 0.834 0.786
Additional Adjusted-R² 0.594 0.009 -0.026 0.654 0.657 0.179 0.131
Country Random Effect 0.762 0.763
(0.114) (0.115)
Company Random Effect 0.519 0.519
(0.01) (0.01)
Residual 0.504 0.503
(0.004) (0.004)
Additional R ²
0.14%
22.27
Table VI
Panel A: OLS and xtmixed Results for FTSE Developed Economies Only
The tables below show the coefficient estimates from FTSE corporate governance ratings for developed economies only. In Panel A there are the OLS and xtmixed models; in Panel B there are
the nested Anova results. These developed economies are defined as those countries that were a member of the OECD by 1990. The regressions below explore the relative importance of
countries and firm in explaining corporate governance ratings of firms in developed economies. We run three sets of firm variables - the first is the original set of observable firm
characteristics used in previous analyses. The second is more inclusive and adds 17 additional observable firm characteristics to capture any firm effect not previously accounted for. The third
are fixed effects to capture unobservable firm characteristics. The Anova models similarly capture unobservable firm characteristics. Because firms are necessarily nested within countries,
which are nested within years, we analyze years, then add country effects, and then add firm effects. Xtmixed models specifications include random effects, which allows for the hierarchical
nature of the country and firm data, enabling us to do a single regression and analyze the importance of firms and countries. The results in Panel A below show that for developed economies
in the FTSE data, countries explain more of the ratings variance than firms do. In contrast to the FTSE emerging market data, these results are not limited; their sample size was much larger
and all models yield reliable results. Looking just at observable characteristics in Models 2 and 4, countries explain 59.4% of the ratings variance while firms explain -0.026%, or nothing. When
we add in unobservable firm and country characteristics with fixed effects, countries explain even more. In Model 6, countries explain 65.7% of the variance while firms explain 13.1%. And
finally, looking at random effects results in Model 10, we see that country random effects account for 52.7% of ratings variance while firm random effects account for only 24.4% of ratings
variance. Correlations are marked with an * for 5% significance, ** for 1% significance, and *** for 0.1% significance. The results in Panel B confirm those found in Panel A. Countries explain
more ratings variance than firms in developed economies.
Panel B: Nested Anova Results for FTSE Developed Economies Only
Source of Variation Additional Adjusted-R ²
0.10%
65.47
17.59
65.54
Year
Country
Company
Country Observations Mean St. Dev. Median Min Max 25th Per 75th Per
Argentina 3 59.65 9.97 59.6 52.6 66.7 52.6 66.7
Brazil 68 60.163 11.781 61.3 34.6 89.3 53.5 68
Chile 17 59.938 7.556 59.95 44.4 72.2 58.05 63.8
China 418 47.294 15.096 48.905 0 84.595 38.596 58.054
Colombia 3 51.4 (one observation)
Czech Republic 3 42.75 5.162 47.75 44.1 51.4 44.1 51.4
Hong Kong 740 59.77 12.812 60.482 3.75 93.5 52.868 67.77
Hungary 5 51.925 6.878 51 45.3 60.4 46.4 57.45
India 641 52.44 12.646 51.44 3.333 93.75 43.41 61.1
Indonesia 184 41.654 16.681 39.862 4 79.345 32.4 51.155
Malaysia 314 58.146 13.341 58.854 12.047 91.042 50.638 66.209
Mexico 20 63.94 9.336 66.7 39 74.2 62.1 69.9
Pakistan 15 33.973 13.501 30.7 18.9 65.6 25.3 43
Peru 3 73.08 2.965 71.5 71.24 76.5 71.24 76.5
Philippines 117 50.217 17.381 53.7 7.709 83 36.6 63.5
Poland 5 40.525 6.926 38.9 34 50.3 36.2 44.85
Russia 2 22.05 9.405 22.05 15.4 28.7 15.4 28.7
Singapore 320 59.525 10.752 59.616 32.4 88.042 51.387 83.304
South Africa 59 70.607 8.93 70.7 45 82.6 64.9 78.79
South Korea 386 53.555 15.164 54.649 0 81.012 45.276 64.399
Taiwan 500 53.606 12.913 54.558 0 97.64 47.463 61.565
