Contents lists available at ScienceDirect
Energy Policy
journal homepage: www.elsevier.com/locate/enpol
Are oil and gas ﬁrms more likely to engage in unethical practices than other
ﬁrms?
Kartick Gupta
Centre for Applied Financial Studies, University of South Australia, 37 North Terrace, Adelaide 5000, Australia
A R T I C L E I N F O
JEL classiﬁcations:
D73
Q40
Q43
Keywords:
Unethical practices
Oil and gas ﬁrms
Competition
Legal determinants
A B S T R A C T
In this study we investigate the unethical practices undertaken by the oil and gas ﬁrms vis-à-vis other ﬁrms. We
ﬁnd that oil and gas ﬁrms operating in non-competitive industries are more likely to engage in unethical
practices. The results are particularly strong in countries that rank poorly in legal enforcement, regulations and
institutional framework, and import less oil. Policy makers should consider undertaking steps to encourage
competition in the industry to limit unethical practices. Further, the country-level enforcement laws,
regulations, and institutional quality need to be reformed, especially in developing countries, to discourage
ﬁrms from gaining undue beneﬁts.
1. Introduction
The role of oil and gas ﬁrms engaged in unethical practices1
to gain
undue beneﬁts remains controversial with a number of studies
indicating that corruption is prevalent in the oil and gas industry. In
an inﬂuential report published by OECD (Foreign Bribery Report),2
more than 400 cases of bribery are investigated. Out of these cases
extractive industries (19%) emerged a most corrupt industry. Similarly,
a report published by Transparency International, a global advocate of
the ﬁght against corruption, found the oil and gas sector was third
amongst the 19 bribery-prone sectors surveyed.3
Al-Kasim et al. (2013)
conduct an in-depth interview with Norwegian specialists in oil
regulation, while Ofori and Lujala (2015) study the Ghana market.
Both the papers indicate that corruption is a concern in an oil and gas
industry. However, as highlighted by Al-Kasim et al. (2013), empirical
evidence into the relationships between oil and gas industries and
corruption is limited to date.
The role of country-level determinants in discouraging oil and gas
ﬁrms from engaging in unethical practices has been overlooked in the
literature. For example, as per the resource curse phenomenon,
resource boom leads to lower GDP (Robinson et al., 2006). Countries
that are rich in natural resources can experience either positive or
negative economic growth (Kolstad and Wiig, 2009; Al-Kasim et al.,
2013; Leite and Weidmann, 1999). In fact, Mehlum et al., (2006a,
2006b) and Robinson et al. (2006) suggest that the relationship
between natural resources and growth of a country is conditional on
the quality of institutions. Abundant natural resources and better
institutional quality play a complementary role in increasing the
economic growth. On the country, countries that have poor institutional
quality have lower economic growth even if there is abundant
supply of natural resources. Further Majbouri (2016) shows that
resource-rich countries have a lower level of entrepreneurship as the
focus turns from increasing productivity and eﬃciency to rent-seeking
and patronage.
Additionally, there is a dearth of studies investigating the role of
competition in driving ﬁrms to engage in unethical practices. This is
highlighted by Al-Kasim et al. (2013), where the authors indicate
competition levels and governance inﬂuence unethical practices.
Similarly, Gupta (2016) shows that competition in an industry is a
strong determinant of oil and gas stock returns.
We address this gap using a comprehensive dataset of 25,702 ﬁrmyear
observations drawn from 53 countries spanning 2002–2012. We
show that oil and gas ﬁrms are not more likely to engage in unethical
practices as compared to other ﬁrms. However, the relationship
between oil and gas ﬁrms and unethical practices is stronger and
statistically signiﬁcant when we condition our results on the competition
level of the industry. We show that oil and gas ﬁrms operating in a
non-competitive industry are more likely to engage in unethical
practices.
Next, we examine the role of country-level determinants in
http://dx.doi.org/10.1016/j.enpol.2016.10.009
Received 18 January 2016; Received in revised form 26 September 2016; Accepted 12 October 2016
E-mail address: Kartick.gupta@unisa.edu.au.
1
We include a range of misconducts, such as corruption, bribery, lobbying, money laundering etc. under the common term “unethical practices”.
2
http://www.oecd.org/corruption/oecd-foreign-bribery-report-9789264226616-en.htm.
3
http://archive.transparency.org/publications/publications/bribe_payers_index_2011.
Energy Policy 100 (2017) 101–112
0301-4215/ © 2016 The Author. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by/4.0/).
Available online 18 October 2016
crossmark
dissuading ﬁrms from engaging in unethical practices. We ﬁnd the oil
and gas ﬁrms operating in non-competitive industries are more likely
to be involved in unethical practices and this is predominantly seen in
countries that rank poorly in legal enforcement, regulations, and
institutional quality. This suggests that a weak institutional framework
may exacerbate corruption and potentially increases the ﬁrm's intention
to engage in unethical practices. Additionally, we ﬁnd the eﬀect is
stronger in oil-rich countries, i.e., countries that depend less on oil
imports. Thus our results support the resource curse phenomenon
where resource-rich countries suﬀer from rent-seeking, patronage and
corruption primarily due to poor institutional quality.
This study contributes to the literature in a number of ways. First,
prior studies mostly use case studies to highlight the corruption
practices in the oil and gas industries. Although, these studies enhance
our understanding of why ﬁrms engage in unethical practices, the
empirical evidence that includes ﬁrm-level data from multiple countries
is limited to date. To the best of our knowledge, this is the ﬁrst
study to test this link using a comprehensive ﬁrm-level data covering
multiple countries.
Second, previous studies mostly undertake an analysis on a single
country and therefore cross-country variation in regulations, laws and
institutional quality are not properly explored. For instance, considerable
evidence suggests that the tendency to undertake unethical
practices is linked to the socioeconomic condition of the country, such
as the resource curse phenomenon. This is also consistent with Jeong
and Weiner (2012), where the authors note that country-level determinants
play an important role in controlling the ﬁrm's intention to
engage in unethical practices. Therefore, we complement the literature
by empirically documenting that country-level determinants are important
attributes in explaining the likelihood of ﬁrms engaging in
unethical practices.
Third, considerable evidence indicates that product market competition
inﬂuences corporate decision-making. For instance, Shleifer and
Vishny (1997 p.738) claim that product market competition is “the
most powerful force toward economic eﬃciency in the world”. In the
context of oil and gas ﬁrms, Gupta (2016) show that the stock returns
and oil price shocks are diﬀerent for the ﬁrms operating in a
competitive vis-à-vis non-competitive industry. In this paper, we show
that the competition level is also an important factor that inﬂuences a
ﬁrm's tendency to engage in unethical practices.
The results from this paper may be of particular interest to policy
makers, regulators and market participants. Extractive industries, such
as oil and gas, have a signiﬁcant impact on the economy of the country
through revenue generation, tax collection and export. One of the
prime challenges is to control rent-seeking, patronage and corruption
and to develop strong institutional framework that encourages productivity,
eﬃciency and innovation (Kolstad and Søreide, 2009). The
results from this study suggest that low competition in the oil and gas
industry increases the likelihood of ﬁrms engaging in unethical
practices. Therefore, policy makers should consider undertaking steps
to encourage competition in the industry. Encouraging competition will
increase eﬃciency, productivity, and cost control and at the same time
limit the ill eﬀects of rent-seeking and patronage. The results also
suggest that country-level enforcement laws, regulations and institutional
quality need to be reformed, especially in developing countries,
to discourage ﬁrms from gaining undue beneﬁts.
Extractive Industries Transparency Initiative (EITI) and other
initiatives are undertaken to encourage countries and ﬁrms to be more
transparent in disclosing revenues. However, becoming a member of
EITI is although an important step, but not the only way to limit
unethical practices in the extractive industries. Following Kolstad and
Søreide (2009), policy makers should focus more on curbing rentseeking,
patronage and corruption problems than macroeconomic
management.
The rest of this paper is organised as follows: The literature is
reviewed in Section 2. Data and Methodology is presented in Section 3.
In Section 4, we present empirical results and their implications.
Finally, Section 5 summarises research ﬁndings and concludes.
2. Background literature
The issue of corruption and its negative eﬀect on the business
environment has been well documented in literature. The focus of the
literature is on the resource curse phenomenon and this arises mainly
from rent-seeking, patronage, poor institutional quality and corruption
(Al-Kasim et al., 2013; Kolstad and Søreide, 2009). Rent-seeking
occurs when individuals are more interested in obtaining economic
beneﬁts from society through manipulation of the social and political
environment. Patronage refers to a situation where government
attempts to stay in power by providing economic and other incentives
to its supporters. Consistent with this, Frynas (2010) alleges that the
incentives to engage in unethical practices are higher in the oil and gas
industry as it has great political power. Resource curse phenomenon is
primarily seen in resource-rich countries with poor institutional
quality, but is not limited to these countries. For instance, evidence
suggests that energy ﬁrms deliberately reported lower prices than the
actual market price in order to pay a lower royalty cost (New York
Times, 1/23/2006).4
Several theoretical models are developed to explain why ﬁrms are
engaged in unethical practices. For instance, Harstad and Svensson
(2011) present a theoretical model and illustrates that ﬁrms can either
comply, bribe or lobby in order to extract favourable regulations.
