Page 1
1
The Effect of the UK Bribery Act 2010 on Growth, Cost of Equity and Value
Suhee Kim1 and William Rees2
Abstract
We examine the impact of the U.K. Bribery Act of 2010 on expected growth, cost of equity and firm
valuation. The Act is seen as being unexpected and severe and hence caused a noticeable shock to equity
markets. We estimate the impact on U.K. firms with high exposure to bribery and find a reduction in cost
of equity, expected growth, bid-ask spreads and an improvement in anti-bribery governance. Taken together
these findings challenge the conventional view that anti-bribery regulation negatively impacts on the value
of regulated firms operating in bribery-sensitive environments. For our sample the value increasing impact
of reduced cost of equity outweighs the reduction in expected growth and we find that the net effect of the
Bribery Act on firm value is positive in the medium term. The effect is stronger for bribery-sensitive firms
that have higher leverage and greater ownership concentration.
EFM Classification: 750-Law, Ethics and Finance
Keywords: bribery, U.K. Bribery Act 2010, anti-bribery law, cost of equity, growth rate, firm value,
discounted residual income valuation, corporate transparency, information asymmetry, anti-bribery
management
1Accounting and Finance, University of Edinburgh Business School, 29 Buccleuch Place, Edinburgh EH8
9JS, UK; Correspondence and presenter: [email protected]
2Accounting and Finance, University of Edinburgh Business School, 29 Buccleuch Place, Edinburgh EH8
9JS, UK; [email protected]
Page 2
2
1. Introduction
Transparency International defines bribery as “offering, promising, giving, accepting or soliciting of an
advantage as an inducement for an action which is illegal, unethical or a breach of trust” (Transparency
International, 2017). It has been the focus of academic and practitioners’ attention due to its prevalence and
economically distortive effects. On the basis of the World Bank Enterprise Surveys, bribery between firms
and the public sector is estimated at $1 trillion per year worldwide. 1 The Dow Jones State of Anti-
corruption (2011) reports that over 40% of surveyed companies mention unethical activities conducted by
competitors as the main cause losing businesses. There is not only an economic but also a moral case for
understanding the effect of bribery. The majority of large bribery instances that come to light involve bribes
by firms in developed economies where the target is a developing country (Cheung, Rau, & Stouraitis,
2012).2 The contract may be won by firms in the EU or US, and paid to intermediaries in Asia or Africa,
but it is the eventual customers in these developing economies who suffer the impact. Given the economic
impact of bribery, it is perhaps surprising that more research has not been conducted since Rose-
Ackermon’s (1975) seminal study on the economics of corruption. However, studying bribery is obviously
handicapped by the fact that most activity is hidden. For example, Karpoff, Lee, and Martin (2014) estimate
that there is a one in sixteen chance of bribery by US firms being discovered.
Previous studies on the effects of corporate bribery on firm value have provided contradictory
evidence. Some find a positive value impact from using bribes resulting from the contracts obtained
(Cheung et al., 2012; Karpoff, Lee, & Martin, 2015; Zeume, 2017), whereas others point out the decrease
in firm value caused by the increase in risk (Murphy, Shrieves, & Tibbs, 2009), the decrease in investment
1 The estimation figure is obtained from the appendix in a chapter by Kaufmann in the World Economic Forum’s
2005-2006 Global Competitiveness Report, which was based on data from the 2000 World Bank Enterprise Survey
and the World Economic Forum’s 2004 Global Competitiveness Survey carried out in several countries.
2 Cheung et al. (2012) analyse 166 cases of bribery during 1971-2007 originating in Japan, 43, USA, 41, France, 23,
Germany, 16, UK, 10, South Korea, 8, Italy, 5, and 20 others. These bribes targeted Japan, 27, South Korea, 13,
Nigeria, 10, Philippines, 8, Indonesia, 7, Lesotho, 7, and 94 others.
Page 3
3
(Birhanu, Gambardella, & Valentini, 2016), or the increased strains on the firms’ relation with their
stakeholders (Serafeim, 2014). The contradictory results may come from focusing on the short-run impact
by using an event study, a cross-sectional analysis based on survey data, or hypothetical data on bribery
occurrence. The studies relying on short-term data cannot capture the influence of bribery on firms’ strategic
and managerial prospects, which are critical for firm value. Furthermore, our understanding of how market
participants incorporate bribery risk into their valuation remains limited. Zeume (2017) measures the two-
day market reaction for U.K. firms affected by U.K. Bribery Act 2010 and non-U.K. competitors without
exposure to the anti-bribery law. He suggests that the passage of the Act was unexpected and surprising in
its rigor, but his analysis of value requires that the market does not anticipate the act and is able to evaluate
its impact rapidly. If the market either anticipates or gradually incorporates, the value impact of Zeume’s
(2017) analysis will be partial. In addition to the value effect, Zeume (2017) is able to show lower economic
activity by UK firms affected by the act when contrasted with competitors that were unaffected. He
concludes that the “evidence is consistent with the notion that bribes facilitate doing business in certain
countries. Imposing unilateral anti-bribery regulation on some firms hurts these firms but benefits their
unregulated competitors” (Zeume, 2017, p1481). Our interpretation of his evidence is that the Bribery Act
hurt growth in the firms covered, but growth is not the only dimension of interest.
The focus of our study is to examine the net effect of the Bribery Act on firm valuation in the medium
term. We estimate the impact of the Bribery Act on implied growth rate, implied cost of equity and the
three-year abnormal returns for these UK firms classified as bribery sensitive. We classify firms as bribery
sensitive according to their activities in countries where bribery is thought to be relatively common.3 Using
a sample of publicly listed U.K. firms that are active around the passage of Bribery Act, we employ a
3 We construct a measure of bribery risk using Transparency International (TI)’s Corruption Perceptions Index (CPI).
Transparency International assigns each country a score between zero to 100, indicating the higher the score, the less
exposure to corruption. Instead of using ‘operational subsidiaries’ which was used in Zeume’s study, we obtain a
weighted average score of the CPI for all ‘geographic sales segments’ by its sales ratio. We classify “high bribery-
risk” firms as those with the weighted CPI score below 55.
Page 4
4
difference-in-differences (DiD) design.4 Previous literature studying the effect of bribery on firm value
have difficulty identifying a causal relationship. Firstly, studies examining the value impact of identified
bribes suffer from relatively severe sample selection bias as most bribes are not identified. According to
Karpoff, Lee, and Martin (2014), the probability that a bribe-paying firm is identified is only 6.4%, whereas
the estimated tendency of US-listed firms with foreign sales to engage in bribery is 22.9%. Secondly, equity
market reactions to bribery may be unclear. It is not apparent when the market incorporates bribery activity
into prices. Typically, a contract won will impact prices as the news is announced and the results become
apparent. The potential downside from discovery may be incorporated into that reaction or deferred until
the controversy becomes public. To overcome these identification challenges, we exploit the Bribery Act
as an external shock using a version of difference in difference estimation.5 Based on a simple revision of
the standard discounted residual income model of valuation (Easton, Taylor, Shroff, & Sougiannis, 2002),
we simultaneously estimate the changes in the implied cost of equity and the expected growth rate before
and after the Bribery Act and compare the changes between firms with high and low exposure to bribery.
Overall, we document two opposing channels through which the U.K. Bribery Act influences firm
valuation. First, we find the Act reduces expected growth rate of high bribery-risk firms by 3.8 percentage
point, implying a reduction in firm value. This effect is economically significant compared to the estimate
of long-term growth rate of 3.3% for non-bribery-sensitive firms and 6.7% for bribery-sensitive firms. Our
result is consistent with the market expecting bribery-sensitive firms to forgo positive NPV-projects after
4 The draft Bribery Act was passed on 25 March 2009, received Royal Assent on 8 April 2010, and came into forced
on 1 July 2011. In this paper, we use the date of the passage of draft Act (25 March 2009) as a reference time instead
of the actual date of enactment (1 July 2011). This is because our study relies on stock prices, which would have
incorporated the effects of the law since it was passed.
5 The U.K. Bribery Act is used as an exogenous shock for the DiD analysis because 1) it was not covered by media
before the passage of the draft of the Act; 2) there was no residual uncertainty about the passage of the Act due to
OECD sanctions against the U.K. government; and 3) it has a significant impact on firm behavior as a strong shock.
Based on hand-collected data on criminal, administrative, and civil foreign bribery cases and other offence related to
foreign bribery from the OECD annual data on enforcement of the Anti-Bribery convention, the cumulative sanctioned
cases in the U.K. increased from 0 in 2008 to 37 in 2016. Further, the level of foreign bribery enforcement of the U.K.
government has been changed from moderate to active since 2009.
Page 5
5
the Act. Second, we find that the Bribery Act increases the valuation of firms with high exposure to bribery
through the reduction in cost of equity. Specifically, the implied cost of equity of high bribery-risk firms
decreases by 2.7% relative to low bribery-risk firms. As a comparison, the estimated cost of equity for high-
bribery risk firms in our sample firms is 9.9%, implying that these firms experience a reduction of more
than 27% in their cost of equity. The increase in value from the reduction in cost of equity suggests that
investors expect the firms to adjust their business operations and engage in projects that are less exposed to
bribes.6
We perform several robustness checks to validate our main findings. First, we validate our parallel
trend assumptions by splitting the effect of anti-bribery law into different time periods relative to the
baseline year using a series of year dummies (Bertrand & Mullainathan, 2003; Ni & Yin, 2018). We find
no significant and persistent pre-treatment trend of changes in the cost of equity and the long-run growth
rate before the passage of the Bribery Act. We also alleviate concerns that our results are due to any
unobservable shocks which are not relevant to the Bribery Act in the data by performing placebo tests with
artificial event years (Atanasov & Black, 2016). Our placebo tests show no significant effect during the
pseudo-event periods. Further, to mitigate the effects of unobserved omitted variables on our results, we
repeat our analysis with a propensity-score-matched (PSM) sample to compare firms with similar
characteristics prior to the enactment of the Bribery Act and find similar results.