Thailand 227 61.524 12.659 63.5 21.648 88.042 54.8 69.709
Country Mean St. Dev. Median Min Max 25th Per 75th Per
Australia 74 62.562 21.135 70.539 0 86.209 53.065 78
Canada 7 56.607 15.654 61.658 30.9 71 45.5 68.929
Greece 3 57.15 5.162 57.15 53.5 60.8 53.5 60.8
Japan 209 57.817 16.432 55.817 3.75 86.334 50.459 69.917
New Zealand 2 89.386 3.536 89.386 86.886 91.887 86.886 91.997
Norway 1 80.2
Spain 2 45.55 1.626 45.55 44.4 46.7 44.4 46.7
Switzerland 2 82.6 0 82.6 82.6 82.6 82.6 82.6
Turkey 31 41.214 13.609 39.4 10.5 60.4 34.7 50.6
United Kingdom 32 73.511 11.988 77 46.9 93.5 66.179 81.683
United States 29 55.411 12.743 54.065 22.766 88.959 48.107 62.6
Country Mean St. Dev. Median Min Max 25th Per 75th Per
Bermuda 113 63.578 25.084 65.5 2.7 99.7 49.3 81.9
Cayman Islands 47 59.489 18.188 59.2 14.6 99.1 47.3 74.7
Gibralter 4 69.625 4.65 65.5 73.8 69.6 65.6 73.65
Guernsey 5 76.74 4.841 77.4 72.3 84.1 72.4 77.5
Hong Kong 660 39.646 19.194 43.15 1.7 95 24.6 53.15
Israel 10 39.17 16.531 43.45 11.6 59.7 30.9 51.5
Jersey 7 69.8 2.443 70.9 67 72.4 67.3 72.4
Liberia 5 77.66 20.859 73.5 53.1 99.2 63.5 99
Marshall Islands 5 63.1 6.789 60.7 56.7 70.4 57.4 70.3
Netherlands Antilles 9 50.7 47.111 74.6 0.9 99.8 2.4 4.3
Panama 2 42.35 0.495 42.35 42 42.7 42 42.7
Singapore 474 54.228 21.034 54.7 0.5 99.6 42.9 68.8
South Korea 67 46.337 15.092 47.6 4 76.1 38.6 57.1
Country Mean St. Dev. Median Min Max 25th Per 75th Per
Australia 696 66.423 18.905 66.2 1.4 100 54.05 79.6
Austria 156 41.787 25.126 43.65 0.1 97.7 23 58.1
Belgium 176 29.146 22.058 27.55 0 82.4 8.05 46.55
Canada 1320 52.657 28.566 54.55 0.5 100 28.45 76.7
Denmark 173 28.147 22.458 23 0.4 85.7 7.7 46.1
Finland 229 54.325 26.289 59.7 2.4 99.8 35.2 75.7
France 587 58.885 23.363 63.3 0.1 99.3 48 75
Germany 631 51.376 19.08 52.6 2.1 99.4 41 64.6
Greece 286 16.95 19.902 7.3 0 78.3 2.1 25.5
Ireland 118 76.883 15.165 78.65 6 99.7 69.8 86.4
Italy 500 43.082 22.767 50 0.2 92.7 22.15 59.2
Japan 4145 28.282 16.254 26.7 0.1 90.2 15.9 37
Luxembourg 29 27.969 17.372 27.7 2.6 60.1 14.6 42.3
Netherlands 319 50.135 26.995 56.8 0.5 100 26.3 69.3
New Zealand 124 58.718 16.989 59.25 10.1 96.8 45.6 70.15
Norway 173 30.881 21.936 27.3 0.3 89.3 11.5 47.9
Portugal 96 14.035 16.078 7 0.1 63.9 2 21.2
Spain 375 36.418 25.243 40.5 0.1 95.5 10.7 55.6
Sweden 350 40.183 26.223 43.35 0.3 98.8 13 60.2
Switzerland 411 66.871 22.187 71.1 1.1 100 49.2 83.3
Turkey 61 27.73 13.838 25.3 0.1 57.6 18.7 40.2
United Kingdom 3022 83.733 12.593 86.2 0 100 77.2 93.1
Appendix 1: Country Statistics
This table presents the summary statistics for all countries represented in our dataset, broken down by market type and by data. The first set of country
statistics come from the GRI data, specifically, their Index Corporate Governance Quotient. We chose to only look at the index-based score, as any trends in
the country statistics should be visible in either score. As can be seen in this set of statistics, the GRI developed economies observations span a number of
countries and account for most developed economies. The countries span geography, from Europe to Asia, to South America, and span size, from Australia
to Marshall Islands. The next set of summary statistics presents GRI's index-based score's emerging economies observations. Comparing the emerging and
developed economies, we see that the developed economies have on average much higher scores overall. The emerging economies GRI country means
range from 28-54 and the developed economy GRI country means are generally in the 50's and 60's. Portugal and Greece are outliers with scores in the
teens. The United Kingdom has the highest mean of 84. The same pattern is visible in the CLSA data, which composes the last two sets of statistics. Looking
at the CLSA statistics, again, we see that the developed economiess generally received higher corporate governance scores than the emerging economies did.