Similarly, Fredriksson et al. (2004) present a theoretical model and
show that an increase in corruption leads to a less stringent energy
policy.
Another strand of literature highlights the nexus between unethical
practices, regulation and lobbying. Bjertnæs and Fæhn (2008) document
that energy-dependent industries are often exempt from paying
taxes in many countries. The possible reason for this favourable
treatment is likely to be attributed to powerful lobby groups. They also
add that the likelihood of engaging in corruption increases signiﬁcantly
in the oil and gas industry as the sector involves high-value investment
and interaction with government oﬃcials. Schweitzer (2010) allege that
oil companies are likelier to spend a signiﬁcant amount of money on
lobbying to promote their self-interest. Further, Slack (2012) documents
that ﬁrms are more interested in window dressing socially
responsible practices and this is prevalent in developing countries.
Literature also suggests that group lobbying can inﬂuence regulation.
For instance, four major oil companies (BP, Caltex, Mobil and
Shell) in Australia successfully lobbied against the Australian
Government to remove the regulation of unleaded petrol prices
(Valadkhani, 2013). Similarly, Marques et al. (2010) states that oil
and gas companies lobby against alternative energy to protect their
self-interest. This ﬁnding is supported by Huang et al. (2007) and
Sovacool (2009) who note that strong lobbying from traditional energy
sources results in a delay of alternative energy usage. Collectively, the
evidence suggests that ﬁrms engage in unethical practices in order to
extract favourable regulation and to generate positive perception
through lobbying and other activities.
Several studies show that in addition to ﬁrm-level determinants,
country-level determinants also drive incentive to engage in unethical
practices. For instance, Jeong and Weiner (2012) ﬁnd that countrylevel
determinants play an important role in controlling the ﬁrm's
intention to engage in unethical practices. The authors note that ﬁrms
are likelier to engage in unethical practices in countries that have poor
legal enforcement and regulations. Similarly, Baughn et al. (2010)
indicate that the chances of ﬁrms engaging in corruption are likely to be
greater in countries where corruption is tolerated. In a diﬀerent
context, Voyer and Beamish (2004) show that the ﬂow of Foreign
4
http://faculty.washington.edu/mturn/NYT1-27-06Oilproﬁts.pdf.
K. Gupta Energy Policy 100 (2017) 101–112
102
Direct Investment (FDI) is impeded the presence of a weak countrylevel
legal framework and regulations. The authors show that a
negative relationship exists between corruption in emerging economies
and Japanese FDI. Thus, trust and reputation inﬂuences the investment
ﬂow in a country.
With a view to curb corruption, especially in resource-rich countries,
EITI was launched in 2002. The major objective of EITI is to
increase transparency and raise public awareness in Government
revenue and company payments in the extractive industry. However,
the eﬀectiveness of EITI in curbing corruption is debatable. For
instance, Kasekende et al. (2016) show that EITI has not helped in
reducing corruption scores.
3. Data and methodology
We use the Asset4 dataset which primarily covers ﬁrm-level
information related to environmental, social, economic and governance
(ESG) indicators. Within this dataset, a binary indicator reports
whether a ﬁrm is suspected of engaging in unethical practices. This
indicator is deﬁned as, “Is the company under the spotlight of the
media because of a controversy linked to bribery and corruption,
political contributions, improper lobbying, money laundering, parallel
imports or any tax fraud?” Asset4 returns “Y” if the ﬁrm is reportedly
engaged in unethical practices. We term this indicator as “Unethical
Practices”. Asset4 does not solely rely on the feedback of the company
as multiple sources, such as stock exchanges ﬁlings, annual reports,
company websites and various other media outlets are used to verify
the accuracy and quality of the information. We restrict our sample
period from 2002 to 2012 as the coverage of the Asset4 dataset starts
from 2002.
3.1. Product market competition
We construct Sales Herﬁndahl-Hirschman Index (HHI) as a proxy
for product market competition. This measure is extensively used in
the literature, particularly in the industrial organisation literature (see
Tirole, 1988, among others). It is calculated by squaring the market
share of each ﬁrm in the industry within the country and then adding
the resulting numbers. The competition measure is calculated for each
industry within a country and represented as a formula below:
∑Sales HHI s=j c t
i
N
i j c t, ,
=1
, , ,
2
j
(1)
where si,j,c,t is the sales market share of ﬁrm i in industry j, country c,
in year t. By the measure of construction, the theoretical range of HHI
measure lies between zero and one. A low HHI suggests that the
industry is competitive, whereas a high HHI suggests that the industry
is non-competitive. We use the Worldscope database to obtain sales
ﬁgures of publicly listed ﬁrms.
3.2. Control variables
In order to control the eﬀect of other variables that inﬂuence a
ﬁrms’ intention to engage in unethical practices, we use the size of the
ﬁrm, risk, growth opportunities, and block holdings. To control the
inﬂuence of ﬁrm size on unethical practices, we include a log of total
assets, with the value of total assets converted from domestic currency
to equivalent US$. We include leverage, the stock return standard
deviation and illiquidity as the measures of riskiness of the ﬁrm. We
measure leverage as a long-term debt divided by the total assets.
Return SD is computed as the daily standard deviation of stock returns
over the last one year. We calculate illiquidity as per the Lesmond et al.
(1999) model, where a stock with no change in price over a period of
time is considered illiquid. To proxy growth opportunities, we include
PTBV and ROA. PTBV is the price to book value and ROA is the return
on assets. We include Blockholding (percentage shares closely held) as
a measure of monitoring by large owners. We retrieve these variables
from Datastream and Worldscope. We winsorize the control variables
at the 1 and 99 percentile to minimize the eﬀect of outliers. To improve
the quality of the data, we follow Ince and Porter (2006) and Chui et al.
(2010) to address the data issues noted in Datastream and Worldscope.
The empirical analysis is undertaken using Stata Special Edition
(version 13.1).5
3.3. Data description
In Table 1, we present a brief description of our sample and
country-level determinants. The ﬁnal sample consists of 25,702 ﬁrmyear
observations from 53 countries, dependent territories, and oﬀshore
ﬁnancial centres. We ﬁnd that many countries have a small
number of ﬁrm-year observations, such as Jordan, Nigeria and the
United Arab Emirates. We consider excluding countries with less than
10 ﬁrm-year observations but these do not alter our main ﬁndings.
Similarly, excluding dependent territories and oﬀ-shore ﬁnancial
centres, such as the British Virgin Islands and Cayman Islands, among
others, do not aﬀect our main ﬁndings.
We also report country-level determinants retrieved from the World
Economic Forum (WEF) report. The ﬁrst country-level determinant is
Irregular Payment and Bribes. As per the WEF, this variable is the
average score across the ﬁve components of the Executive Opinion
Survey deﬁned as, “In your country, how common is it for ﬁrms to
make undocumented extra payments or bribes connected with (a)
imports and exports; (b) public utilities; (c) annual tax payments; (d)
awarding of public contracts and licenses; (e) obtaining favourable
judicial decisions? ”. The higher the score, the less likely irregular
payments and bribes are made.
The second country-level determinant we include in this study is
Favoritism in Decisions of Government Oﬃcials. This variable is also
derived from the Executive Opinion Survey and deﬁned as, “To what
extent do government oﬃcials in your country show favoritism to wellconnected
ﬁrms and individuals when deciding upon policies and
contracts?” The higher the score, the less likely the government oﬃcials
show favoritism in the country.
The third country-level determinant is Ethical Behaviour of Firms.
This variable is deﬁned as, “How would you compare the corporate
ethics (ethical behaviour in interactions with public oﬃcials, politicians,
and other enterprises) of ﬁrms in your country with those of
other countries in the world?” A higher score indicates less likelihood
of the ﬁrms engaging in unethical practices.
The last country-level determinant we include in this study is
Trustworthiness and Conﬁdence. Similar to the previous three determinants,
the higher the score, the more trustworthy the ﬁrms in the
country.
In Table 2 report, we group the ﬁrms into oil and gas ﬁrms and
other ﬁrms. We use the International Classiﬁcation Benchmark (ICB)
provided by FTSE International to assign stocks to either oil and gas or
other industries. As expected, the oil and gas ﬁrms comprise a small
percentage (6.92%) of the overall sample. We ﬁnd that the sample
coverage has progressively increased from 2002 except in the year
2012.
In Table 3, we report the summary statistics of the key variables
used in this study. We ﬁnd on average 10% of ﬁrms reported in the
media are suspected of bribery, corruption, political corruption or
lobbying. The mean and median Sales HHI is 0.152 and 0.091
respectively, indicating that on average ﬁrms operate in a moderately
competitive industry. The indicators of riskiness in the ﬁrm-average
leverage ratio, return SD and illiquidity are reported as 18.37%, 2.30%
and 7.26% respectively. The growth indicators of the ﬁrm-average
5
The data and commands are available from the author on request.