To further corroborate our results, we show that the passage of the Bribery Act indeed leads to an
improvement in transparency and anti-bribery management. According to Jin and Myers (2006), firms
lacking transparency pass risks to outside investors and experience large negative returns. Thus, if the anti-
bribery law affects firm transparency, it will change the regulated firms’ risks and returns. We find that
bribery-sensitive firms have experienced a significant decline in information asymmetry measured by bid-
6 These estimates are based on the results in table 4, regression 3, where the control group estimates are the constant
(r) and the BTMD (g) the pre-Act treatment group and constant plus bribery-risk (r) and BTMD plus BTMD*bribery-
risk (g) and post-Act treatment group are constant plus bribery-risk plus passage-bribery (r) and BTMD plus
BTMD*bribery-risk plus BTMD*passage*bribery-risk (g).
Page 6
6
ask spread up to four years after the Act. This is consistent with bribery-sensitive firms communicating
their anti-bribery efforts to investors, leading to the reduction of the information asymmetry between
outside investors and firm managers. Furthermore, using Asset4 anti-bribery provision data, we find that
bribery-sensitive firms experience a significant increase in anti-bribery management systems after
enactment of the Bribery Act. The improvement in anti-bribery management will decrease the riskiness of
the firms’ business activities by internalizing bribery risk into their risk assessment and due diligence,
resulting in the reduction in the implied cost of equity.
We also perform a set of cross-section analyses to examine how firm characteristics influence the
effect of anti-bribery law on the firm growth and cost of equity. We find that the decrease in implied cost
of equity is stronger among large bribery-sensitive firms. Following the regulatory shock, large firms can
easily alter their business models for the less exposed to bribery because of having more alternative projects
to choose (Aguilera & Vadera, 2008; Collins, Uhlenbruch, & Rodriguez, 2009; D’Souza & Kaufmann,
2013). Thus, the firms’ cash flows become less sensitive to markets, lowering their implied cost of equity.
On the contrary, small firms have difficulty in adjusting their business models immediately. Furthermore,
we find that the results for both reduced cost of equity and reduced growth opportunities are stronger
amongst firms with greater monitoring from external shareholders and debtholders, as proxied by
ownership concentration and leverage level respectively. A greater concentrated ownership will deter firms
from committing misconduct due to investors’ enhanced monitoring power over managerial decisions
(Ding, Qu, & Wu, 2016; Morck, Percy, Rian, & Yeung, 2005; Zhou & Peng, 2012). Moreover, higher
leverage will encourage bribery-sensitive firms facing financial constraints to avoid the risk of severe
penalties due to pressure from debtholders. Therefore, high bribery-risk companies with greater monitoring
pressure from external investors are expected to adjust to the Act more effectively leading to a lower cost
of equity.
As we document two opposing forces on firm valuation, the natural next step is to examine the net
impact of the Bribery Act on firm value. We examine the net impact of the Act by measuring the ex-post
Page 7
7
long-term changes in firm value. We find that market-adjusted returns of bribery-sensitive firms increase
up to 18 months after the passage of the Bribery Act. On Average, high bribery-risk firms are expected to
increase their abnormal returns by 12% for 18 months compared to low bribery-risk firms. Our result
suggests that any forgone growth opportunity is offset by the reduction in cost of equity, resulting in the
positive market-adjusted performance. Additionally, we find that the positive abnormal returns are greater
amongst large firms, which is consistent with our prior results that large firms are in a better position to put
a system in place to reduce bribery.
This study contributes to the literature on the impact of bribery on firm value. Zeume (2017) finds
that firms affected by the Bribery Act exhibit negative abnormal returns on the day the draft law was passed
and argues that bribes facilitate business transactions in some countries. Our results are consistent with that:
bribery-sensitive firms exhibit a decline in expected growth rate under the Bribery Act. However, our results
also show that these firms experience a reduction in implied cost of equity, which improves their valuation.
In particular, the effect is driven by improved corporate transparency through the reduction in information
asymmetry between firm managers and external investors, and the strengthened anti-bribery management
systems. On balance the net effect of the reduction in both cost of equity and growth rate result in better
market-adjusted performance for bribery-sensitive firms. Our results are consistent with avoiding bribery
enhance firm value in the long run.
Further, our results can help guide policymakers in jurisdictions that have not yet implemented anti-
bribery laws. While the negative effect of anti-bribery laws on firm growth may discourage lawmakers
from enacting such legislation, our results suggest that discouraging firms from corrupt activities also
decreases their cost of equity. More importantly, the significant increase in long-term performance amongst
firms that are affected by the Act suggests that the effect on the cost of equity dominates the effect on
growth. Our results also indicate that firms with more resources to reorganize their businesses away from
high bribery-risk regions benefit the most from the Bribery Act. Of course the value impact of anti-bribery
legislation is not only justified or refuted by an economic analysis. Above all bribery damages the eventual
Page 8
8
customers, and as most bribery appears to affect activities in relatively poor economies, its impact can be
severe.
The remainder of this paper is organized as follows. Section 2 introduces the theoretical background
of the impact of bribery risk on firm growth and cost of equity and establishes hypotheses. Section 3
describes empirical methodology and sample section. Section 4 provides the main results of DiD
regressions and examines the channels through which bribery risk works. It is followed by further analyses
of the effects of bribery risk on bribery-sensitive firms’ bid-ask spread, anti-bribery management, and
cumulative abnormal returns. Next, robustness tests are conducted to validate our findings, investigating
the timing of the effects of the Act, using placebo tests with artificial event years and a propensity-score-
matched (PSM) sample, and sensitivity analyses of bribery risk indicator. Section 5 concludes.
2. Theoretical Framework and Hypothesis Development
2.1. Institutional Background
Before the Bribery Act came into force, UK bribery laws were reformed several times. The reform
can be traced back to the Public Bodies Corrupt Act 1889 that confined bribery to the public sector and
criminalized the soliciting or the receiving of a bribe by a public officer. The law was reformed by the
Prevention of Corruption Act 1906 which expanded bribery to the private sector. This was further revised
by the Prevention of Corruption Act 1916 to lighten the burden of proof of corruption. The U.K. then
adopted the OECD Convention on Combating Bribery of Foreign Public Officials in International Business
Transactions into U.K. law under Part 12 (ss 108-110). Bribery committed outside the U.K. by public
officials was first dealt with the Anti-Terrorism Crime and Security Act 2001.
Despite the history of reforms of U.K. bribery laws, there were severe pressures from within and
outside the U.K. on tightening of the anti-bribery regime. Within the U.K., there were concerns raised by
Page 9
9
the Nolan report (1995)7 on public officials who were taking up positions in firms with which they had had
official dealings. In addition, OECD sanctioned the U.K. government in June 2008 as the government had
not resolved the complexity and uncertainty among the different anti-bribery laws and did not bring a single
foreign bribery case to court. As a result, the U.K. government reformed its bribery laws and introduced a
more stringent and consolidated anti-bribery law, the Bribery Act 2010. The Act was passed on 25 March
2009, received Royal Assent on 8 April 2010 and came into forced on 1 July 2011.
The Bribery Act imposes potentially unlimited fines and jail terms up to ten years for bribing another
person or a foreign public official, taking bribes, and the failure of a commercial organization to prevent
bribery. It applies to individuals or companies which use bribes in the U.K. or elsewhere. Both domestic
and foreign firms with U.K. operations employing U.K. citizens or providing services to any U.K.
organization are subject to the Act. Unlike the previous U.K. anti-bribery laws and the U.S. Foreign Corrupt
Practices Act (FCPA), the Act covers facilitation payments which have the aim of inducing performance
of routine government tasks that are already obligated to be performed (Trautman & Altenbaumer, 2013).
It also has a wider scope of application including all forms of bribes, not only foreign public officials but
also the private sector. The Act also imposes substantial fines for the corporate offense of failing to prevent
bribery.
Under the new legal regime, the regulated businesses are required to establish effective anti-bribery
systems and controls such as (1) adequate procedures; (2) top (board) level commitment; (3) risk assessment;
(4) due diligence; (5) communication and training; and (6) monitoring and review. The application of risk-
based diligence is extended to the regulated firms’ contractual counterparties like contractors and suppliers.
Thus, the Bribery Act changes the basis for corporate criminal liability from focusing on personnel
7 At the request of the Prime Minister, the Committee on Standards in Public Life published the first report on ethical
principles of public life, known as the Nolan principles, in 1995.
Page 10
10
misconduct within the firm to focusing on the quality of the system governing the company’s activities
(Mukwiri, 2015).
The Bribery Act has several properties as an exogenous shock. First of all, there was no residual
uncertainty about the passage of anti-bribery regulation because of the OECD sanctions. At that time, the
U.K. government was under pressure from the OECD’s anti-bribery working group to reform its outdated
anti-corruption laws and to strengthen the legal enforcement of foreign bribery cases. Second, the Act has
had a strong impact on firms. It imposes unlimited penalties for the use of bribes, much higher than the
fines specified in previous U.K. anti-bribery laws, in the U.S. FCPA or by the OECD Anti-Bribery
Convention. The strength of the Act as a shock is also supported by the cases in the U.K where it has been
enforced. Based on hand-collected data on criminal, administrative and civil foreign bribery cases and other
offences related to foreign bribery on from the OECD annual data on enforcement of the Anti-Bribery
Convention8, the number of cases brought under the Act increased from 0 in 2008 to 37 in 2016. The level
of foreign bribery enforcement in the U.K. has been changed from moderate to active since 2009. In sum,
the Bribery Act was a severe, unexpected and external shock to firms operating in bribery-sensitive markets.
2.2. Bribery Risk and Firm Growth
In this paper, we examine the impact of bribery risk on firm-level long-term growth and implied cost of
equity following the passage of Bribery Act. The Act will cause changes in business opportunities and firm
competitiveness, especially for bribery-sensitive firms. First, the Act may negatively affect the regulated
firms’ growth rate due to missing contracts with positive net present value (NPV) by not bribing. According
to Cheung et al. (2012), a net value of using bribes is positive. The authors hand collected a sample of
bribery cases from 20 stocks markets during 1971-2007 and measured the change of bribe-paying firms in
market capitalization on the initial date of award of the contract. Even after subtracting the amount of bribe,
8 The enforcement data of the Anti-Bribery Convention has been gathered from the Parties to the OECD Anti-Bribery
Convention Decisions on Foreign Bribery Cases from 1999 to 2016. The OECD Working Group on Bribery (WGB)
annual updates the number of criminal, administrative, and civil cases of foreign bribery.