Panel A: CLSA Emerging Economies
Panel B: CLSA Developed Economies
Panel B: GRI Emerging Economies - Index CGQ Only
GRI Developed Economies
Independent Variables *(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed
Expanded Firm Variables yes yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Company FE yes yes
R² 0.080 0.133 0.125 0.223 0.212 0.230 0.730 0.796
Adjusted-R² 0.0782 0.130 0.116 0.190 0.206 0.222 0.606 0.631 Country=26.4%
Additional Adjusted-R² for All Firms 0.052 -0.014 0.06 0.128 0.144 0.4 0.41 Company=37.3%
Adjusted-R² for Multinationals 0.114 0.125 0.111 0.196 0.169 0.158 0.628 0.687 Country=6.1%
Additional Adjusted-R² for Multinationals 0.011 -0.014 0.071 0.055 0.044 0.459 0.529 Company=53.69%
Adjusted-R² for Single Market Firms 0.071 0.155 0.150 0.311 0.246 0.266 0.629 0.649 Country=35.49%
Additional Adjusted-R² for Single Market Firms 0.044 -0.005 0.156 0.175 0.195 0.383 0.383 Company=32.73%
Company 51.6 39.2
Year 7.20% 6.78%
Country 18 17.32
Company 62.9 48.2
Panel C: Anova Results for Single Market Firms
Source of Variation Additional R ² Additional Adjusted-R ²
Year 1.21% 1.14%
Country 0.52 3.3
Appendix 2 - CLSA Multinationals Robustness Tests
The models below explore the relative importance of firms and countries in explaining corporate governance variance and what impact multinationals firms have on this importance. The table shows that
regardless of whether we look at multinationals or single market firms in emerging economies, the importance of firm characteristics is greater than the country characteristics. We determined multinationals
by matching the firms in the CLSA data to firms listed in the Directory of Corporate Affiliations (DCA). Multinationals were determined by whether or not they had subsidiaries. Panel A explores the OLS and
xtmixed results whiles Panel B and C looks at Anova models. In the top highlighted row of Panel A, we see the entire sample of firms in emerging economies. The middle highlighted row shows multinations in
emerging economies. And the bottom highlighted row show single market firms in emerging economies. Comparing all three samples on top of each other, we see that the results are roughly the same across
the board. Firms take on a greater importance in emerging markets regardless of whether they are multinationals or single market firms. The random effects models suggest that firms are even more
important in multinationals and that in single market firms countries are statistically equal to firms in importance.