K. Gupta Energy Policy 100 (2017) 101–112
103
PTBV and ROA are 2.737% and 6.18% respectively. We also ﬁnd that
on average 25.58% of shares are closely-held. In summary, we ﬁnd the
values of the variables fall within the expected range.
Before undertaking any empirical analysis, we check if the variables
have a unit root problem. We perform a modiﬁed Dickey-Fuller t-test,
proposed by Elliott et al. (1996) to check if the unit root problem exists.
In unreported results, we ﬁnd that all our variables are stationary. The
DF-GLS tau- statistic is outside of the one percent critical value in all
four lags and therefore the null hypothesis of unit root is rejected.
In Table 4, we present the correlation coeﬃcient between our key
variables. We ﬁnd that the relationship between unethical practices of
the oil and gas ﬁrms dummy is statistically not signiﬁcant. Larger ﬁrms
are likelier to engage in unethical practices, while more proﬁtable ﬁrms
and closely held ﬁrm are less likely to do so. We also investigate the
possibility of high multi-collinearity between our independent variables.
In Table 4, we do not ﬁnd a correlation coeﬃcient exceeding
absolute 0.50.
4. Empirical results
4.1. Main results
We model the likelihood of oil and gas ﬁrms engaging in unethical
practices by performing the following logit model:
Table 1
Sample description and country-level determinants.
Country Firm-year Irregular payments &
bribes
Favoritism in decisions of
government oﬃcials
Ethical behaviour of
ﬁrms
Trustworthiness &
conﬁdence
1 Australia 1591 5.71 4.04 5.45 6.32
2 Austria 147 5.39 3.87 5.56 4.98
3 Belgium 213 5.61 4.04 5.34 4.61
4 Bermuda 373 NA NA NA NA
5 Brazil 174 3.87 2.9 3.75 4.96
6 British Virgin Islands 7 NA NA NA NA
7 Canada 1438 5.77 4.22 5.69 5.76
8 Cayman Islands 211 NA NA NA NA
9 Chile 77 5.66 4.12 5.1 5.25
10 China 310 3.99 4 4.17 4.62
11 Czech Republic 14 3.69 2.42 3.55 4.94
12 Denmark 204 6.08 4.42 6.15 5.4
13 Egypt 24 3.35 2.95 4.09 3.48
14 Finland 231 6.65 5.27 6.44 6.18
15 France 796 5.4 3.85 5.22 5.1
16 Germany 545 5.67 4.63 5.67 5.06
17 Gibraltar 5 NA NA NA NA
18 Greece 111 3.57 2.55 3.48 3.16
19 Hong Kong 375 6.11 4.05 5.62 6.53
20 Hungary 13 4.33 2.49 3.73 4.84
21 India 327 3.22 2.79 3.75 5.47
22 Ireland 221 6.14 4.12 5.42 4.41
23 Israel 51 5.39 3.32 4.88 5.63
24 Italy 412 3.84 2.37 3.63 3.73
25 Japan 3535 6.13 4.8 5.81 5.28
26 Jersey 73 NA NA NA NA
27 Jordan 3 4.94 3.78 4.42 3.93
28 Kuwait 7 4.45 2.75 3.95 4.31
29 Luxembourg 40 6.24 4.44 5.92 5.32
30 Malaysia 145 4.7 4.05 5.01 6
31 Mexico 52 3.59 2.91 3.75 4.95
32 Netherlands 303 6.09 5.12 6.01 4.92
33 New Zealand 87 6.69 5.31 6.57 6.46
34 Nigeria 3 2.57 2.16 3.24 4.93
35 Norway 165 6.29 4.99 6.19 5.59
36 Panama 19 4 2.9 3.96 5.06
37 Philippines 60 3.31 3.03 3.98 4.65
38 Poland 79 4.76 3.13 4.08 5.53
39 Portugal 101 5.16 3.05 4.39 3.78
40 Russia 131 3.24 2.59 3.66 3.49
41 Saudi Arabia 23 5.52 4.24 5.11 5.08
42 Singapore 346 6.51 5.42 6.33 6.47
43 South Africa 336 4.59 2.46 4.73 6.72
44 South Korea 358 4.35 2.96 3.81 4.59
45 Spain 389 4.75 3.14 4.14 4.14
46 Sweden 449 6.16 5.33 6.17 5.76
47 Switzerland 565 6.17 4.88 6.24 5.69
48 Taiwan 427 5.11 4.23 5 4.97
49 Thailand 83 3.76 2.83 3.98 4.88
50 Turkey 91 4.48 3.35 4.21 4.76
51 United Arab
Emirates
3 6.36 5.07 5.67 4.8
52 United Kingdom 2624 5.99 4.35 5.76 5.51
53 United States 7335 4.94 3.34 4.9 5.54
Total 25,702
K. Gupta Energy Policy 100 (2017) 101–112
104
U β OILDummy β LogofAsset β Leverage β PTBV
β RoA β Blockholding β ReturnSD β Illiquidity
YearDummies CountryDummies ε
= ∝ + + + +
+ + + +
+ + +
i t i t i t i t i t
i t i t i t i t
i t
, 0 1 , 2 , 3 , 4 ,
5 , 6 , 7 , 8 ,
, (2)
where Ui,t is a dummy variable of ﬁrm i and year t taking the value of one if
the ﬁrm is reported in the media as involved in bribery, corruption,
political corruption, money laundering or lobbying. Oil Dummyi,t is a
dummy variable indicating whether or not the ﬁrm is in the oil and gas
industry.
The model includes a set of control variables that can potentially
inﬂuence a ﬁrm's decision to engage in unethical practices. Following
Gormley and Matsa (2014), we use ﬁxed eﬀects to capture unobservable
heterogeneity and omitted factors that are potentially related to
both unethical practices and industry. We use year-ﬁxed eﬀects to
control for time trends. Our results are qualitatively unchanged if we
use the probit model.
In Table 5, we present our baseline regressions. In Model 1, we ﬁnd
that the Oil Dummy coeﬃcient is 0.0335 but statistically is not
signiﬁcant (t-stat 0.38), suggesting that on an average oil and gas
ﬁrms are not more likely to engage in unethical practices. The results
suggest that larger ﬁrms with more growth opportunities, proﬁtable
ﬁrms, with less closely-held shares (i.e. less monitoring), and liquid
ﬁrms are likelier to engage in unethical ﬁrms. The pseudo R-square of
the model is 20.4%.
In Model 2, we introduce the eﬀect of Sales HHI and check whether
product market competition can inﬂuence a ﬁrm's decision to engage in
unethical practices. We interact Oil Dummy with the Sales HHI of the
industry using the following equation:
U β OILDummy β LogofAsset β Leverage β PTBV
β RoA β Blockholding β ReturnSD β Illiquidity
β OilDummy SalesHHI YearDummies CountryDummies
ε
= ∝ + + + +
+ + + +
+ ( × ) + +
+
i t i t i t i t i t
i t i t i t i t
i t
i t
, 0 1 , 2 , 3 , 4 ,
5 , 6 , 7 , 8 ,
9 ,
, (3)
We ﬁnd the Oil Dummy X Sales HHI coeﬃcient is 0.905 and is
statistically signiﬁcant at the one percent level. This suggests that the
engagement in unethical practices is strongly inﬂuenced by the
competition level within the industry. Oil and gas ﬁrms that operate
in a non-competitive industry are more likely to engage in unethical
practices.
To verify the robustness of our result, we create (2×1) vector of
Sales HHI dummies and interact with the Oil Dummy in Model 3. The
Table 2
Year-wise distribution of sample.
Year Oil ﬁrm-years Non-oil ﬁrm-years Total
2002 39 780 819
2003 40 1117 1157
2004 109 1715 1824
2005 126 1836 1962
2006 137 1858 1995
2007 150 1968 2118
2008 220 2500 2720
2009 240 2878 3118
2010 258 3217 3475
2011 271 3447 3718
2012 190 2606 2796
Total 1780 23,922 25,702
In Table 2, we report the year-wise breakdown of a full sample into oil and gas ﬁrms and
other ﬁrms. We use the International Classiﬁcation Benchmark (ICB) provided by FTSE
International to assign stocks to the oil and gas and other industries.
Table 3
Summary statistics of key variables.
Variable Mean StDev Median Skewness Kurtosis
Unethical Dummy 0.101 0.301 0.000 2.655 8.051
Oil Dummy 0.069 0.254 0.000 3.393 12.514
Sales HHI 0.152 0.147 0.091 2.319 9.644
Log of Asset (~US$) 15.701 1.677 15.558 0.379 3.659
Leverage 18.37% 15.74% 15.88% 0.953 3.840
PTBV 2.737 3.185 1.900 5.441 50.259
ROA (%) 6.183 8.460 5.380 0.512 14.918
Blockholding (%) 25.581 23.869 19.220 0.806 2.671
Return SD 2.30% 1.20% 2.01% 3.599 62.934
Illiquidity 7.26% 6.02% 5.75% 4.183 35.973
In Table 3, we report summary statistics of the key variables used in this study. Unethical
Dummy is a binary indicator deﬁned as, “Is the company under the spotlight of the media
because of a controversy linked to bribery and corruption, political contributions,
improper lobbying, money laundering, parallel imports or any tax fraud?” Oil Dummy
is a dummy variable indicating whether the ﬁrm is located in the oil and gas industry. Log
of Assets is the value of total assets converted from domestic currency to equivalent US$.