Page 11
11
the firms’ market value is found to increase by 11 dollars for each dollar of a bribe payment. Consistent
with this evidence, a case study of Karpoff et al. (2015) also shows the positive NPV of bribes. Using 143
enforcement cases against foreign bribery under the Foreign Corrupt Practices Act (FCPA), they report that
the benefits of bribe-paying contracts offset the direct and indirect (reputational) costs of bribery except for
cases of corporate fraud such as financial misrepresentation. Therefore, unless bribery-sensitive companies
find an effective way of competing with bribe-paying firms, they are likely to lose projects with positive
NPV and lose competitiveness in high-corruption countries.
Another major factor is a decrease in bribery-sensitive firms’ competitiveness because of the
additional costs of reorganizing their businesses. Hines (1995) investigates the effect of the FCPA on the
operations of U.S. firms in countries perceived as corrupt. They find an unusual decline of U.S. companies
in foreign direct investment (FDI), capital/labor ratios, joint venture activity, and exports due to weakening
competitive positions compared to their counterparts in foreign business transactions. It is also supported
by Zeume (2017)’s evidence that U.K. firms regulated by the Bribery Act reduce the expansion of
subsidiary networks and experience sales decreases in countries perceived as corrupt. Moreover, according
to Iriyama, Kishore, and Talukdar (2016), companies strategically choose between bribery and anti-bribery
practices such as HR-training based on the type of rivals. Thus, the regulated U.K. firms may face a
competitive disadvantage compared to their rivals after anti-bribery regulation.
In sum, the decline in firm competitiveness of bribery-sensitive companies due to the loss of
profitable business and the additional cost of reorganizing their business to comply with the Act suggests
that the passage of the Bribery Act will slow down the bribery-sensitive companies’ growth.
H1: The long-term growth rate of bribery-sensitive firms will decline after the passage of the
U.K. Bribery Act.
2.2. Bribery Risk and Cost of Equity
Page 12
12
Prior literature shows mixed evidence on the effects of bribery on firm risk. Thus, we hypothesize two
different relationships between bribery and the implied cost of equity. First, despite negative market
reaction to anti-bribery legislation, investors recognize the adverse effects of bribery on firm risk. This
includes potential prosecution from higher exposure to anti-bribery laws, loss of business, or a negative
impact on reputation following misconduct. A study on how the engagement in misconduct influences firm
profitability and risk conducted by Murphy et al. (2009) supports the negative impact of misconduct,
evidenced by an increase in risk measured as the standard deviation of return. Moreover, Serafeim (2014)
shows the costs of bribery regarding stakeholder relations after bribery detection. Using an analysis of a
cross-national survey data by PwC in 2009 and 2011, the author finds a significant negative effect of bribery
on employee morale plus damaged business relations, reputation and, through worsening stakeholder
relations, an increase in risk.
Due to the negative effects of bribery on firm risk, investors might expect that the severeness of the
fines for violating the Bribery Act induces the regulated firms’ commitment to avoid bribery related risk.
The Act imposes substantial and open-ended fines on the use of bribes inside and outside the U.K. as well
as for not implementing internal anti-bribery controls. Moreover, the fines under the Act are much higher
than the fines under previous legislation in the U.K. by the OECD Anti-Bribery Convention or by
comparable legislation in the U.S. Therefore, bribery-sensitive firms will become more selective in the
bidding for contracts to avoid being caught up in legal action.
Furthermore, the Bribery Act could serve as an external monitoring device that makes accepting
bribes costly to the managers, and aligns their incentives with shareholders’ interests. Consistent with this
view, Desai, Dyck, and Zingales (2007) find that increased tax enforcement enhances the value of Russian
oil firms by reducing agency problem. The regulated firms’ adoption of anti-bribery policies and internal
controls under the Act also lead to gaining investors’ trust in the stable business environment without
engaging in misconduct or being caught up in legal action. This analysis suggests that the Bribery Act may
result in a lower implied cost of equity of bribery-sensitive firms.
Page 13
13
H2a: The implied cost of equity of bribery-sensitive firms will decline after the passage of the
U.K. Bribery Act.
However, a bribery-sensitive company may continue using bribes to increase the likelihood of
winning contracts even under the new regime, due to the cost of building up new networks and social capital
in order to compete effectively without bribery. According to De Jong, Tu, and Van Ees (2012), especially
in emerging economies, bribery allows entrepreneurs to easily enter new markets by developing a network
of informal relationships with public officials. Although there is a non-linear relationship between bribery
and revenues, bring entrepreneurs are said to enjoy private gains in transition economies such as Vietnam.
This logic is supported by pervasive bribery cases of multinational corporations occurring in emerging
economies such as Wal-Mart Store Inc.’s long-lasting bribery in Mexico. Wal-Mart de Mexico, the brightest
success story of Wal-Mart pitched as a model for future growth to investors, engaged in $24 million bribery
from 2005 to 2011. The primary motive was to speed up approval of government officials to open new
stores and to win local market dominance. As seen in many other bribery cases, establishing networks and
social capital for easy entry into new markets through bribery is an effective route to growth. Firms may
resist abandoning already-built networks based on bribery and invest in an alternative strategy.
Consequently, the Bribery Act cannot easily change bribery-sensitive firms’ behaviour. This suggests that
investors may require a premium for investing in bribery-sensitive firms due to potential risks of engaging
in bribery, resulting in a higher cost of equity. Therefore, it is hypothesized as follows:
H2b: The implied cost of equity of bribery-sensitive firms will increase after the passage of the
U.K. Bribery Act.
3. Methodology and Data
3.1. Model Specification
Page 14
14
In the financial economics literature estimating the cost of capital typically employs an implied cost of
capital approach based on the discounted residual income valuation (Claus & Thomas, 2001; Gebhardt, Lee,
& Swaminathan, 2001) or the abnormal earnings growth valuation models (Gode & Mohanram, 2003;
Ohlson & Juettner-Nauroth, 2005). These models assume that the market value of equity accurately reflects
the market’s expectations about discounted future cash flows. They then estimate the implied cost of capital
by equating the present value of the expected future payoffs, such as residual earnings or dividends, to the
current stock price. In this process, analysts’ short- and long-term earnings forecasts are used as proxies for
the markets’ expectation of future earnings. However, analysts’ forecasts are subject to bias and timeliness
problems that might negatively affect the accuracy of the implied cost of capital estimates. According to
Lys and Sohn (1990), analysts’ short-term earnings forecasts contain only 66% of the information that is
reflected in security prices before the forecast-release date. Furthermore, Guay, Kothari, and Shu (2011)
show that analysts’ sluggishness in revising their forecasts following stock price changes is common for
analysts’ earnings forecasts. For these reasons, we use a simple revision of the standard discounted residual
income model to simultaneously estimate the implied cost of equity and the expected growth rate, based on
Easton et al. (2002). We simultaneously estimate both of these rates, avoid using analysts’ forecasts (Easton,
2009), and only need to assume that one year ahead actual earnings are an unbiased estimate of market
expectations.
Following the implied cost of capital literature, we assume that both future discount and growth rates
can be approximated by a constant, and the following test equation can be obtained by simple algebraic
calculation (See Appendix A):
𝐸𝑌𝑖,𝑡 = 𝑟 + 𝑔 𝐵𝑇𝑀𝐷𝑖,𝑡 + ɛ𝑖,𝑡 (1)
where EYit is the forward earnings yield for firm i in year t, which is measured as earnings at t+1 divided
by market value of equity at t. BTMDit is the book to market discount measured by book value of equity
minus market value scaled by market value (Easton et al. 2002). The constant r, the estimated cost of equity,
and the slope coefficient g, the long-run growth estimate, are the main variables of interest.
Page 15
15
To estimate the effect of the Bribery Act on U.K. firms’ long-run growth and cost of equity, a quasi-
experiment is used. As an exogenous shock, the passage of Bribery Act is used to establish a causal link
between firm bribery and long-run growth and cost of equity. Our identification strategy relies on the
assumption that the Bribery Act is exogenous and uncorrelated with other determinants of the forward
earnings yield (EY) and the book to market discount (BTMD). To compare the changes in the estimated
cost of equity and long-run growth before and after the legislation, we add a dummy variable, Passage that
takes the value of one for the post-legislation period and zero otherwise:
𝐸𝑌𝑖,𝑡 = 𝑟 + 𝑟′ 𝑃𝑎𝑠𝑠𝑎𝑔𝑒 + 𝑔 𝐵𝑇𝑀𝐷𝑖,𝑡 + 𝑔′ 𝐵𝑇𝑀𝐷𝑖,𝑡 𝑃𝑎𝑠𝑠𝑎𝑔𝑒 + 𝛾𝑋𝑖,𝑡 + ɛ𝑖,𝑡 (2)
where r’ and g’ represent the difference in implied cost of equity and expected growth for the post-
legislation period compared to the pre-legislation period. We clarify 2003-2008 as the pre-legislation period
and 2010-2015 as the post-legislation period as the draft of the U.K. Bribery Act was passed on March 25,
2009. 2009 is excluded from our sample period as this is the year in which the Act was passed and its
impact on performance in that year is unclear.