Panel A: OLS and xtmixed Results
Panel B: Anova Results for Multinationals
Source of Variation Additional R ² Additional Adjusted-R ²
Independent variables *(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth 0.000** -0.030 -0.028
(0.000) (0.030) (0.033)
Financial Dependence -0.053 -0.007 -0.004
(0.076) (0.060) (0.046)
Closely Held Shares -0.200** -0.188** 0.012
(0.028) (0.037) (0.068)
Log (Assets) 3.078** 3.056** 3.432
(0.447) (0.748) (3.697)
Cash/Total Assets 21.78** 8.709 -13.890
(5.325) (11.510) (16.090)
Antidirector x Legal 2.755** 1.708** 1.987** 0.438 2.736
(0.350) (0.471) (0.610) (0.811) (4.471)
GDP per capita 0.000 -0.000** -0.001** 0.000 0.002
(0.000) (0.000) (0.000) (0.000) (0.001)
Stock Market Cap/GDP 0.09** 0.085** 0.092** 0.017 0.069+
(0.013) (0.016) (0.023) (0.013) (0.040)
Expanded Firm Variables yes yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Company FE yes yes
Observations 6,810 6,659 3,382 1,832 6,810 6,659 6,810 1,832 6,810 6,810
R² 0.009 0.096 0.129 0.159 0.265 0.263 0.655 0.647
Adjusted-R² 0.008 0.095 0.126 0.145 0.262 0.260 0.568 0.513
Additional Adjusted-R² 0.087 0.031 0.05 0.254 0.252 0.314 0.253
Country Random Effect 16.210 16.245
(2.665) (2.671)
Company Random Effect 15.346 15.368
(0.405) (0.403)
Residual 18.167 17.940
(0.174) (0.172)
Independent variables *(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth 0.000** -0.013 -0.034
(0.000) (0.028) (0.030)
Financial Dependence -0.025 0.009 -0.023
(0.065) (0.050) (0.052)
Closely Held Shares -0.254** -0.207** 0.006
(0.028) (0.037) (0.073)
Log (Assets) 3.398** 2.682** 4.069
(0.432) (0.737) (3.623)
Cash/Total Assets 23.28** 15.00 -16.12
(5.074) (11.00) (15.83)
Antidirector x Legal 4.129** 2.969** 3.071** 0.240 1.367
(0.355) (0.476) (0.588) (0.704) (4.141)
GDP per capita 0.000* -0.001** -0.001** 0.000115 0.003+
(0.000) (0.000) (0.000) (0.000182) (0.001)
Stock Market Cap/GDP 0.122** 0.113** 0.119** 0.00526 0.017
(0.013) (0.016) (0.023) (0.0142) (0.037)
Expanded Firm Variables yes yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Company FE yes yes
Observations 6,810 6,659 3,382 1,832 6,810 6,659 6,810 1,832 6,810 6,810
R ² 0.005 0.154 0.200 0.214 0.356 0.356 0.685 0.670
Adjusted-R² 0.00424 0.153 0.196 0.201 0.354 0.353 0.606 0.546
Additional Adjusted-R² 0.149 0.043 0.048 0.35 0.349 0.254 0.193
Country Random Effect 17.097 17.187
(2.781) (2.796)
Company Random Effect 13.944 14.026
(0.377) (0.376)
Residual 17.544 17.348
(0.168) (0.166)
Appendix 3 - GRI Developed Economies Except the United Kingdom and Japan
Panel A: OLS and xtmixed Results for Index CGQ
The tables below examine the relative importance of firms and countries in explaining corporate governance in GRI developed economies. Specifically, these tables are intended to explore the
importance of the United Kingdom and Japan in our developed economy results. We test this by excluding these two markets together and compare our results to those for the full set of
developed economies. Panel A gives the results for models using the Index-based Corporate Governance Quotient while Panel B gives the results for the Industry-Based Corporate Governance
Quotient. We see below that removing the UK and Japan weakens the importance of countries relative to firms, and that this is especially true for the Index-based corporate governance
quotient. However, we can also see that removing these two countries does not change our finding about the importance of country characteristics in developed economies. Thus, we can be
confident that our trend is not driven by specific countries. Correlations are marked with an * for 5% significance, ** for 1% significance, and *** for 0.1% significance.