We measure leverage as a long-term debt divided by the total assets. Return SD is
computed as daily standard deviation of stock returns over the last one year. We calculate
illiquidity as per the Lesmond et al. (1999) model, where a stock with no change in price
over a period of time is considered illiquid. PTBV is price to book value and ROA is return
on assets. We include Block holding (% shares closely held) as a measure of monitoring of
the ﬁrm.
Table 4
Correlation coefficient.
Unethical dummy Oil dummy Log of asset (~US$) Leverage PTBV ROA (%) Blockholding (%) Return SD Illiquidity
Unethical Dummy 1
Oil Dummy 0.000971 1
Log of Asset (~US$) 0.306***
−0.0237***
1
Leverage −0.00133 0.000512 0.0790***
1
PTBV −0.0138*
−0.0249***
−0.218***
0.0126*
1
ROA (%) −0.0187**
0.0333***
−0.197***
−0.122***
0.328***
1
Blockholding (%) −0.0882***
−0.0284***
−0.0682***
−0.0600***
0.0189**
0.0484***
1
Return SD −0.0256***
0.0722***
−0.150***
0.0142*
−0.0895***
−0.214***
0.0443***
1
Illiquidity −0.101***
−0.0344***
−0.189***
−0.0428***
−0.0611***
−0.0707***
0.187***
−0.0240***
1
In this table, we report a correlation among the key variables used in the study. Unethical Dummy is a binary indicator defined as, “Is the company under the spotlight of the media
because of a controversy linked to bribery and corruption, political contributions, improper lobbying, money laundering, parallel imports or any tax fraud?” Oil Dummy is a dummy
variable indicating whether the firm is located in the oil and gas industry. Log of Assets is the value of total assets converted from domestic currency to equivalent US$. We measure
leverage as a long-term debt divided by the total assets. Return SD is computed as daily standard deviation of stock returns over the last one year. We calculate illiquidity as per the
Lesmond et al. (1999) model, where a stock with no change in price over a period of time is considered illiquid. PTBV is price to book value and ROA is return on assets. We include
Block holding (% shares closely held) as a measure of monitoring of the ﬁrm.
*
Statistically signiﬁcant at the 10%.
**
Statistically signiﬁcant at the ﬁve percent.
***
Statistically signiﬁcant at the one percent level.
K. Gupta Energy Policy 100 (2017) 101–112
105
quantiles are formed on the basis of whether the ﬁrm is located in the
lowest HHI industry or the highest. This procedure will capture
varying degrees of competition within the industry. We also include
Low Sales HHI dummy to control for the direct eﬀect of competition
on the Unethical Dummy but do not report in the table due to
paucity of space. We model this relationship using the following
equation:
U β OILDummy β LogofAsset β Leverage β PTBV
β RoA β Blockholding β ReturnSD β Illiquidity
β OilDummy LowSalesHHIDummy
β OilDummy HighSalesHHIDummy
β LowSalesHHIDummy YearDummies CountryDummies
ε
= ∝ + + + +
+ + + +
+ ( × )
+ ( × )
+ + +
+
i t i t i t i t i t
i t i t i t i t
i t
i t
i t
i t
, 0 1 , 2 , 3 , 4 ,
5 , 6 , 7 , 8 ,
9 ,
10 ,
11 ,
, (4)
Table 5
Oil and gas firms and unethical practices.
(1) (2) (3) (4) (5) (6)
Logistic Regression Panel Regression
Dependent Variable Unethical Dummy Unethical Dummy Unethical Dummy Total Controversies Total Controversies Total Controversies
Oil Dummy 0.0335 0.0903***
(0.706) (0.000)
Oil Dummy X Sales HHI 0.905***
0.362***
(0.001) (0.000)
Oil Dummy x Low HHI −0.671*
−0.00406
(0.084) (0.826)
Oil Dummy x High HHI 0.256**
0.120***
(0.014) (0.000)
Log of Asset (~US$) 0.648***
0.655***
0.679***
0.143***
0.146***
0.147***
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Leverage −0.0795 −0.163 −0.219 −0.215***
−0.229***
−0.232***
(0.588) (0.277) (0.149) (0.000) (0.000) (0.000)
PTBV 0.0395***
0.0418***
0.0405***
0.00948***
0.00973***
0.00988***
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
ROA (%) 0.0238***
0.0213***
0.0214***
0.00329***
0.00317***
0.00315***
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Blockholding (%) −0.00546***
−0.00523***
−0.00498***
−0.00120***
−0.00118***
−0.00116***
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Return SD 1.369 1.935 2.331 2.860***
3.056***
3.033***
(0.580) (0.441) (0.355) (0.000) (0.000) (0.000)
Illiquidity −2.591***
−2.607***
−2.507***
0.493***
0.497***
0.498***
(0.001) (0.002) (0.002) (0.000) (0.000) (0.000)
Country F.E.? Yes Yes Yes Yes Yes Yes
Year F.E.? Yes Yes Yes Yes Yes Yes
N 25,702 25,151 25,151 25,702 25,170 25,170
Pseudo R2
0.204 0.208 0.209
Adjusted R2
0.108 0.110 0.110
In this table, we run a logistic regression. The dependent variable is a binary variable, Unethical Dummy, defined as, “Is the company under the spotlight of the media because of a
controversy linked to bribery and corruption, political contributions, improper lobbying, money laundering, parallel imports or any tax fraud?” Oil Dummy is a dummy variable
indicating whether the firm is located in the oil and gas industry. Sales HHI is industry concentration ratio. Low HHI and High HHI are dummy variables formed using the median
values of Sales HHI. Log of Assets is the value of total assets converted from domestic currency to equivalent US$. We measure leverage as a long-term debt divided by the total assets.
Return SD is computed as daily standard deviation of stock returns over the last one year. We calculate illiquidity as per the Lesmond et al. (1999) model, where a stock with no change
in price over a period of time is considered illiquid. PTBV is price to book value and ROA is return on assets. We include Block holding (% shares closely held) as a measure of monitoring
of the ﬁrm. The model includes country and year ﬁxed eﬀect.
In specification 1, we use the following model: U β OILDummy β LogofAsset β Leverage β PTBV β RoA
β Blockholding β ReturnSD β Illiquidity YearDummies CountryDummies
ε
= ∝ + + + + +
+ + + + +
+
i t i t i t i t i t i t
i t i t i t
i t
, 0 1 , 2 , 3 , 4 , 5 ,
6 , 7 , 8 ,
,
In specification 2, we use the following model: U β OILDummy β LogofAsset β Leverage β PTBV β RoA
β Blockholding β ReturnSD β Illiquidity β OilDummy SalesHHI
YearDummies CountryDummies ε
= ∝ + + + + +
+ + + + ( × )
+ + +
i t i t i t i t i t i t
i t i t i t i t
i t
, 0 1 , 2 , 3 , 4 , 5 ,
6 , 7 , 8 , 9 ,
,
In specification 3, we use the following model: U β OILDummy β LogofAsset β Leverage β PTBV β RoA
β Blockholding β ReturnSD β Illiquidity β OilDummy LowSalesHHI
β OilDummy HighSalesHHI β LowSalesHHI YearDummies
CountryDummies ε
= ∝ + + + + +
+ + + + ( × )
+ ( × ) + +
+ +
i t i t i t i t i t i t
i t i t i t i t
i t i t
i t
, 0 1 , 2 , 3 , 4 , 5 ,
6 , 7 , 8 , 9 ,
10 , 11 ,
,
In specification 4, 5, and 6 we replace the dependent variable Unethical Dummy with Total Controversies.
Standard errors are clustered at the firm-level.
*
statistically signiﬁcant at the 10%.
**
statistically signiﬁcant at the ﬁve percent.
***
statistically signiﬁcant at the one percent level. p-values are given in parentheses and are based on robust standard errors.
K. Gupta Energy Policy 100 (2017) 101–112
106
The results indicate that oil and gas ﬁrms operating in a competitive
industry are less likely to engage in unethical practices. The Oil Dummy
X Low HHI dummy is −0.671 and is statistically signiﬁcant only at the
10% level, suggesting that the evidence is not very strong. On the
contrary, we ﬁnd that oil and gas ﬁrms that operate in a lesscompetitive
industry are more likely to engage in unethical practices.
The Oil Dummy X High HHI dummy is positive and statistically
signiﬁcant at the ﬁve percent level.