To compare changes in the cost of equity and long-run growth for treatment firms with high bribery
risk to those for control firms with low bribery risk, a Difference-in-Differences (DiD) specification is used
as follows:
𝐸𝑌𝑖,𝑡 = 𝑟 + 𝑟′𝑃𝑎𝑠𝑠𝑎𝑔𝑒 + 𝛽1𝐵𝑟𝑖𝑏𝑒𝑟𝑦𝑅𝑖𝑠𝑘𝑖,𝑡 + 𝑟′′𝑃𝑎𝑠𝑠𝑎𝑔𝑒 𝐵𝑟𝑖𝑏𝑒𝑟𝑦𝑅𝑖𝑠𝑘𝑖,𝑡 + 𝑔 𝐵𝑇𝑀𝐷𝑖,𝑡
+ 𝑔′ 𝐵𝑇𝑀𝐷𝑖,𝑡 𝑃𝑎𝑠𝑠𝑎𝑔𝑒 + 𝛽2𝐵𝑇𝑀𝐷𝑖,𝑡𝐵𝑟𝑖𝑏𝑒𝑟𝑦𝑅𝑖𝑠𝑘𝑖,𝑡
+ 𝑔′′𝐵𝑇𝑀𝐷𝑖,𝑡 𝑃𝑎𝑠𝑠𝑎𝑔𝑒 𝐵𝑟𝑖𝑏𝑒𝑟𝑦𝑅𝑖𝑠𝑘𝑖,𝑡 + 𝛾𝑋𝑖,𝑡 + ɛ𝑖,𝑡
(3)
where BriberyRiskit is an indicator that takes the value of one of the test firms with high bribery risk and
zero otherwise. As discussed later we also use a continuous variable (Segment_CPI), which is the
underlying computation on which the zero-one dummy is based. The DiD method controls for omitted
trends that might be correlated with cost of equity and long-run growth in both the treatment and the control
firms. Since the test is conducted around the passage of the Bribery Act which causes a change in firms’
Page 16
16
growth and risk, we can rule out the concern of reverse-causality. In addition, we include firm fixed effects
which control for unobserved time-invariant differences between the treated and the control groups. DiD
methodology traditionally uses a single dummy to distinguish the treatment from the control group both
before and after the event. We assume that controlling for firm specific differences should be more precise
but replicate all our firm fixed effect results with traditional treatment group fixed effect as a sensitivity
test. The major variables of interest are r’’ and g’’ which capture the additional changes in the implied cost
of equity and long-run growth from the pre- to the post-legislation period for firms with high bribery risk
relative to the change for firms with low bribery risk.
3.2. Data and Sample
We initially sample all U.K. firms listed on the U.K. stock markets which consist of 1,506 active (FBRIT)
and 9,118 dead companies (DEADUK) from Datastream as of 30 August 2017. After restricting the sample
to primary equites, listed on the London Stock Exchange (LSE) or the Alternative Investment Market (AIM),
and active in March 2009, we are left with 1,884 firms. Stock return data and accounting data of the
companies for 2003-2016 are obtained from Datastream/Worldscope. The data for 2016 is used for forward
earnings yield, so the period of analysis is from 2003 to 2015. Firms with missing data, negative equity (EY
less than 0 and BTMD less than -1), priced at less than £1, and in the financial sector are excluded. All
continuous variables are truncated the 1st and 99th percentiles to reduce the impact of outliers. Table 1
describes the sample selection process.
(Insert Table 1 here)
3.3. Variables
Measurement of Bribery Risk
We estimate firms’ bribery risk using geographic segments. The high correlation between firms’ foreign
sales and their propensity to use bribes is supported by Karpoff, et al. (2014) and Chung et al. (2012). The
Page 17
17
bribery risk is estimated by weighting the Transparency International (TI)’s Corruption Perceptions Index
(CPI) 2016 for each geographic region by the firm’s sales in those regions. Transparency International
assigns each country a score between zero to 100, with higher scores indicating lower exposure to
corruption. Where a company reports the geographic segment as combined regions or continents, the
average of CPI scores is used. The measure is calculated as follows:
𝑆𝑒𝑔𝑚𝑒𝑛𝑡_𝐶𝑃𝐼𝑖,𝑡 = ∑ (𝑆𝑒𝑔𝑚𝑒𝑛𝑡 𝑆𝑎𝑙𝑒𝑠𝑖,𝑡,𝑠
𝑇𝑜𝑡𝑎𝑙 𝑆𝑎𝑙𝑒𝑠𝑖,𝑡 × 𝐺𝑒𝑜𝑔𝑟𝑎𝑝ℎ𝑖𝑐 𝑆𝑒𝑔𝑚𝑒𝑛𝑡𝑠
′𝑠 𝐶𝑃𝐼)
where 𝑆𝑒𝑔𝑚𝑒𝑛𝑡𝑠′𝑠 𝐶𝑃𝐼 is the CPI score for each geographic segment s of firm i in year t. In this measure,
a higher score means lower bribery risk. Bribery risk is assigned 1 if a firm’s total score (Segment_CPI) is
less than or equal to 55 and 0 otherwise. We use 55 as a reference point for bribery risk by considering the
distribution of Segment_CPI scores. 55 represents a value 1 standard deviation below the mean of
Segment_CPI scores (See Table 2). We examine the sensitivity of our results to different cut-off points.
Based on 55, the proportion of treated (high bribery-risk) firms is 17.06%. The distribution of Segment_CPI
scores and the sensitivity analysis of the bribery risk indicator are described in session 4.6.4.
Control Variables
Following previous studies on the determinants of bribery we use the following control variables: firm size,
concentration of ownership, return on assets, return on equity, earnings growth, working capital ratio and
sales growth. Firm Size (LnMarketCap) is measured by the logarithm value of market capitalization
(Aguilera & Vadera, 2008; Amato & Amato, 2007; Collins et al. 2009; D’Souza & Kaufmann, 2013). Prior
studies expect conflicting relationships between firm size and a likelihood of bribery. Smaller firms are
likely to focus more on survival and growth rather than considering ethical issues (Amato & Amato, 2007):
hence, firm size decreases the tendency of bribery. On the other hand, smaller firms have fewer resources
for developing sophisticated governance structures (Collins et al. 2009; D’Souza & Kaufmann, 2013) or
code of ethics (Aguilera & Vadera, 2008), in which case, firm size increases the likelihood of bribery.
Concentrated Ownership (ConcenOwn) is measured by a percentage of closely-held equity of shareholders
Page 18
18
at least 5 percent of equity ownership within the firm (Ding et al., 2016; Morck et al., 2005; Zhou & Peng,
2012). A higher level of ownership concentration is expected to prevent firms from committing frauds
because it strengthens investors’ control over managerial decisions.
Measures of basic financial performance and business environment are also included: Return on
Asset (ROA) measured by net income divided by total assets, Return on Equity (ROE) calculated by net
income divided by shareholders’ equity (Povel , Singh, & Winton, 2007), Forward Earnings Yield (EY)
measured by earnings per share scaled by stock price (Wang & Winton, 2010), Sales Growth (SalesGrowth)
calculated by a difference between current and past sales scaled by past sales (Zahra, Priem, & Rasheed,
2005) and Working Capital Ratio (WCR) measured a ratio of current assets to current liabilities (Martin,
Cullen, Johnson, & Parboteeah, 2007). The tendency to commit fraud increases in good business conditions
with investor belief in strong markets (Povel et al., 2007; Wang & Winton, 2010). In contrast, the likelihood
to bribe is seen to increase with the level of financial constraints (Martin et al., 2007) and environmental
hostility (Zahra et al., 2005).
Following prior literature on firm risk and cost of equity (Ashbaugh-Skaife, Collins, Kinney, &
Lafond, 2009), we include Leverage measured by total debt divided by total assets, Cash Flow from
Operation (CFO) calculated by cash flow from operations scaled by total assets, Market Return (Return)
measured by a ratio of a difference between dividend-adjusted stock prices at calendar year end of t+1 and
t to adjusted price at t and Book-to-Market (BM), measured by book value of equity scaled by market value
of equity. All the variables are summarized in Appendix B.
4. Empirical Results
4.1. Descriptive Statistics
Table 2 presents descriptive statistics for our main variables. Panel A includes the full sample and reveals
on average earnings yield of 8.1%, and book-to-market discount of -34.7%. The estimation implies
Page 19
19
reasonable average P/E of 12.3 and MTB of 1.5 during 2003-2015. The treatment and control groups are
divided according to Segment_CPI score resulting in the test group which represents 17.06% of the total
sample. In Panel B, we observe that the treated and control firms have differences in several firm
characteristics including forward earning yield (EY) and book-to-market discount (BTMD). Relative to the
control group, the treated firms have higher foreign sales ratio (ForeignSales), lower Segment_CPI, larger
firm size (LnMarketCap and LnTotalAssets), while having similar leverage level(Leverage) and ownership
concentration (ConcenOwn). Panel C includes more firm characteristics used to balance the treated and the
control groups for the PSM sensitivity test. After we use PSM to match the control and treated firms, the
significant differences in covariates are eliminated. The correlation matrix in Table 3 shows negative
correlation between Segment_CPI and firm size (LnMarketCap and LnTotalAssets), foreign sales
(ForeignSales), and working capital ratio (WCR).
(Insert Table 2 here)
(Insert Table 3 here)
4.2. Effects of Bribery Risk on Firm Growth and Cost of Equity: a DiD approach
Table 4 reports the results from the estimation of equations (1) and (2) and from our DiD regression of (3).
In column (1), using firm-fixed effects, we estimate the cost of equity (r) as 9.1% (p<0.01) and the long-
term growth rate (g) as 3.3% (p<0.01). The estimated values are reasonable for the U.K. economy from
2003 to 2015. In column (2), introducing the passage variable of Bribery Act, the passage of the Act is
accompanied by a significant decrease in the cost of equity (r’) by 1.0% (p<0.01) but no significant effect
on long-run growth (g’). In column (3) the bribery risk dummy and interactions are introduced. The
estimated coefficients show that the passage of the Bribery Act is associated with a significant decrease in
the cost of equity (r’’) by 2.7% (p<0.01) and long-term growth (g’’) by 3.8% (p<0.05) for treated firms.
However, the Bribery Risk and BTMD*Bribery results show that these firms with higher bribery risk had
higher estimates of cost of equity (2.3%) and growth (3.5%) than control firms in the period before the Act.