Panel B: OLS and xtmixed Results for Industry CGQ
Source of Variation Additional R² Additional Adjusted-R²
Year 17.10% 16.70%
Industry 14.40% 12.10%
Country 13 12.9
Firm 30.3 25.8
Source of Variation Additional R² Additional Adjusted-R²
Year 14.50% 14.10%
Industry 14.60% 12.20%
Country 11.1 10.8
Firm 33.5 29
Source of Variation Additional R² Additional Adjusted-R²
Year 17.10% 16.70%
Industry 24.40% 19.60%
Country 10.9 11.4
Firm 22.4 19.8
Source of Variation Additional R² Additional Adjusted-R²
Year 14.50% 14.10%
Industry 24.70% 19.70%
Country 9 9.3
Firm 25.5 23
Panel C: Index CGQ, 3-digit SIC Codes
Panel B: Industry CGQ, 2-digit SIC Codes
Panel D: Industry CGQ, 3-digit SIC Codes
Appendix 4: GRI Emerging Economies Anova Results with Industry Included
The tables below shows the coefficient estimates from the Anova models of GRI corporate governance
ratings for emerging economies. In contrast to previous emerging GRI economies results using Anova
specifications, the models below include industry as an intermediate level of analysis. We understand
industry to be embedded within years, but crossing countries, so we proceed with the following
hieararchy in our analysis: year, industry, country, and firm. These tables are intended to explore
whether our previous results that firms explain greater variance than countries in emerging economies
is actually capturing industry effects. What we see in the results below is that industry does capture
some of the variation in corporate governance ratings. However, the main result holds even to the
inclusion of industry effects: the importance of firm effects in explainin ratings variation is still larger
than country effects. Panels A and B focuses on the 2-digit SIC codes for the Index and Industry CGQ's,
respectively. Panels C and D focus on the 3-digit SIC codes for the Index and Industry CGQ's, again
respectively.
Panel A: Index CGQ, 2-digit SIC Codes
Source of Variation Additional R² Additional Adjusted-R²
Year 0.08% 0.04%
Industry 5.32% 4.85%
Country 51.45 51.64
Firm 20.08 14.82
Source of Variation Additional R² Additional Adjusted-R²
Year 0.08% 0.04%
Industry 4.24% 3.76%
Country 54.66 54.87
Firm 18.74 13.66
Source of Variation Additional R² Additional Adjusted-R²
Year 0.08% 0.04%
Industry 12.59% 10.86%
Country 46.03 46.88
Firm 18.23 13.57
Source of Variation Additional R² Additional Adjusted-R²
Year 0.08% 0.04%
Industry 11.27% 9.52%
Country 49.79 50.71
Firm 16.58 12.06
Panel C: Index CGQ, 3-digit SIC Codes
Panel D: Industry CGQ, 3-digit SIC Codes
Appendix 5: GRI Developed Economies Anova Results with Industry Included
Panel A: Index CGQ, 2-digit SIC Codes
The tables below shows the coefficient estimates from the Anova models of GRI corporate
governance ratings for developed economies. In contrast to previous emerging GRI economies results
using Anova specifications, the models below include industry as an intermediate level of analysis.
We understand industry to be embedded within years, but crossing countries, so we proceed with
the following hieararchy in our analysis: year, industry, country, and firm. These tables are intended
to explore whether our previous results that countries explain greater variance than firms in
developed economies is actually capturing industry effects. What we see in the results below is that
industry does capture some of the variation in corporate governance ratings. However, the main
result holds even to the inclusion of industry effects: the importance of country effects in explainin
ratings variation is still larger than firm effects in developed economies. Panels A and B focuses on
the 2-digit SIC codes for the Index and Industry CGQ's, respectively. Panels C and D focus on the 3-
digit SIC codes for the Index and Industry CGQ's, again respectively.