In Model 4, 5, and 6 we use total number of controversies reported
for a ﬁrm.6
This addresses the concern that in a logit model we are
grouping together a ﬁrm that is reported in media as engaged in one
unethical practice vis-à-vis multiple unethical practices. We retrieve
the total number of controversies from Asset4, deﬁned as “Number of
controversies published in the media linked to public health or
industrial accidents harming the health & safety of third parties
(non-employees and non-customers); business ethics in general,
political contributions or bribery and corruption; tax fraud, parallel
imports or money laundering; anti-competitive behaviour (e.g., antitrust
and monopoly), price-ﬁxing or kickbacks; and activities in critical,
undemocratic countries that do not respect fundamental human rights
principles.”
In Model 4, we ﬁnd that Oil Dummy is positive and statistically
signiﬁcant at the one percent level. This suggests that, as compared to
logit model, oil and gas ﬁrms appear to be more likely to engage in
unethical practices. The results are diﬀerent compared to the logit
model as in Model 1 we do not ﬁnd statistically signiﬁcant results for
the Oil Dummy. In Model 5, we interact Oil Dummy with Sales HHI
and ﬁnd a statistically signiﬁcant coeﬃcient of 0.362. This indicates
that the competition level is an important determinant of unethical
practices. In fact, the results from Model 6 suggest that the tendency of
ﬁrms involved in unethical practices is prevalent in non-competitive
industries. Summing up, the results from Table 5 suggest that oil and
gas ﬁrms operating in non-competitive industries are more likely to
engage in unethical practices. Thus, ﬁrms are more likely to take
advantage of the dominant position in the industry in order to maintain
dominance over the long run and to potentially create entrance barriers
for new ﬁrms by undertaking activities, such as bribing, corruption and
lobbying.
4.2. Country-level determinants
In Table 5, we document that oil & gas ﬁrms operating in noncompetitive
industries are more likely to engage in unethical practices.
However, we have not explicitly considered cross-country diﬀerences in
institutional quality. For instance, the resource curse phenomenon
suggests that countries rich in natural resources can experience either
positive or negative economic growth (Kolstad and Wiig, 2009; AlKasim
et al., 2013; Leite and Weidmann, 1999). In fact, Mehlum et al.
(2006a, 2006b) and Robinson et al. (2006) suggest that the relationship
between natural resources and growth of a country is conditional
on the quality of institutions. Abundant natural resources and better
institutional quality play a complementary role in increasing the
economic growth. On the country, countries that have poor institutional
quality have lower economic growth even if there is abundant
supply of natural resources. Similarly, the country-level legal enforcement,
regulatory and institutionary framework can discourage ﬁrms
from engaging in unethical practices (Baughn et al., 2010; Jeong and
Weiner, 2012). Therefore, in Table 6, we investigate the role of
country-level determinants in inﬂuencing the ﬁrm's decision to engage
in unethical practices. We consider four diﬀerent proxies of countrylevel
determinants that indicate the level of unethical practices
acceptable in the country. We report our ﬁndings using the interacted
variable of oil and gas ﬁrms and Sales HHI score. This step is
undertaken to avoid a few observations resulting from the interaction
of two dummy variables (Oil Dummy and Sales HHI dummy). For
instance, the interaction term will take the value of zero if the ﬁrm is
not located in the oil and gas industry and if it operates in a competitive
industry.7
This problem is exacerbated when we split the sample using
various country-level determinants.
In Model 1 and Model 2, we bifurcate the sample into Low and
High, based on the median value of Irregular Payments & Bribes. We
document that countries in which unethical practices are widespread,
ﬁrms operating in non-competitive industries are more likely to engage
in unethical practices. The Low and High Oil Dummy X Sales HHI
coeﬃcient is 1.281 (t-stat 3.84) and 0.243 (t-stat 0.54) respectively.
This is consistent with our initial expectation that ﬁrms located in a
weak institutional framework are more likely to engage in unethical
practices. Thus, it appears that ﬁrms take advantage of weak regulations,
laws and poor enforcement in pursuing rent-seeking and patronage
behaviour.
We next split the sample using the median values of Favoritism in
Decisions of Government Oﬃcials in Model 3 and Model 4. We ﬁnd
consistent results as the Oil Dummy X Sales HHI coeﬃcient is positive
and statistically signiﬁcant only for the countries that rank low in the
Favoritism index. The third country-level determinant we include is
Unethical Behaviour of Firms. It can be argued that ﬁrms are more
likely to engage in unethical practices if they observe other ﬁrms doing
the same and deriving undue beneﬁts. Similar to our previous two
country-level proxies, we ﬁnd that ﬁrms are more likely to engage in
unethical practices if the unethical behaviour of ﬁrms in a country is
tolerated. The Low and High Oil Dummy X Sales HHI coeﬃcient is
1.292 (t-stat 3.59) and 0.435 (t-stat 1.04) respectively.
The last proxy we include is Trustworthiness & Conﬁdence in
Model 7 and Model 8. We show that the tendency to engage in
unethical practices increases as the level of trustworthiness and
conﬁdence falls. The above results collectively suggest that the competition
level within the industry and the country-level legal enforcement,
regulatory and institutional framework are important determinants of
the ﬁrm’s tendency to engage in unethical practices. Our results are in
line with Baughn et al. (2010) and Jeong and Weiner (2012) who
indicate that the chances of a ﬁrm engaging in corruption are likely to
increase in countries where corruption is tolerated.
4.3. Level of oil import
In Table 7, we check whether the level of oil imports can explain the
ﬁrm's likelihood to engage in unethical practices. Majbouri (2016)
show that ﬁrms operated in oil-rich countries are less keen on
increasing productive and transparency and are more inclined towards
rent-seeking and patronage. This can be due to a number of reasons,
including to maintain dominance in the domestic market and to create
obstacles in the free ﬂow of oil trade through unfair practices. To this
end, we retrieve country-level oil import data available from the US
Energy Information Administration (USEIA) website. We split the
sample into Low and High oil importing countries, using the median oil
import of all countries each year as a cut-oﬀ.
We ﬁnd that Oil Dummy X Sales HHI coeﬃcient is statistically
signiﬁcant only for low oil importing countries. The Low and High Oil
Dummy X Sales HHI coeﬃcient is 0.937 (t-stat 2.80) and 0.919 (t-stat
1.60) respectively. This indicates that when the oil and gas ﬁrms
operate in a non-competitive industry, it is more likely that they are to
maintain dominance in the domestic market and to limit the free ﬂow
trade of oil and gas trading. Our results are in a similar context to Lee
and Weng (2013) who note that bribery in one’s home country
dampens a ﬁrm’s exports.
6
We thank an anonymous referee for suggesting this additional test.
7
As reported in Table 2, around 93% of the observations pertain to non-oil and gas
ﬁrms.
K. Gupta Energy Policy 100 (2017) 101–112
107
4.4. Robustness checks
4.4.1. Product market competition and classiﬁcation
We report a number of robustness checks in this section. The ﬁrst
concern we address is whether the Sales HHI is a poor proxy of product
market competition. We address this concern by using Asset HHI index
where we substitute sales with total assets. The rationale is that large
ﬁrms in an industry may create entry barriers for new companies or
ready to sustain losses in the short term to discourage other ﬁrms from
entering the industry (Benoit, 1984). Also, due to their large asset base,
these companies may have economies of scale (Bolton and Scharfstein,
1990).
In addition to Asset HHI, we also use Employee HHI. We deﬁne
Employee HHI as the concentration of employees in a given industry.
The notion here is that ﬁrms operating in non-competitive industries
will employ a much larger workforce than their competitors. Therefore,
a high Employee HHI suggests that the majority of the workforce is
employed in a small number of ﬁrms.
The second concern we address is whether ﬁrms are incorrectly
identiﬁed into oil and gas ﬁrms. To negate this concern, we use 48
industry classiﬁcations (FF48) provided by Fama and French (1997).
This classiﬁcation in turn is drawn from the Standard Industrial
Classiﬁcation (SIC)‘s four-digit code. We report the results using
alternative proxy of industry concentration and industry classiﬁcation
in Table 8. We ﬁnd that the results reported in Table 8 are largely
consistent with our baseline ﬁndings. We ﬁnd that oil and gas ﬁrms
operating in non-competitive industries are likelier to engage in
unethical practices using alternative measures of product market
competition or industry classiﬁcation. The interacted coeﬃcient of
Oil Dummy and product market competition measure are positive and
statistically signiﬁcant at the one percent level in all three models
reported in Table 8.
4.4.2. Alternative measure of unethical practices
We primarily base our main ﬁndings using logistic regression.
However, the main dependent variable (Unethical Dummy) is binary
and therefore the extent and severity to which the ﬁrms are engaged in
unethical practices is not very clear.8
Further, we retrieve the data from
Asset4 where the ﬁrm is given a score of one, if the ﬁrm is reported in
the media with the news of corruption, bribery, political contribution or
lobbying. However, it is important to verify the robustness of this
indicator. Although ﬁnding a perfect substitute is diﬃcult, we argue
that ﬁrms engaged in unethical practices are also prone to earnings
management. The motivation to use earnings management comes from
Halter et al. (2009) and Al-Kasim et al. (2013), who suggest that the
ﬁrm should increase transparency in order to reduce corruption.