Page 20
20
We are able to include the indicator for bribery risk into the DiD regression (3) together with firm-fixed
effects as it is a time-variant measure based on firms’ annual Segment_CPI.
When we use Segment_CPI, a continuous variable, instead of the dummy variable, BriberyRisk, in
column (4), an increase in Segment_CPI is positive and statistically significant, but no significant effect on
the long-term growth. As higher Segment_CPI score means lower bribery risk, the result is similar to the
finding of the BribeyRisk indicator.
Our results are consistent with the Bribery Act 2010 reducing the cost of equity for bribery-sensitive
firms by a statistically and economically significant amount. This could be explained by firms pulling back
from risky activities and improving their anti-bribery management. We examine both of these explanations
later. Our initial results using a bribery risk dummy are also consistent with a reduction in expected growth
which is again statistically and economically significant. However, when we replace the bribery risk
dummy with the continuous variable the statistical significance disappears. At this stage we are unable to
robustly conclude that Bribery Act was associated with a reduction of expected growth for bribery-sensitive
firms. A decrease in the long-run growth rate would be consistent with the finding of Zeume (2017)’s study,
yet the drop in the cost of equity implies that the anti-bribery legislation also reduced risk for bribery-
sensitive firms.
We conduct two simple sensitivity tests which control for variation in the risk-free rate and in forward
yield. Column (5) includes risk-free rate which uses Bank of England’s ‘Annual Average of UK Official
Bank Rate in the DiD regression (3). After controlling yearly risk-free rate, the coefficients for test variables
(r’’ and g’’) remain the same. Column (6) adds annual average of forward earnings yield (EY) into the DiD
regression (3) to control the effect of yearly trends of forward earnings yield. The estimation results are not
significantly different from the results of DiD regression (3).
(Insert Table 4 here)
4.3. Bribery Risk and Corporate Transparency
Page 21
21
Following up on the apparent reduction in cost of equity we investigate the effect of bribery risk on the
bribery-sensitive firms’ transparency. Jin and Myers (2006) argue that corporate transparency affects firm
risk and returns. The Bribery Act requires the affected companies to establish anti-bribery policies and also
to internalize bribery risk into their risk assessment and due diligence. The explicit intention of the act is
also to reduce illicit activity which, given the enhanced possibility of discovery, implies a reduction in
actual risk. Thus, the anti-bribery law is intended to increase the regulated firms’ transparency and reduce
its risk-taking activities, which should result in the decrease in the firms’ cost of equity.
4.3.1. Information asymmetry
Merton (1989) and Healy and Palepu (2001) argue that if there is an information asymmetry between firms’
managers and outside investors regarding the firms’ future prospects, the investors ask for a premium
bearing information risk, thereby increasing the firms’ cost of external financing. The high bribery-risk
firms’ efforts to comply with the Act by reporting on their anti-bribery activities and risk management will
reduce the information asymmetry between managers and investors regarding the firms’ bribery risk, and
so affect the firms’ cost of equity. To validate the reasoning, we examine the change in information
asymmetry of bribery-sensitive firms after the passing of the Act, which is measured by firms’ bid-ask
spreads. Modifying the model of bid-ask spreads implemented in extant literature (Amiram, Owens, &
Rozenbaum, 2016), we conduct a DiD regression of yearly average of daily bid-ask spreads (LnSpread) on
the firms’ bribery risk and several control variables to absorb its non-information asymmetry components
as follows:
𝐿𝑛𝑆𝑝𝑟𝑒𝑎𝑑𝑖,𝑡 = 𝛽0 + 𝛽1𝑃𝑎𝑠𝑠𝑎𝑔𝑒 + 𝛽2𝐵𝑟𝑖𝑏𝑒𝑟𝑦𝑅𝑖𝑠𝑘𝑖,𝑡 + 𝛽3𝑃𝑎𝑠𝑠𝑎𝑔𝑒 𝐵𝑟𝑖𝑏𝑒𝑟𝑦𝑅𝑖𝑠𝑘𝑖,𝑡
+ 𝛽4𝑀𝑎𝑟𝑘𝑒𝑡𝑃𝑟𝑖𝑐𝑒𝑖,𝑡 + 𝛽5𝐿𝑛𝑀𝑎𝑟𝑘𝑒𝑡𝐶𝑎𝑝𝑖,𝑡 + 𝛽6𝑉𝑜𝑙𝑎𝑡𝑖𝑙𝑖𝑡𝑦𝑖,𝑡 + 𝛽7𝑉𝑜𝑙𝑢𝑚𝑒𝑖,𝑡 + ɛ𝑖,𝑡 (4)
where i and t represents firm and year, respectively. LnSpread is the natural logarithm of firms’ bid-ask
spread, which is measured by the difference between daily ask and daily bid prices scaled by the midpoint
of them. MarketPrice is an adjusted closed stock price at the calendar year end. Volatility is measured by
Page 22
22
the annual standard deviation of daily stock returns. Volume is a daily trading share volume averaged over
the year.
Firms with a larger spread tend to have higher information asymmetry. Thus, we expect that bribery-
sensitive firms are more likely to engage in decreasing their bribery risk, and resultantly reduce the firms’
information asymmetry (lower spread). The DiD regression results of bid-ask spreads for ±5 years are
presented in Table 5. The significant and negative estimated coefficient β3, 29.8% (p<0.01) with firm-fixed
effects in column (1) supports this logic. After including control variables in column (2), the magnitude of
the effect decreases to 13.7% (p=0.10). Furthermore, the result is consistent with the findings of using
Segment_CPI score (Segment_CPI) instead of the bribery risk indicator. When a firm’s Segment_CPI score
decreases by a unit, it will decrease its bid-ask spread around by 1.5% (p<0.01) without control variables
in column (3), and 0.6% with control variables in column (4). The result also remains the same when we
use a different measure of bid-ask spread which scales the difference between ask and bid prices by the
natural logarithm of total assets. These results suggest that bribery-sensitive firms show a significant
decrease in information asymmetry after the passage of the draft law. It is also consistent with our results
which indicate a reduction of cost of equity.
(Insert Table 5 here)
4.3.2. Anti-bribery management
A further measure of transparency is firms’ anti-bribery management. Under the Act’s new regime, bribery-
sensitive companies are more likely to improve anti-bribery policy or internal control systems to alleviate
their bribery risk. To measure the effect of the Bribery Act on the firms’ anti-bribery management, we use
Asset4 assessment data of anti-bribery provisions for 2003-2014.9 The anti-bribery provisions consist of
six indicators: 1) whether the company mentions a public commitment to anti-bribery/corruption at the
9 At the moment of data-gathering, the Asset4 anti-bribery data is available up to 2014.
Page 23
23
senior management or the board level, 2) strengthens anti-bribery/corruption in its code of conduct, 3)
adopts an internal management tool for bribery/corruption such as hotlines or whistle-blowing systems, 4)
has a policy to cope with bribery/corruption in business transactions, 5) communicates bribery-relevant
issues with employees in organizational processes, and 6) conducts employee training on anti-
bribery/corruption. Asset4 records “Yes” or “No” for each indicator so that we assign the value of one to
“Yes” and zero to “No”. All values are summed up from zero to six. Higher score in anti-bribery provisions
means better anti-bribery management.
Table 6 reports the DiD regression results of the Asset4 anti-bribery provisions on the firms’ bribery
risk and several controls based on evidences found in prior literature. The passage of the Bribery Act leads
to a significant increase in anti-bribery management of bribery-sensitive firms both without controls in
column (1) and with controls in columns (2). Specifically, the bribery-sensitive firms improve by 0.5 unit
(p<0.05) of the anti-bribery provisions relative to the control firms after the passing of the Bribery Act. The
magnitude of the effect decreases to 0.4 unit (p<0.05) when including control variables. When we use
Segment_CPI score (Segment_CPI) instead of the bribery-risk indicator, a one-unit decrease of firms’
Segment_CPI score improves the firms’ anti-bribery management by 0.03 unit (p<0.01). A lower
Segment_CPI score means a higher bribery risk so the result is consistent with the finding of using bribery-
risk indicator. The improvement in anti-bribery management and the information environment of high
bribery-risk firms will lead to the decline in implied cost of equity after the enactment of Bribery Act.
(Insert Table 6 here)
Although we are focused on the impact of the Bribery Act on bribery-sensitive firms it is apparent
that our sample spans a period in which anti-bribery management techniques appeared to undergo a
transformation. The coefficient on the Passage variable is positive and statistically significant and
significantly larger than the coefficient on the treatment variable. Our results suggest that bribery-sensitive
firms improved their anti-bribery management more than did the control firms, but all firms experienced a
Page 24
24
substantive improvement. The most obvious explanation would be that the Act stimulated this change, but
without a control group we are unable to prove the general result.
4.4. Cross-Sectional Analyses
We conduct cross-sectional analyses to investigate how firm characteristics affect the role of the Act on the
firms’ cost of equity and growth rate. We use three factors: firm size, leverage, and ownership concentration.
We partition our sample into two subsamples based on the median values of total assets, debt-to-asset ratio,
and concentrated ownership, respectively, and compare the estimated coefficients of r’’ and g’’ between
the two subsamples. Table 7 reports the results of the sub-analyses.
Large firms have more resources for improving governance structures (Collins et al., 2009; D’Souza
& Kaufmann, 2013) and anti-bribery practices (Aguilera & Vadera, 2008) and more easily alter their
business models to achieve less exposure to bribery than do smaller firms. We expect that large bribery-
sensitive firms may utilize corporate resources to actively reduce their bribery risk, resulting in lower cost
of equity. Conversely, small bribery-sensitive companies have difficulty in adjusting their business models
immediately as they need to focus on survival and growth (Amato & Amato, 2007) rather than adopt anti-
bribery management practices. The result supports our reasoning. In columns (1) and (2), the estimated
coefficient of r’’ is significant and negative, -3.6% (p<0.1) only for larger firms, whereas the estimated
coefficient of g’’ is not significant -4.2% (p>0.1) for larger companies, but significant -5.2% (p<0.05) for
smaller companies. This indicates that the impact of anti-bribery law is conditioned by corporate resources:
large firms benefit from the law as investors perceive them less risky, whereas small firms experience
decreasing long-run growth.