Panel B: Industry CGQ, 2-digit SIC Codes
Independent variables *(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth 0.006 0.010 -0.002
(0.008) (0.027) (0.030)
Financial Dependence 0.145 -0.732 1.973
(0.184) (0.640) (1.589)
Closely Held Shares 0.035 0.0804 0.090
(0.072) (0.074) (0.097)
Log (Assets) -0.292 -0.843 5.726
(1.005) (1.454) (6.546)
Cash/Assets -2.391 2.809 -27.34
(10.18) (11.76) (22.97)
Antidirector x Legal 3.299 3.611 8.203+ -0.745 11.63
(2.433) (3.235) (4.547) (2.564) (7.886)
GDP per capita 0.000 -0.001 -0.001 0.000 -0.002+
(0.001) (0.001) (0.001) (0.001) (0.001)
Stock Market Cap/GDP -0.024+ -0.021 -0.031 0.031* 0.032+
(0.012) (0.017) (0.022) (0.012) (0.018)
Expanded Firm Variables yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Firm FE yes yes
Observations 1,223 1,110 666 323 1,223 1,110 1,223 323 1,223 1,223
R² 0.202 0.257 0.251 0.408 0.331 0.351 0.730 0.824
Adjusted-R² 0.199 0.251 0.236 0.347 0.325 0.343 0.655 0.708
Additional Adjusted-R² 0.055 -0.003 0.117 0.112 0.13 0.336 0.359
Country Random Effect 11.536 13.725
(7.725) (7.044)
Company Random Effect 12.240 11.939
(0.782) (0.673)
Residual 15.528 12.321
(0.352) (0.278)
Appendix 6 - GRI Emerging Economies, Excluding Tax Havens
The tables below show the coefficient estimates of various models for the emerging economies in the GRI dataset, but exlcuding tax havens. These small, island countries are present throughout the GRI dataset. The
regressions here are intended to test whether including them along with the emerging economies changes our results. Thus, we have rerun all of our initial models for emerging economies, but on a restricted sample that
excludes the tax havens. More generally, the regressions below explore the relative importance of countries and firm in explaining corporate governance ratings of firms in emerging economies. The OLS models include
different combinations of observable firm and country characterics including both the original, limited set of firm variables and the expanded firm variables. We analyze additional adjusted R² to determine additional
variance explained by each set of variables. The results below show that the importance of firms effects in emerging economies does not depend on the inclusion of tax havens. Firm variables continue to explain greater
governance variance than country variables do, even on this restricted sample of emerging economies without tax havens. At times, the results in the random effects models are statistically not differentiable, given the large
standard errors. We take these results together to confirm our overall finding that, in emerging economies, firm characteristics range from anywhere to roughly equal to significantly more important than country
charcteristics in explain corporate governance variance. Panels A and B, respectively, present OLS and xtmixed results for the Index-based corporate governance quotient, and the Industry-based corporate governance
quotient. Panels C and D present Anova results, again for the Index- and Industry-based corporate governance quotient, respectively. Correlations are marked with an * for 5% significance, ** for 1% significance, and *** for
0.1% significance.
Panel A: OLS and xtmixed Results for Index CGQ