We use two diﬀerent models of the earnings management to check
the consistency of our result. In Model 1, we use the Dechow and
Dichev (2002) (DD) model for estimating accruals quality. More
speciﬁcally, we regress changes in working capital on the current, past
and future cash ﬂow from operation, i.e.,
WC b b CFO b CFO b CFOj ε∆ = + * + * + * +j t o j t j t t j t, 1 , −1 2 , 3 +1 , (5)
where WC∆ j t, is total current accruals calculated as ∆current assets∆current
liabilities- ∆cash + ∆debt in current liabilities of ﬁrm j in year
t. The changes here refer to the current and previous year. CFO is cash
Table 6
Role of country-level determinants.
Irregular payments & bribes Favoritism Unethical behaviour of ﬁrms Trustworthiness and conﬁdence
Low High Low High Low High Low High
(1) (2) (3) (4) (5) (6) (7) (8)
Oil Dummy X Sales HHI 1.281***
0.243 1.122***
0.531 1.292***
0.435 1.031***
0.203
(0.000) (0.590) (0.001) (0.247) (0.000) (0.299) (0.000) (0.797)
Log of asset (~US$) 0.671***
0.633***
0.693***
0.608***
0.674***
0.631***
0.646***
0.683***
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Leverage −0.440**
0.266 −0.436**
0.373 −0.459**
0.294 −0.239 0.326
(0.028) (0.240) (0.025) (0.117) (0.023) (0.191) (0.138) (0.447)
PTBV 0.0382***
0.0527***
0.0370***
0.0569***
0.0386***
0.0521***
0.0402***
0.0489***
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.008)
ROA (%) 0.0216***
0.0197***
0.0250***
0.0125*
0.0214***
0.0198***
0.0208***
0.0197**
(0.000) (0.001) (0.000) (0.085) (0.000) (0.001) (0.000) (0.011)
Blockholding (%) −0.00456***
−0.00592***
−0.00441***
−0.00568***
−0.00439***
−0.00611***
−0.00546***
−0.00345
(0.006) (0.004) (0.007) (0.008) (0.008) (0.003) (0.000) (0.251)
Return SD 4.477 −4.960 2.058 2.404 4.546 −5.150 3.142 −11.51
(0.125) (0.340) (0.483) (0.641) (0.121) (0.315) (0.229) (0.263)
Illiquidity −1.421 −3.597***
−2.519**
−2.959**
−1.387 −3.599***
−2.322**
−2.443*
(0.248) (0.001) (0.030) (0.013) (0.259) (0.001) (0.026) (0.079)
Country F.E.? Yes Yes Yes Yes Yes Yes Yes Yes
Year F.E.? Yes Yes Yes Yes Yes Yes Yes Yes
N 12,063 12,456 13,010 11,509 11,942 12,577 18,913 5606
Pseudo R2
0.186 0.207 0.200 0.191 0.186 0.206 0.194 0.227
In this table, we run a logistic regression to investigate the role of country-level determinants in influencing the firm's decision to engage in unethical practices. We consider four
different proxies of country-level determinants that indicate the level of unethical practices acceptable in the country. These proxies are from the WEF 2013 report and the definition is
provided in Appendix A of this paper. The dependent variable is the binary variable, Unethical Dummy, deﬁned as, “Is the company under the spotlight of the media because of a
controversy linked to bribery and corruption, political contributions, improper lobbying, money laundering, parallel imports or any tax fraud?” Oil Dummy is a dummy variable
indicating whether the ﬁrm is located in the oil and gas industry. Sales HHI is industry concentration ratio. Log of Assets is the value of total assets converted from domestic currency to
equivalent US$. We measure leverage as a long-term debt divided by the total assets. Return SD is computed as daily standard deviation of stock returns over the last one year. We
calculate illiquidity as per the Lesmond et al. (1999) model, where a stock with no change in price over a period of time is considered illiquid. PTBV is price to book value and ROA is
return on assets. We include Block holding (% shares closely held) as a measure of monitoring of the ﬁrm. The model includes country and year ﬁxed eﬀect. Standard errors are clustered
at the ﬁrm-level.
*
Statistically signiﬁcant at the 10%.
**
Statistically signiﬁcant at the ﬁve percent.
***
Statistically signiﬁcant at the one percent level. p-values are given in parentheses and are based on robust standard errors.
8
We partly address this concern in Table 5 by including total number of controversies
reported against a ﬁrm.
K. Gupta Energy Policy 100 (2017) 101–112
108
ﬂow from operations calculated as net income before extraordinary
items less total accruals. Total accruals is calculated as current accruals
less depreciation and amortization expense. All the variables are scaled
by the average of total assets over the current and previous year. A
higher absolute value of residual suggests lower earnings quality.
In Model 2, we use earnings volatility (the standard deviation of
earnings) as a measure of a ﬁrms’ earnings management (EV). We
follow Dechow and Dichev (2002) and measure the earnings standard
deviation as the standard deviation of a ﬁrm's net income before
extraordinary items scaled by beginning of year total assets. A rolling
window of 10 years is used to compute standard deviation. Larger
values of the earnings volatility suggest that earnings are less trans-
parent.
We present both the models in Table 9. In Model 1, we ﬁnd
evidence that oil and gas ﬁrms operating in non-competitive markets
are more likely to manage earnings. However, the evidence is weak as
the Oil Dummy X Sales HHI coeﬃcient is statistically signiﬁcant only
at the 10% level. In Model 2, we use the EV model and ﬁnd strong
evidence that oil and gas ﬁrms operating in non-competitive markets
have lower earnings quality. We ﬁnd the Oil Dummy X Sales HHI
coeﬃcient positive and statistically signiﬁcant (t-stat 2.19).
We also consider the possibility that our results are driven primarily
by the US-domiciled ﬁrms as they comprise around 28% of the sample
observation. In unreported results, we exclude US-domiciled ﬁrms and
ﬁnd qualitatively similar results. Additionally, we cluster the standard
errors at the country-and industry-level in order to allow the observations
within a country or industry to correlate in some unknown way.
In un-tabulated results, we still ﬁnd that oil and gas ﬁrms operating in
non-competitive industries are more likely to engage in unethical
practices.
5. Conclusions and policy implications
Extractive industries, such as oil and gas have a signiﬁcant impact
on the economy of the country through revenue generation, tax
collection and export. Unethical practices in the oil and gas industry
Table 7
Level of oil import.
Oil Import
Low High
(1) (2)
Oil dummy X sales HHI 0.937***
0.919
(0.005) (0.109)
Log of asset (~US$) 0.627***
0.698***
(0.000) (0.000)
Leverage 0.241 −0.492**
(0.337) (0.014)
PTBV 0.0489***
0.0429***
(0.000) (0.000)
ROA (%) 0.0220***
0.0233***
(0.000) (0.000)
Blockholding (%) −0.00449**
−0.00685***
(0.018) (0.000)
Return SD −6.315 6.338**
(0.227) (0.040)
Illiquidity −2.596**
−2.809*
(0.023) (0.066)
Country F.E.? Yes Yes
Year F.E.? Yes Yes
N 10,877 12,677
Pseudo R2
0.217 0.200
In this table, we run a logistic regression to investigate whether the level of oil import can
explain the firm's likelihood to engage in unethical practices. We retrieve oil import data
available from the US Energy Information Administration (USEIA) website. We split the
sample into Low and High oil importing countries, using the median oil import of all
countries each year as a cut-off measure. The dependent variable is the binary variable,
Unethical Dummy, defined as, “Is the company under the spotlight of the media because
of a controversy linked to bribery and corruption, political contributions, improper
lobbying, money laundering, parallel imports or any tax fraud?” Oil Dummy is a dummy
variable indicating whether the firm is located in the oil and gas industry. Sales HHI is
industry concentration ratio. Log of Assets is the value of total assets converted from
domestic currency to equivalent US$. We measure leverage as a long-term debt divided
by the total assets. Return SD is computed as daily standard deviation of stock returns
over the last one year. We calculate illiquidity as per the Lesmond et al. (1999) model,
where a stock with no change in price over a period of time is considered illiquid. PTBV is
price to book value and ROA is return on assets. We include Block holding (% shares
closely held) as a measure of monitoring of the ﬁrm. The model includes country and year
ﬁxed eﬀect. Standard errors are clustered at the ﬁrm-level.
*
Statistically signiﬁcant at the 10%.
**
Statistically signiﬁcant at the ﬁve percent.
***
Statistically signiﬁcant at the one percent level. p-values are given in parentheses
and are based on robust standard errors.
Table 8
Alternative industry concentration and alternative industry classification.