Second, firm leverage and ownership concentration are used to investigate the effect of monitoring
pressure from external investors such as debtholders and shareholders. When firms face financial
constraints from a higher leverage, they may tend to be selective, avoiding contracts with high bribery
related risks. Similarly, concentrated ownership is expected to prevent firms from risky behaviour as it
Page 25
25
enhances the monitoring power of investors over managerial decisions (Ding et al., 2016; Morck et al.,
2005; Zhou & Peng, 2012). Therefore, bribery-sensitive firms with higher leverage and a greater ownership
concentration have an incentive to reduce their bribery risk responding to the greater monitoring pressure
from external investors. The results in columns (3)-(6) support this expectation. The estimated coefficients
r’’ and g’’ are significant and negative, -4.3% (p<0.05) and -5.3% (p<0.05), respectively, for highly
leveraged and bribery-sensitive firms, but not significant for bribery-sensitive firms with lower leverage.
The result is consistent for ownership concentration. The estimated coefficients r’’ and g’’ are significant
and negative, -5.3% (p<0.01) and -8.8% (p<0.01), respectively, for bribery-sensitive firms with greater
concentrated ownership, but not significant for firms with more dispersed ownership.
The result remains the same, except the effect of firm size on the expected growth rate (not
statistically significant), when we use Segment_CPI score (Segment_CPI) instead of the bribery-risk
indicator (BriberyRisk). The result infers that the effects of the Bribery Act on firms’ cost of equity and
growth rate are more pronounced for bribery-sensitive companies under greater monitoring pressure from
external investors.
(Insert Table 7 here)
4.5. Bribery Risk and Market-Adjusted Returns
It is difficult to anticipate the ultimate direction of change in firm value after the passing of the Bribery Act
since the law can be expected to reduce both the implied cost of equity and the long-term growth rate of
bribery-sensitive firms. Thus, we investigate this by examining medium-term equity returns. Table 8 reports
the results of cross-sectional regressions of cumulative relative returns to market returns (Ri-Rm) using the
‘Market Return Model’ with different event windows from 6 months in (1) to 36 months in (6) after the
passing of the Bribery Act. To calculate the cumulative relative returns to market returns, we use FTSEALL
as a measure of the market return and firms’ monthly Return Index (RI) on a monthly basis from DataStream.
The regression includes firm characteristics as controls such as natural logarithm of market capitalization
Page 26
26
(Size) and book-to-market ratio (BM) as of 25 March 2009 (event date), and annual returns in percentages
for price momentum (Momentum) from a year prior to the event date.
With controlling for industry-fixed effects, BriberyRisk leads to a significant and positive cumulative
abnormal return with coefficients of 0.1 (p<0.1) for 18 months after the passing of the Bribery Act. The
result indicates that high bribery-risk firms are likely to better off than control firms after the passing of the
Act. It infers that the decline in cost of equity offsets the reduced growth opportunities of bribery-sensitive
firms. The Bribery Act has a positive net effect on the value of bribery-sensitive firms. The results are also
supported by the findings of cumulative abnormal returns using Segment_CPI instead of BriberyRisk. Firms
with higher Segment_CPI (low bribery risk) experience significant and negative cumulative abnormal
returns with coefficient of -0.006 (p<0.05) up to 24 months.
When we obtain cumulative abnormal returns (CARs) using the ‘(Beta-Adjusted) Market Model’
which regresses the stocks’ monthly returns on the monthly market returns (FTSEAll) over three years, the
estimated coefficients of BriberyRisk are positive but not significant. This might be because our sample
includes U.K. firms listed on the LSE or on the AIM, having a wide range of firm size. Therefore, the
estimated beta of small-sized companies can be too unstable to obtain appropriate expected returns.
(Insert Table 8 here)
Figure 1 presents 3-year performance between high bribery-risk and low bribery-risk groups using
monthly return indexes which equally weighted firms’ monthly relative returns to market returns
(FTSEALL) using the ‘Market Return Model’. The starting point is at the passage date the Bribery Act, 25
March 2009. The result indicates that high bribery-risk firms outperform low bribery-risk firms for three
years after the Bribery Act. On average, monthly relative returns are 1.65% for high bribery-risk group and
1.29% for low bribery-risk group.10
10 The reason why two equally weighted indexes outperform the market returns (FTSEALL shares) might be due to
size effect. This is validated by the fact that smb-factor monthly return in the U.K. market was 0.66% on average from
Page 27
27
(Insert Figure 1 here)
We also divide our sample into two subsamples according to the median value of total assets to
investigate whether firm size affects the companies’ stock performance. The result, which is not included
in the tables, shows that larger bribery-sensitive firms have significant and higher returns, with estimated
coefficients of BriberyRisk ranging from 0.1 (p<0.05) for 12 months to 0.2 (p<0.01) for 30 months. On the
other hand, estimated coefficients of smaller bribery-sensitive companies are lower and not significant,
from 0.2 (p>0.1) for 12 months to 0.007 (p>0.1) for 30 months. This finding is aligned with our reasoning
that the effect of anti-bribery law is conditioned by corporate resources. Larger bribery-sensitive firms may
utilize enough resources to reduce their bribery risk and experience significant and positive market returns.
4.6. Sensitivity Analyses
In this section, we conduct a series of sensitivity tests to confirm whether our findings are robust to
unobservable trends, shocks or variables. These results are available in an Appendix C.
4.6.1. Dynamic effects
Another test for any pre-treatment trends is to investigate the timing of the effects of the Bribery Act.
Following Bertrand and Mullainathan (2003), and Ni and Yin (2018), we include a series of year dummies
and bribery risk indicator times year dummy interactions into the DiD regression, then examine the changes
in implied cost of equity and expected long-run growth rate each year. In this way, we decompose the effect
of the Act into different time periods. If the Bribery Act is exogenous and is not driven by pre-treatment
trends, year dummies from the pre-legislation period should not be significantly correlated to the cost of
equity and the long-run growth rate. The extended DiD specification is as follows:
March 2009 to March 2012. The factor data for the U.K. market is obtained from ‘Xfi Center for Finance and
Investment, University of Exeter Business School’. The factor data is available from: http://business-
school.exeter.ac.uk/research/centres/xfi/famafrench/files/.
Page 28
28
𝐸𝑌𝑖,𝑡 = 𝑟 + 𝑦𝑒𝑎𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 + 𝑓𝑖𝑟𝑚 𝑓𝑖𝑥𝑒𝑑 𝑒𝑓𝑓𝑒𝑐𝑡𝑠 + ∑ 𝑟′′𝑠𝐴𝑐𝑡𝑠
12
𝑠=1 𝐵𝑟𝑖𝑏𝑒𝑟𝑦𝑅𝑖𝑠𝑘𝑖,𝑠 + 𝑔𝐵𝑇𝑀𝐷𝑖,𝑡
+ ∑ 𝑔′𝑠𝐴𝑐𝑡𝑠 𝐵𝑇𝑀𝐷𝑖,𝑠
12
𝑠=1 + ∑ 𝑔′′𝑠𝐴𝑐𝑡𝑠 𝐵𝑇𝑀𝐷𝑖,𝑠
12
𝑠=1 𝐵𝑟𝑖𝑏𝑒𝑟𝑦𝑅𝑖𝑠𝑘𝑖,𝑠 + ɛ𝑖,𝑡 (5)
where BriberyRiskit is an indicator variable for the treated group of firm i in year t, ranging from 2003 to
2015 except 2009. The test variables and firm fixed effects are the same as those in Eq. (3). We decompose
the passage of the Bribery Act into separate time periods: Act1 equals 1 if the year is 2003 and 0 otherwise;
Act2 equals 1 if the year is 2004 and 0 otherwise; and so on.
The coefficients on most of the pre-legislation year dummies except 2003 (BTMD*Act1) and 2004
(Act2, BTMD*Act2) are not significant. The result indicates that there is no significant and persistent pre-
treatment trend of changing the cost of equity and the long-run growth before the passing of the Bribery
Act. When we see the estimated coefficients of r’’ and g’’ for the post-legislation period, which measure
the difference between high and low bribery-risk firms, coefficients of r’’ are significant and negative in
2014 (Act1*Bribery) and coefficients of g’’ change from significant and positive for 2010-2011
(BTMD*Act7*Bribery, BTMD*Act8*Bribery) to significant and negative from 2012 (BTMD*Act9*Bribery,
BTMD*Act11*Bribery). The result infers that the Bribery Act has had different magnitude of impact on the
implied cost of equity and the expected growth rate of bribery-sensitive firms over time after the passage
of the Act.
4.6.2. Placebo analyses using artificial event periods
To test whether unobservable shocks that are not related to the Bribery Act drive our results, we conduct
placebo analyses using artificial event periods. In order to exclude any effect of the Bribery Act, different
times during the pre-legislation period are included into the placebo analyses (Atanasov & Black, 2016).
Instead of 2009 (the actual event year), 2005, 2006, and 2007 are selected as pseudo-event years in the DiD
regressions. We find that the estimated coefficients of the implied cost of equity (r’’) and the long-run
growth (g’’) are not significant around all the pseudo-event years. Namely, unobservable shocks in the
Page 29
29
pseudo-event years which are unrelated to the Bribery Act have no significant impact on the cost of equity
and the growth rate between treatment and control firms.
4.6.3. DiD regression result with propensity-score-matched (PSM) sample
To mitigate the possibility that unobserved omitted variables might explain our result, we replicate the DiD
analysis using a balanced sample by the Propensity Score Matching (PSM). The PSM method pairs treated
and control units that have similar observable characteristics. To implement the PSM, we first estimate a
logit regression which models the probability of being a firm with bribery risk. The regression model
includes variables of firm characteristics in the year 2007 as predictors that may potentially affect a firm’s
propensity to bribe, cost of equity and long-run growth. The dataset 2007 before the Bribery Act is used to
minimize the impact of the financial crisis. The most relevant covariates are included as our predictors such
as forward earning yield (EY), a lagged forward earning (lag1EY), book-to-market ratio (BM), firm size
(LnMarketCap), working capital ratio (WCR), market return (Return), and sales growth rate (SalesGrowth).