Independent variables *(1)* *(2)* *(3)* *(4)* *(5)* *(6)* *(7)* *(8)* *(9)* *(10)*
OLS OLS OLS OLS OLS OLS OLS OLS xtmixed xtmixed
Sales Growth 0.008 0.018 0.008
(0.008) (0.027) (0.030)
Financial Dependence 0.156 -0.440 1.631
(0.160) (0.685) (1.946)
Closely Held Shares -0.032 0.0490 0.014
(0.079) (0.082) (0.097)
Log (Assets) 0.104 -0.076 6.547
(1.090) (1.555) (6.375)
Cash/Assets 4.249 14.10 -23.25
(10.74) (14.46) (25.65)
Antidirector x Legal 3.148 3.493 7.980+ -1.282 12.89
(2.385) (2.933) (4.437) (2.814) (8.103)
GDP per capita 0.000 -0.001 -0.001 0.000 -0.001+
(0.001) (0.001) (0.001) (0.001) (0.001)
Stock Market Cap/GDP -0.026* -0.022 -0.032 0.023 0.025
(0.012) (0.016) (0.022) (0.014) (0.019)
Expanded Firm Variables yes
Year FE yes yes yes yes yes yes yes yes yes
Country FE yes yes yes yes
Firm FE yes yes
Observations 1,223 1,110 666 323 1,223 1,110 1,223 323 1,223 1,223
R² 0.167 0.224 0.212 0.376 0.277 0.297 0.73 0.828
Adjusted-R² 0.164 0.218 0.196 0.311 0.27 0.288 0.655 0.714
Additional Adjusted-R² 0.057 -0.018 0.106 0.093 0.111 0.391 0.423
Country Random Effect 7.088 8.728
(4.66) (7.02)
Company Random Effect 13.668 13.386
(0.81) (0.747)
Residual 15.260 12.588
(0.35) (0.29)
Panel B: OLS and xtmixed Results for Industry CGQ
Panel C: Anova Results for Index CGQ
Source of Variation Additional R ² Additional Adjusted-R ²
Year 20.20% 19.80%
Country 13 12.7
Company 39.8 33
Panel D: Anova Results for Industry CGQ
Source of Variation Additional R ² Additional Adjusted-R ²
Company 45.1 38.3
Year 16.80% 16.40%
Country 11.1 10.8
Variable Median Mean St Dev. Min Max Observations
Corporate Governance Score 3.19 3.48 1.21 1.00 5.99 9,736
Antidirector x Legal 6.01 6.11 1.64 -0.26 8.67 9,720
GDP per capita 34,587 32,873 6,780 2,387 56,625 9,727
Stock Market Cap / GDP 134.12 131.80 89.89 17.51 617.05 9,720
2yr Sales Growth 9.00 107.50 6,342.60 -1,114.69 481,801.50 9,178
Financial Dependence -2.02 -12.35 288.44 -11,268.59 230.00 6,911
Closely Held Shares 22.08 27.23 23.15 0.00 100.00 8,283
Log (Assets) 15.64 15.70 1.85 8.94 22.05 9,229
Cash/Total Assets 0.08 0.16 1.42 0.00 101.96 7,663
Fixed Assets/Total Assets 0.21 0.28 0.25 0.00 0.99 9,127
SEC Compliance 0.00 0.10 0.30 0.00 1.00 9,729
CurrentRatio 1.34 1.74 1.75 0.00 61.55 7,304
Leverage 0.59 0.81 6.26 -0.10 290.80 8,719
PE Ratio 15.95 17.88 75.46 -1,771.98 2,211.00 7,556
Price-to-book Ratio 1.98 2.58 10.32 -400.67 178.85 7,333
Quick Ratio 0.91 1.25 1.65 0.00 60.72 7,283
Return on Assets 5.72 6.37 9.51 -142.97 169.87 9,170
R&D Intensity 1,812.78 9,519.02 98,476.00 0.00 ########## 4,147
CapitalExpenditure 4.05 5.48 6.35 0.00 155.29 8,680
Cash/Dividends 4.84 19.03 731.13 -26,302.00 46,526.00 6,938
3yr Dividends Growth 9.56 11.21 27.92 -100.00 367.23 8,531
5yr Income Growth 10.49 13.82 23.51 -100.00 272.13 7,849
5yr Sales Growth 7.57 10.18 19.52 -100.00 507.86 8,918
Short-Term Debt 183,524 6,554,576 39,800,000 0 921,000,000 9,152
5yr Assets Growth 7.10 9.68 15.11 -61.08 259.14 8,878
Total Debt 53.89 121.52 1,265.85 -76,200.00 68,188.54 9,225
Appendix 7 - FTSE Variables
Panel A: Summary Statistics
Corporate Governance Score 1.00
Antidirector x Legal 0.57*** 1.00
GDP per capita -0.53*** -0.22*** 1.00
Stock Market Cap / GDP -0.04*** 0.27*** 0.08*** 1.00