Using public and private ﬁrms data
Using Fama and French industry classiﬁcation
(1) (2) (3)
Oil dummy X sales
HHI
1.209***
(0.000)
Oil dummy X asset
HHI
1.066***
(0.000)
Oil dummy X
employee HHI
1.067***
(0.000)
Log of asset (~US$) 0.666***
0.650***
0.650***
(0.000) (0.000) (0.000)
Leverage −0.116 −0.0847 −0.0858
(0.447) (0.568) (0.566)
PTBV 0.0394***
0.0413***
0.0398***
(0.000) (0.000) (0.000)
ROA (%) 0.0238***
0.0232***
0.0216***
(0.000) (0.000) (0.000)
Blockholding (%) −0.00576***
−0.00603***
−0.00623***
(0.000) (0.000) (0.000)
Return SD −0.117 2.348 2.035
(0.965) (0.340) (0.413)
Illiquidity −2.653***
−2.478***
−3.249***
(0.002) (0.002) (0.000)
Country F.E.? Yes Yes Yes
Year F.E.? Yes Yes Yes
N 23,448 25,273 24,709
Pseudo R2
0.215 0.207 0.208
In this table, we report the results using alternative proxy of industry concentration and
industry classification. The dependent variable is the binary variable, Unethical Dummy,
defined as, “Is the company under the spotlight of the media because of a controversy
linked to bribery and corruption, political contributions, improper lobbying, money
laundering, parallel imports or any tax fraud?” Oil Dummy is a dummy variable
indicating whether the firm is located in the oil and gas industry. We use Fama and
French (1997) FF48 industry classiﬁcation. Sales HHI is industry concentration ratio.
Asset HHI is calculated by squaring the market asset share of each ﬁrm in the industry
within the country and then adding the resulting numbers. Employee HHI is the
concentration of employees in a given industry. Log of Assets is the value of total assets
converted from domestic currency to equivalent US$. We measure leverage as a longterm
debt divided by the total assets. Return SD is computed as daily standard deviation
of stock returns over the last one year. We calculate illiquidity as per the Lesmond et al.
(1999) model, where a stock with no change in price over a period of time is considered
illiquid. PTBV is price to book value and ROA is return on assets. We include Block
holding (% shares closely held) as a measure of monitoring of the ﬁrm. The model
includes country and year ﬁxed eﬀect. Standard errors are clustered at the ﬁrm-level.
* Statistically significant at the 10%.
** Statistically significant at the five percent.
***
Statistically signiﬁcant at the one percent level. p-values are given in parentheses
and are based on robust standard errors.
K. Gupta Energy Policy 100 (2017) 101–112
109
can have a serious repercussion on the economy of the country.
Available evidence suggests that unethical practices are widespread
in the oil and gas industry but empirical evidence to support this claim
is limited to this date. We have shown that oil and gas ﬁrms are not
more likely to engage in unethical practices. However, we ﬁnd that
industry competition is an important determinant. Oil and gas ﬁrms
operating in a non-competitive industry are more likely to engage in
unethical practices than oil and gas ﬁrms operating in a competitive
industry. This suggests that oil and gas ﬁrms are likely to take
advantage of their dominant position in the industry in order to
maintain control over the long-run and to potentially create entrance
barriers for new ﬁrms by undertaking activities, such as bribing,
corruption and lobbying. The results also suggest that country-level
institutional factors are important, as the occurrence of unethical
practices is widespread in countries where corruption is tolerated.
Our study has strong policy implications for government policy
makers, regulators and market participants. Policy makers should
consider undertaking steps to encourage competition in the industry
to increase eﬃciency, productivity, cost control, and to limit unethical
practices. Further, the country-level enforcement laws, regulations, and
institutional quality need to be reformed, especially in developing
countries, to discourage ﬁrms from gaining undue beneﬁts.
EITI and other initiatives are undertaken to encourage countries
and ﬁrms to be more transparent in disclosing revenues. However,
(Kolstad and Wiig, 2009; Kolstad and Søreide, 2009; Papyrakis et al.,
2016) suggests that EITI is not the only solution to corruption and
multifaceted reforms are needed to control corruption in the extractive
industries. For instance, Kolstad and Søreide (2009) and Kolstad and
Wiig (2009) argue that EITI unintentionally increases rent-seeking
behaviour with the formation of multi-stakeholder committees where
the members have a vested interest. Similarly, Ofori and Lujala (2015)
argue that participating in EITI does not necessarily increase transparency
as observed in Ghana.
Also, as highlighted by Kolstad and Søreide (2009), policy makers
should focus more on addressing factors responsible for resource curse,
such as rent-seeking, patronage and corruption problems than macroeconomic
management. This will encourage ﬁrms to pay more attention
in improving productivity, eﬃciency and innovation (Majbouri,
2016). High penalties, disciplinary actions, enforceable anti-corruption
measures and eﬀective detection are needed to dissuade ﬁrms from
undertaking unethical practices (O’Higgins, 2006).
As noted by Slack (2012), ﬁrms window dress socially responsible
practices that are unlikely to curb unethical practices. A more rigorous,
accountable and operational socially responsible practices are needed
at the ﬁrm-level to address rent-seeking and patronage. Additionally, a
better cooperation is needed among countries as oil and gas ﬁrms may
be domiciled in one country but draw revenue from another country.
An eﬀective collaboration among countries will discourage ﬁrms from
undertaking unethical practices either in the headquartered country or
the producing country. Following Al-Kasim et al. (2013), a future
empirical research investigating the link between corruption and
suboptimal oil production and reserves will further enhance our
understanding of how unethical practices is related to suboptimal
production. The main challenge here is to access oil production and
reserve data at the ﬁrm-level across countries.
Appendix A. Variable deﬁnitions
Variable Deﬁnition Source
A. Country-level variables
1 Irregular payments & bribes Average score across the ﬁve components of the following Executive Opinion Survey
question: In your country, how common is it for ﬁrms to make undocumented extra payments
or bribes connected with (a) imports and exports; (b) public utilities; (c) annual tax
payments; (d) awarding of public contracts and licenses; (e) obtaining favourable judicial
decisions? In each case, the answer ranges from 1 (very common) to 7 (never occurs).
WEF 2013
Report
2 Favoritism in Decisions of
Government Oﬃcials
To what extent do government oﬃcials in your country show favoritism to well-connected
ﬁrms and individuals when deciding upon policies and contracts? [1 = always show favoritism;
7=never show favoritism]
WEF 2013
Report
Table 9
Oil and gas firms and earnings management.
DDstdev MDDstdev SMstdev
(1) (2) (3)
Oil dummy X sales HHI 0.00341*
0.00316 0.0365**
(0.091) (0.112) (0.028)
Log of asset (~US$) −0.00612***
−0.00704***
−0.0553***
(0.000) (0.000) (0.000)
Leverage −0.0107***
−0.00476 −0.165***
(0.001) (0.186) (0.000)
PTBV 0.000944***
0.000698***
−0.00293**
(0.000) (0.005) (0.011)
ROA (%) 0.000327**
0.000136 0.00433***
(0.017) (0.433) (0.000)
Blockholding (%) 0.0000613 0.0000292 0.000237
(0.183) (0.320) (0.284)
Return SD 0.889***
0.684***
7.821***
(0.000) (0.000) (0.000)
Illiquidity 0.0223**
0.0587***
−0.260***
(0.040) (0.002) (0.000)
Country F.E.? Yes Yes Yes
Year F.E.? Yes Yes Yes
N 11,116 6867 12,516
Adj. R2
0.126 0.229 0.084
In this table we use an alternative measure of unethical practices, i.e., earnings
management. We use two different models of the earnings management to check the
consistency of our result. In Model 1, we use the Dechow and Dichev (2002) (DD) model
for estimating accruals quality. In Model 2, we use earnings volatility (the standard
deviation of earnings) as a measure of a ﬁrms’ earnings management (EV). We follow
Dechow and Dichev (2002) and measure earnings standard deviation as the standard
deviation of a ﬁrm's net income before extraordinary items scaled by beginning of year
total assets. A rolling window of 10 years is used to compute standard deviation. Larger
values of the earnings volatility suggest that earnings are less transparent. Oil Dummy is
a dummy variable indicating whether the ﬁrm is located in the oil and gas industry. Sales
HHI is industry concentration ratio. Log of Assets is the value of total assets converted
from domestic currency to equivalent US$. We measure leverage as a long-term debt
divided by the total assets. Return SD is computed as daily standard deviation of stock
returns over the last one year. We calculate illiquidity as per the Lesmond et al. (1999)
model, where a stock with no change in price over a period of time is considered illiquid.
PTBV is price to book value and ROA is return on assets. We include Block holding (%
shares closely held) as a measure of monitoring of the ﬁrm. The model includes country
and year ﬁxed eﬀect. Standard errors are clustered at the ﬁrm-level.
*
Statistically signiﬁcant at the 10%.
**
Statistically signiﬁcant at the ﬁve percent.
***
Statistically signiﬁcant at the one percent level. p-values are given in parentheses
and are based on robust standard errors.
K. Gupta Energy Policy 100 (2017) 101–112
110
3 Ethical Behaviour of Firms How would you compare the corporate ethics (ethical behaviour in interactions with public
oﬃcials, politicians and other enterprises) of ﬁrms in your country with those of other
countries in the world? [1=among the worst in the world; 7 =among the best in the world]
WEF 2013
Report
4 Trustworthiness &
Conﬁdence
This index is a combination of three ﬁnancial market development indicators, namely
Soundness of banks, Regulations of securities exchange, and Legal rights index. Each of the
indicators are deﬁned in turn.