Our logit regression model has a Likelihood Ratio Chi-Squared of 45.2. After estimating the propensity
score for each firm with the predicted probabilities from the logit regression model, we use the propensity
score to match each treated firm to the control firm. Based on the propensity score, using the nearest-
neighbourhood technique, with replacement to reduce the bias of matching, we can match pairs with a
maximum variation in the causal variable while minimizing the difference in firm characteristics (Dehejia
& Wahba, 2002).
The PSM matched sample includes 994 observations from 137 firms for the pre- and post-legislation
periods. Although the sample is somewhat smaller than that of the primary DiD analysis, the PSM
regression produces similar results to the full sample analysis. These show that there is a significant
decrease in the implied cost of equity (r’’) by 2.4% (p<0.1) and a significant decrease in the expected
growth rate (g’’) by 4.8% (p<0.05) for bribery-sensitive firms, as seen in column (1). It is also supported
by the findings of PSM regression with Segment_CPI score instead of BriberyRisk indicator in column (2).
A unit increase in Segment_CPI score leads to a significant increase in the implied cost of equity by 0.1%
Page 30
30
(p<0.05) and a significant increase in the expected growth rate by 0.1% (p<0.05). In other words, bribery-
sensitive firms having smaller Segment_CPI score experience a significant decrease in the cost of equity
and in the growth rate. It implies that the causal impacts of bribery risk on the implied cost of equity and
the long-term growth rate remain valid after controlling for differences between treatment and control firms.
4.6.4. Sensitivity analysis of bribery risk indicator
To address the sensitivity of our results to the proxy indicator of firm-level bribery risk, we compare the
proportion of treated firms and the changes in the implied cost of equity and the long-term growth, by
varying the Segment_CPI score. Our primary measure for bribery risk is based on whether a firm’s
Segment_CPI is 55 or not. When we observe the effects of Segment_CPI scores on the estimated
coefficients of r’’ and g’’, the estimated coefficients remain significant and similar up to around 55.
However, from about 60, the significance of the estimated coefficients reduces as the proportion of
treatment companies becomes highly increasing. Thus, 55 of the Segment_CPI score is valid to be used as
our primary proxy for the treatment group.
Moreover, when comparing our measure of corporate bribery risk with the FTSE4Good bribery risk
indicator, the proportion of matched companies is 77.3%, or 231 out of 299 firms.11 The matching ratio
seems to support the validity of our measure of bribery risk.
5. Conclusions
We examine the effect of anti-bribery regulation on firm valuation using the regulatory shock of the U.K.
Bribery Act 2010. Since the Act not only imposes substantial fines on the use of bribes but also requires
regulated firms to develop anti-bribery policies and management systems, the Bribery Act significantly
11 We could only count the matching ratio of our measure of bribery risk to the FTSE4Good classification not for the
whole sample period but for 2009, due to the lack of data availability.
Page 31
31
changes firm behaviour and may affect firm valuation. Previous evidence (Zeume, 2017) has suggested that
the Act had a negative impact on growth prospects for bribery-sensitive firms. To measure the impact of
the new regime, we simultaneously estimate the changes in the implied cost of equity and the expected
growth rate following the Act’s implementation. We find that the Bribery Act was associated with a
reduction in the implied cost of equity, but also a reduction in expected growth. The net effect results in an
increase in firm value. Our results are consistent with bribery-sensitive firms passing up positive NPV-
projects to reduce their exposure to bribery, but the reduction in their cost of capital due to reduced risk is
more than compensating.
To examine the mechanism through which the Bribery Act affects firm value, we estimate the effect
of bribery on transparency. After the passing of the Bribery Act, we find that high bribery-risk firms
experience a significant decline in information asymmetry between firm managers and outside investors,
as evidenced by bid-ask spread, and an improvement in anti-bribery management systems. The results
indicate that bribery-sensitive companies are likely to improve corporate transparency, actively engaging
in bribery-management systems and communicating their anti-bribery efforts to the public as evidence of
complying with the Act. The increase in transparency will become a channel for high bribery-risk firms to
reduce their cost of equity. This reasoning is also supported by our findings that the decline in implied cost
of equity is stronger amongst bribery-sensitive firms with greater monitoring pressure from shareholders
and debtholders.
To investigate the net impact in firm value of the Bribery Act between the reduction in expected
growth rate and the decline in implied cost of equity, we measure long-run market-adjusted performance.
We find that bribery-sensitive firms’ market-adjusted returns increase up to 24 months after the passage of
the draft law. The result indicates that the bribery-sensitive companies have been better off on average since
the law was enacted. Overall, the net effect of the Act on firm value is positive.
This study adds the literature on bribery and firm value. Our evidence shows that, following the
Bribery Act 2010, investors re-evaluate bribery-sensitive firms as having less potential for growth, but also
Page 32
32
as being less risky. The drop in the expected growth rate is consistent with Zeume (2017)’s finding that the
Bribery Act reduces the U.K. firms’ market value and competitiveness compared to their unregulated non-
U.K. competitors. In contrast, our result provides a new insight into the effectiveness of the Bribery Act, in
particular on reducing firm risk through the reduction in information asymmetry and the strengthened anti-
bribery management systems.
Page 33
33
References
Aguilera, R.V., & Vadera, A.K. (2008). The dark side of authority: Antecedents, mechanisms, and
outcomes of organizational corruption. Journal of Business Ethics, 77(4), 431-449.
Amato, L.H., & Amato, C.H. (2007). The effects of firm size and industry on corporate giving. Journal of
Business Ethics, 72, 229-241.
Amiram, D., Owens, E., & Rozenbaum, O. (2016). Do information release increase or decrease information
asymmetry? New evidence from analyst forecast announcements. Journal of Accounting and
Economics, 62, 121-138.
Ashbaugh-Skaife, H., Collins, D.W., Kinney, W.R., & Lafond, R. (2009). The effect of SOX internal
control deficiencies on firm risk and cost of equity. Journal of Accounting Research, 47(1), 1-43.
Atanasov, V., & Black, B. (2016). Shock-based causal inference in corporate finance and accounting
research. Critical Finance Review, 5, 207-304.
Bertrand, M., & Mullainathan, S. (2003). Enjoying the quiet life? Corporate governance and managerial
preferences. Journal of Political Economy, 111, 1043-1075.
Birhanu, A.G., Gambardella, A., & Valentini, G. (2016). Bribery and investment, Strategic Management
Journal, 37, 1865-1877.
Cheung, Y.L., Rau, P.R., & Stouraitis, A. (2012). How much do firms pay as bribes and what benefits do
they get? NBER working paper.
Claus, J., & Thomas, J. (2001). Equity premia as low as three percent? Evidence from analysts’ earnings
forecasts for domestic and international stock markets. Journal of Finance, 56, 1629-1666.
Collins, J.D., Uhlenbruck, K., & Rodriguez, P. (2009). Why firms engage in corruption: A top management
perspective. Journal of Business Ethics, 87(1), 89-108.
Dehejia, R.H., & Wahba, S. (2002). Propensity score-matching methods for nonexperimental causal studies.
Review of Economics and statistics, 84(1), 151-161.
De Jong, G., Tu, P.A., & Ees, H.V. (2012). Which entrepreneurs bribe and what do they get from it?
Exploratory evidence from Vietnam. Entrepreneurship Theory and Practice, 36(2), 323-345.
Desai, M.A., Dyck, A., & Zingales, L. (2007). Theft and taxes. Journal of Financial Economics, 84, 591-
623.
Page 34
34
Ding, S., Qu, B., & Wu, Z. (2016). Family control, socioemotional wealth, and governance environment:
The case of bribes. Journal of Business Ethics, 136, 639-654.
Dow Jones [http://www.dowjones.com]. (2011). The State of Anti-Corruption Compliance Survey [Cited
2011 February 9]. Available from:
http://www.corporatecompliance.org/Portals/1/PDF/Resources/past_handouts/CEI/2011/407.pdf.
D’Souza, A., & Kaufmann, D. (2013). Who bribes in public contracting and why: worldwide evidence from
firms. Economics of Governance, 14, 333-367.
Easton, P., Taylor, G., Shroff, P., & Sougiannis, T. (2002). Using forecasts of earnings to simultaneously
estimate growth and the rate of return on equity investment. Journal of Accounting Research, 40(3),
657-676.
Easton, P. (2009). Estimating the cost of capital implied by market prices and accounting data. Foundations
and Trends in Accounting, 2(4), 241-364.
Gebhardt, W., Lee, C., & Swaminathan, B. (2001). Toward an implied cost of capital. Journal of Accounting
Research, 39, 135-176.
Gode, D., & Mohanram, P. (2003). Inferring the cost of capital estimates using the Ohlson-Juettner model.
Review of Accounting Studies, 8, 399-431.
Guay, W., Kothari, S.P., & Shu, S. (2011). Properties of implied cost of capital using analysts’ forecasts.
Australian Journal of Management, 36(2), 125-149.
Healy, P.M., & Palepu, K.G. (2001). Information asymmetry, corporate disclosure, and the capital markets:
A review of the empirical disclosure literature. Journal of Accounting and Economics, 31, 405-440.
Hines, J.R. (1995). Forbidden payment: Foreign bribery and American business after 1977. NBER
Working Paper No. 5266.
Iriyama, A., Kishore, R., & Talukdar, D. (2016). Playing dirty or building capability? Corruption and HR
training as competitive actions to threats from informal and foreign firm rivals. Strategic
Management Journal, 37, 2152-2173.
Jin, Li., & Myers, S.C. (2006). R2 around the world: New theory and new tests. Journal of Financial
Economics, 79(2), 257-292.
Karpoff, J.M., Lee, D.S., & Martin, G.S. (2014). The economics of foreign bribery: Evidence from FCPA
enforcement actions. http://www.fmaconferences.org/Tokyo/Papers/JonathanKarpoff.pdf.