2yr Sales Growth -0.02 -0.01 0.01 0.01 1.00
Financial Dependence 0.01 0.03 -0.02 0.00 0.00 1.00
Closely Held Shares -0.10*** 0.10*** -0.13*** 0.22*** 0.03*** 0.00 1.00
Log (Assets) -0.20*** -0.42*** 0.09*** -0.10*** 0.00 -0.08*** -0.18*** 1.00
Cash/Total Assets 0.01 0.03*** -0.01 0.00 0.00 0.00 0.00 -0.08*** 1.00
Fixed Assets/Total Assets 0.01 0.05*** -0.01 0.02 0.01 0.03*** 0.02 -0.08*** -0.05*** 1.00
SEC Compliance 0.21*** 0.03*** -0.14*** 0.00 -0.01 0.01 -0.08*** 0.21*** -0.01 0.05*** 1.00
CurrentRatio -0.08*** 0.01 0.10*** 0.13*** 0.00 -0.02 0.06*** -0.19*** 0.04*** -0.23*** -0.03 1.00
Leverage 0.04*** 0.04*** -0.03*** -0.01 0.00 0.00 0.04*** -0.06*** 0.83*** 0.00 -0.01 -0.01 1.00
PE Ratio -0.02 -0.02 0.02 0.01 -0.01 0.00 0.01 -0.02 0.00 0.01 -0.02 0.00 0.00
Price-to-book Ratio 0.00 -0.02 -0.01 0.03 0.00 0.00 -0.01 -0.05*** 0.02 -0.01 -0.01 0.01 0.01
Quick Ratio -0.07*** 0.02 0.07*** 0.14*** 0.00 -0.03 0.08*** -0.17*** 0.04*** -0.20*** -0.01 0.96*** -0.02
Return on Assets 0.03*** 0.01 -0.06*** 0.10*** -0.01 0.01 0.02 -0.10*** -0.03*** 0.07*** 0.03*** 0.07*** -0.04***
R&D Intensity 0.06*** 0.09*** -0.05*** 0.01 0.00 0.00 -0.04 -0.15*** 0.46*** -0.07*** -0.01 0.18*** 0.54***
CapitalExpenditure 0.03 0.02 -0.04*** 0.00 0.02 0.03 0.00 -0.09*** -0.03 0.53*** 0.06*** -0.10*** 0.00
Cash/Dividends -0.03 -0.02 0.02 -0.01 0.00 0.00 -0.01 0.01 0.00 -0.01 0.00 -0.01 0.00
3yr Dividends Growth -0.07*** -0.03*** 0.07*** 0.03*** 0.00 0.01 0.01 0.07*** -0.05*** 0.02 0.03*** 0.04*** -0.04***
5yr Income Growth -0.03 -0.01 -0.01 0.02 0.24*** 0.03 -0.01 0.05*** -0.02 -0.06*** 0.02 0.03 0.00
5yr Sales Growth -0.04*** 0.00 -0.01 0.09*** 0.29*** 0.01 0.04*** 0.00 -0.02 0.02 -0.01 0.04*** -0.03
Short-Term Debt 0.03*** -0.07*** -0.05*** -0.04*** 0.00 -0.35*** -0.05*** 0.40*** -0.01 -0.15*** 0.13*** -0.08*** 0.00
5yr Assets Growth 0.01 0.02 -0.08*** 0.07*** 0.04*** 0.01 0.00 0.06*** -0.02 -0.05*** -0.04*** 0.07*** -0.02
Total Debt -0.01 -0.03*** -0.02 -0.01 0.00 -0.07*** 0.00 0.10*** 0.00 -0.02 -0.01 -0.02 0.00
PE Ratio 1.00
Price-to-book Ratio 0.00 1.00
Quick Ratio 0.00 0.01 1.00
Return on Assets 0.02 0.11*** 0.08*** 1.00
R&D Intensity -0.01 0.01 0.19*** -0.17*** 1.00
CapitalExpenditure 0.01 0.02 -0.07*** 0.15*** -0.05** 1.00
Cash/Dividends -0.01 0.01 -0.01 -0.01 0.00 0.00 1.00
3yr Dividends Growth 0.00 0.02 0.04*** 0.23*** -0.06*** 0.08*** 0.00 1.00
5yr Income Growth -0.02 0.02 0.05*** 0.18*** -0.03 0.11*** 0.01 0.24*** 1.00
5yr Sales Growth 0.00 0.02 0.05*** 0.02 0.05*** 0.15*** 0.00 0.09*** 0.55*** 1.00
Short-Term Debt -0.02 -0.01 -0.06*** -0.08*** -0.01 -0.11*** 0.00 -0.03 -0.02 0.00 1.00
5yr Assets Growth 0.01 0.03 0.07*** 0.08*** -0.03 0.17*** -0.01 0.11*** 0.43** 0.58*** 0.05*** 1.00
Total Debt -0.01 0.18*** -0.01 -0.02 0.00 -0.02 0.00 0.00 0.00 0.01 0.11*** 0.02 1.00
5yr Income
Growth
5yr Sales
Growth
Short-Term
Debt
5yr Assets
Growth
Total
Debt
Panel B: Correlations
Quick
Ratio
Return on
Assets
R&D
Intensity
Capital
Expenditure
Cash/
Dividends
3yr Dividends
Growth
Fixed Assets/
Total Assets
SEC
Compliance
Current
Ratio Leverage
PE Ratio
Price-to-book
Ratio
Corporate Governance
Score
Antidirector x
Legal
GDP per
capita
Stock Market
Cap / GDP
2yr Sales
Growth
Financial
Dependence
Closely Held
Shares Log (Assets)
Cash/Total
Assets