WEF 2013
Report
Soundness of banks is deﬁned as, “How would you assess the soundness of banks in your
country? [1 = insolvent and may require a government bailout;7 = generally healthy with
sound balance sheets”
Regulation of securities exchanges is deﬁned as, “How would you assess the regulation
and supervision of securities exchanges in your country? [1 = ineﬀective; 7=eﬀective]”
Legal rights index is deﬁned as, “Degree of legal protection of borrowers and lenders’
rights on a0–10 (best) scale”]
B. Firm-level variables
1 Unethical dummy This binary indicator is deﬁned as, “Is the company under the spotlight of the media because
of a controversy linked to bribery and corruption, political contributions, improper
lobbying, money laundering, parallel imports or any tax fraud?”
Asset4
2 Total Controversies Number of controversies published in the media linked to public health or industrial accidents
harming the health & safety of third parties (non-employees and non-customers);
business ethics in general, political contributions or bribery and corruption; tax fraud,
parallel imports or money laundering; anti-competitive behaviour (e.g., anti-trust and
monopoly), price-ﬁxing or kickbacks; activities in critical, undemocratic countries that do
not respect fundamental human rights principles.
Asset4
3 Oil Dummy Oil Dummy is a dummy variable indicating whether the ﬁrm is located in the oil and gas
industry
ICG
4 Log of Asset (~US$) Log of total assets is the value of total assets converted from domestic currency to
equivalent US$
Worldscope
5 Leverage Leverage is long-term debt divided by the total assets Worldscope
6 PTBV Price to book value ratio Worldscope
7 ROA (%) Return on assets Worldscope
8 Block holding percentage shares closely held Worldscope
9 Return SD Daily standard deviation of stock returns over the last one year Author's own
calculation
10 Illiquidity Illiquidity is calculated as per theLesmond et al. (1999) model, where a stock with no
change in price over a period of time is considered illiquid
Author's own
calculation
11 DD Measure of accruals quality that captures the mapping of current accruals into last period
and current period into next period cash ﬂows
Author's own
calculation
12 EV Measure of earnings standard deviation calculates as the standard deviation of a ﬁrm's net
income before extraordinary items scaled by beginning of year total assets. A rolling window
of 10 years is used to compute standard deviation.
Author's own
calculation
C. Industry concentration
1 Sales HHI It is calculated by squaring the market sales share of each ﬁrm in the industry within the
country and then adding the resulting numbers.
Author's own
calculation
2 Asset HHI It is calculated by squaring the market asset share of each ﬁrm in the industry within the
country and then adding the resulting numbers.
Author's own
calculation
3 Employee HHI We deﬁne Employee HHI as the concentration of employees in a given industry. Author's own
calculation
References
Al-Kasim, F., Søreide, T., Williams, A., 2013. Corruption and reduced oil production: an
additional resource curse factor? Energy Policy 54, 137–147.
Baughn, C., Bodie, N.L., Buchanan, M.A., Bixby, M.B., 2010. Bribery in international
business transactions. J. Bus. Ethics 92, 15–32.
Benoit, J.-P., 1984. Financially constrained entry in a game with incomplete information.
Rand J. Econ. 15, 490–499.
Bjertnæs, G.H., Fæhn, T., 2008. Energy taxation in a small, open economy: social
eﬃciency gains versus industrial concerns. Energy Econ. 30, 2050–2071.
Bolton, P., Scharfstein, D.S., 1990. A theory of predation based on agency problems in
ﬁnancial contracting. Am. Econ. Rev. 80, 93–106.
Chui, A.C., Titman, S., Wei, K.J., 2010. Individualism and momentum around the world.
J. Financ. 65, 361–392.
Dechow, P.M., Dichev, I.D., 2002. The quality of accruals and earnings: the role of
accrual estimation errors. Account. Rev. 77, 35–59.
Elliott, G., Rothenberg, T.J., Stock, J.H., 1996. Eﬃcient tests for an autoregressive unit
root. Économ.: J. Économ. Soc. 64, 813–836.
Fama, E., French, K., 1997. Industry costs of equity. J. Financ. Econ. 43, 153–193.
Fredriksson, P.G., Vollebergh, H.R., Dijkgraaf, E., 2004. Corruption and energy eﬃciency
in OECD countries: theory and evidence. J. Environ. Econ. Manag. 47, 207–231.
Frynas, J.G., 2010. Corporate social responsibility and societal governance: lessons from
transparency in the oil and gas sector. J. Bus. Ethics 93, 163–179.
Gormley, T.A., Matsa, D.A., 2014. Common errors: how to (and not to) control for
unobserved heterogeneity. Rev. Financ. Stud. 27, 617–661.
Gupta, K., 2016. Oil price shocks, competition, and oil & gas stock returns—global
evidence. Energy Econ. 57, 140–153.
Halter, M.V., De Arruda, M.C.C., Halter, R.B., 2009. Transparency to reduce corruption?
J. Bus. Ethics 84, 373–385.
Harstad, B., Svensson, J., 2011. Bribes, lobbying, and development. Am. Polit. Sci. Rev.
105, 46–63.
Huang, M.-Y., Alavalapati, J.R., Carter, D.R., Langholtz, M.H., 2007. Is the choice of
renewable portfolio standards random? Energy Policy 35, 5571–5575.
Ince, O., Porter, R., 2006. Individual equity return data from Thomson datastream:
handle with care!. J. Financ. Res. 29, 463–479.
Jeong, Y., Weiner, R.J., 2012. Who bribes? Evidence from the United Nations' oil-for‐
food program. Strateg. Manag. J. 33, 1363–1383.
Kasekende, E., Abuka, C., Sarr, M., 2016. Extractive industries and corruption:
investigating the eﬀectiveness of EITI as a scrutiny mechanism. Resour. Policy 48,
117–128.
Kolstad, I., Søreide, T., 2009. Corruption in natural resource management: implications
K. Gupta Energy Policy 100 (2017) 101–112
111
for policy makers. Resour. Policy 34, 214–226.
Kolstad, I., Wiig, A., 2009. Is transparency the key to reducing corruption in resourcerich
countries? World Dev. 37, 521–532.
Lee, S.H., Weng, D.H., 2013. Does bribery in the home country promote or dampen ﬁrm
exports? Strateg. Manag. J. 34, 1472–1487.
Leite, C.A., Weidmann, J., 1999. Does mother nature corrupt? Natural resources,
corruption, and economic growth. Natural Resources, Corruption, and Economic
Growth (June 1999). IMF Working Paper 99/85. pp. 1–34.
Lesmond, D., Ogden, J., Trzcinka, C., 1999. A new estimate of transaction costs. Rev.
Financ. Stud. 12, 1113–1141.
Majbouri, M., 2016. Oil and entrepreneurship. Energy Policy 94, 10–15.
Marques, A.C., Fuinhas, J.A., Manso, J.P., 2010. Motivations driving renewable energy in
European countries: a panel data approach. Energy Policy 38, 6877–6885.
Mehlum, H., Moene, K., Torvik, R., 2006a. Cursed by resources or institutions? World
Econ. 29, 1117–1131.
Mehlum, H., Moene, K., Torvik, R., 2006b. Institutions and the resource curse. Econ. J.
116, 1–20.
O’Higgins, E.R., 2006. Corruption, underdevelopment, and extractive resource
industries: addressing the vicious cycle. Bus. Ethics Q. 16, 235–254.
Ofori, J.J.Y., Lujala, P., 2015. Illusionary transparency? Oil revenues, information
disclosure, and transparency. Soc. Nat. Resour. 28, 1187–1202.
Papyrakis, E., Rieger, M., Gilberthorpe, E., 2016. Corruption and the extractive
industries transparency initiative. J. Dev. Stud., 1–15.
Robinson, J.A., Torvik, R., Verdier, T., 2006. Political foundations of the resource curse.
J. Dev. Econ. 79, 447–468.
Schweitzer, D., 2010. Oil Companies and Sustainability: More than Just an Image?
Working paper. University of Munich.
Shleifer, A., Vishny, R.W., 1997. A survey of corporate governance. J. Financ. 52,
737–783.
Slack, K., 2012. Mission impossible?: adopting a CSR-based business model for
extractive industries in developing countries. Resour. Policy 37, 179–184.
Sovacool, B.K., 2009. Rejecting renewables: the socio-technical impediments to
renewable electricity in the United States. Energy Policy 37, 4500–4513.
Tirole, J., 1988. The Theory of Industrial Organization. MIT press, Cambridge,
Massachusetts.
Valadkhani, A., 2013. Do petrol prices rise faster than they fall when the market shows
signiﬁcant disequilibria? Energy Econ. 39, 66–80.
Voyer, P.A., Beamish, P.W., 2004. The eﬀect of corruption on Japanese foreign direct
investment. J. Bus. Ethics 50, 211–224.
K. Gupta Energy Policy 100 (2017) 101–112
112