Page 35
35
Karpoff, J.M., Lee, D.S., & Martin, G.S. (2015). The value of foreign bribery to bribe paying firms.
Working paper, University of Washington School of Business.
Lys, T., & Sohn, S. (1990). The association between revisions of financial analysts’ earnings forecasts and
security price changes. Journal of Accounting and Economics, 13, 341-363.
Martin, K.D., Cullen, J.B., Johnson, J.L., & Parboteeah, K.P. (2007). Deciding to bribe: A cross-level
analysis of firm and home country influences on bribery activity. Academy of Management Journal,
50(6), 1401-1422.
Merton, R.C. (1989). A simple model of capital market equilibrium with incomplete information. The
Journal of Finance, 42, 483-510.
Morck, R.K., Percy, M., Rian, G.Y., & Yeung, B. (2005). The rise and fall of the widely held firm: The
history of corporate ownership in Canada, in: Morck, R. (eds) The History of Corporate
Governance around the World: Family Business Groups to Professional Manager, pp. 65-140.
University of Chicago Press, Chicago.
Mukwiri, J. (2015). British law on corporate bribery. Journal of Financial Crime, 22(1), 16-27.
Murphy, D.L., Shrieves, R.E., & Tibbs, S.L. (2009). Understanding the penalties associated with corporate
misconduct: An empirical examination of earnings and risk. Journal of Financial and Quantitative
Analysis, 44(1), 55-83.
Ni, X., & Yin, S. (2018). Shareholder litigation rights and the cost of debt: Evidence from derivative
lawsuits. Journal of Corporate Finance, 48, 169-185.
Ohlson, J., & Juettner-Nauroth, B. (2005). Expected EPS and EPS growth as determinants of value. Review
of Accounting Studies, 10, 349-365.
Povel, P., Singh, R., & Winton, A. (2007). Booms, busts, and fraud. Review of Financial Studies, 20, 1219-
1254.
Rose-Ackermon, S. (1975). The economics of corruption. Journal of Public Economics, 4(2), 187-203.
Serafeim, G. (2014). Firm competitiveness and detection of bribery. Harvard Business School Working
Paper, No. 14-012.
Transparency International [http://www.transparency.org]. (2017). Corruption Perception Index 2016
[Cited 2017 January 25]. Available from:
https://www.transparency.org/whatwedo/publication/corruption_perceptions_index_2016.
Page 36
36
Trautman, L.J., & Altenbaumer-Price, K. (2013). Lawyers, guns, and money: the bribery problem and the
U.K. Bribery Act. The International Lawyer, 47(3), 481-517.
Wang, T.Y., & Winton, A. (2010). Corporate fraud and business conditions. The Journal of Finance, 65(6),
2255-2292.
Zahra, S.A., Priem, R.L., & Rasheed, A.A. (2005). The antecedents and consequences of top management
fraud. Journal of Management, 31(6), 803-828.
Zeume, S. (2017). Bribes and firm value. Review of Financial Studies, 30(5), 1457-1489.
Zhou, J.Q., & Peng, M.W. (2012). Does bribery help or hurt firm growth around the world? Asia Pacific
Journal of Management, 29, 907-921.
Page 37
37
Table 1. Sample Selection Process
Number of Firms
Entire UK stocks listed on all UK markets: FBRIT (Domestic research stocks) and
DEADUK from Datastream as of 30 August 2017
1,506 (active) and
9,118 (dead)
Less cases with no historical price data around March 2009, of non-primary equity,
and not listing on the LSE or the AIM
(9,496)
=Total number of initial sample 1,884
Less cases with market price less than £1 (358)
Less financial firms (ICB Industry=8) (279)
Less outliers on EY, BTMD, Market Price variables by truncating the 1st and 99th
percentiles
(15)
Less cases with negative equity (EY<0 and BM<0) (263)
Less missing variables of cross-sectional analyses (TotalAssets, Leverage, and
ConcenOwn)
(35)
=Total number of final sample 934
Page 38
38
Table 2. Descriptive Statistics
Panel A-1. Descriptive statistics of the whole sample
Variable Observations Mean
Std.
Dev. Min Q1 Median Q3 Max
EY 6,020 0.081 0.061 0.000 0.043 0.068 0.101 0.438
BTMD 6,020 -0.347 0.576 -0.999 -0.710 -0.512 -0.187 3.606
Segment_CPI 6,020 68.502 13.230 9.608 60.656 72.696 80.075 87.467
LnMarketCap 6,020 11.931 2.177 6.045 10.207 11.783 13.541 18.687
LnTotalAssets 6,020 12.019 0.517 6.795 10.432 11.768 13.504 19.054
Leverage 6,020 0.165 0.154 0.000 0.019 0.140 0.261 0.855
ConcenOwn 6,020 30.440 24.564 0.000 8.860 26.470 48.545 99.550
ForeignSales 6,020 40.427 35.929 0.000 2.095 35.310 73.550 100.000
Panel A reports summary statistics for variables used in the main analyses. EY and BTMD are truncated at 1% and 99%.
Other continuous variables are winsorized at 1% and 99%.
Panel A-2. Observations by year of the whole sample
Year 2003 2004 2005 2006 2007 2008 2009
Obs. 342 396 468 513 527 478 517
Year 2010 2011 2012 2013 2014 2015 Total
Obs. 525 508 489 433 430 394 6020
Panel B. Mean variables by bribery risk (within the whole sample)
Treated Group
(n=1,027)
Control Group
(n=4,993) Difference t-statistics
EY 0.077 0.082 0.005** 2.453
BTMD -0.398 -0.337 0.061*** 3.114
Segment_CPI 44.826 73.372 28.546*** 107.81
LnMarketCap 12.572 11.799 -0.773*** -10.019
LnTotalAssets 12.457 11.929 -0.528*** -7.11
Leverage 0.161 0.166 0.006 1.066
ConcenOwn 31.162 30.292 -0.87 -1.033
ForeignSales 70.631 34.214 -36.418*** -31.998
In comparison between treated (high bribery-risk) and control (low bribery-risk) groups, firms are assigned
to the treatment group if the firms’ Segment_CPI is lower than 55 and the control group otherwise. ***, **,
and * denote statistical significance at the 1, 5, and 10% levels respectively.
Page 39
39
Panel C. Comparison between treated and control groups before/after balancing (Year=2007)
Mean variables of unbalanced sample
Treated Group
(n=88)
Control Group
(n=439) Difference
Normalized
Difference
EY 0.072 0.074 -0.002 0.038
BTMD -0.611 -0.502 -0.109*** 0.336
Segment_CPI 46.313 73.736 -27.423*** 4.031
LnMarketCap 12.149 11.825 0.324 0.146
LnTotalAssets 11.717 11.752 -0.035 0.016
Leverage 0.155 0.179 -0.025 0.160
ConcenOwn 32.212 31.059 1.153 0.047
ROA 0.058 0.062 -0.004 0.030
ROE -0.207 0.181 -0.388** 0.183
SalesGrowth 0.562 0.271 0.291** 0.198
WCR 2.861 1.558 1.303*** 0.406
CFO 0.066 0.095 -0.030 0.222
Return 0.154 0.016 0.138 0.183
BM 0.389 0.498 -0.109 0.335
Mean variables of balanced (propensity score matching) sample
Treated Group
(n=79)
Control Group
(n=58) Difference
Normalized
Difference
EY 0.072 0.076 -0.004 0.073
BTMD -0.608 -0.607 -0.001 0.003
Segment_CPI 46.537 72.400 -25.863 3.671
LnMarketCap 12.280 12.363 -0.083 0.036
LnTotalAssets 11.895 12.086 -0.191 0.082
Leverage 0.162 0.164 -0.002 0.013
ConcenOwn 29.669 31.447 -1.778 0.071
ROA 0.056 0.069 -0.013 0.086
ROE -0.245 0.126 -0.371 0.176
SalesGrowth 0.547 0.580 -0.033 0.016
WCR 2.643 1.866 0.776 0.237
CFO 0.073 0.099 -0.026 0.197
Return 0.129 0.111 0.018 0.046
BM 0.392 0.393 -0.001 0.003
Panel C reports summary statistics for outcome and control variables between treated and control firms before
and after the PSM-balancing. Normalized Difference is the absolute difference between means of treated and
control groups scaled by the squared root of midpoint between standard deviations squared of treated and
standard deviations squared of control groups. If the normalized difference lies between 0.2 and 0.3, the sample
is regarded as well-balanced. EY and BTMD are truncated at 1% and 99%. Continuous covariates are winsorized
at the 1st and 99th percentiles in order to mitigate the effects of outliers. ***, **, and * denote statistical
significance at the 1, 5, and 10% levels respectively.
Page 40
40
Table 3. Correlations
1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 EY 1.00
2 BTMD 0.30 1.00
3 Segment_CPI 0.05 0.09 1.00
4 LnMarketCap -0.21 -0.41 -0.21 1.00
5 LnTotalAssets -0.09 -0.16 -0.16 0.94 1.00
6 Leverage 0.01 -0.03 0.03 0.25 0.36 1.00
7 ConcenOwn 0.11 0.17 0.03 -0.51 -0.51 -0.14 1.00
8 ForeignSales -0.05 -0.08 -0.61 0.26 0.23 0.04 -0.15 1.00
9 ROA -0.02 -0.18 -0.01 0.16 0.06 -0.09 -0.06 0.01 1.00
10 ROE -0.02 0.00 0.03 0.04 0.03 -0.01 0.01 -0.02 0.24 1.00
11 SalesGrowth 0.02 0.01 -0.02 0.00 0.00 -0.01 0.01 0.03 -0.01 0.00 1.00
12 WCR -0.04 0.03 -0.19 -0.12 -0.19 -0.30 0.12 0.11 0.03 0.00 0.00 1.00
13 Return 0.13 -0.10 -0.03 -0.02 -0.06 -0.03 0.02 0.01 0.04 0.01 -0.01 0.03 1.00
14 CFO -0.01 -0.22 -0.02 0.19 0.06 -0.10 -0.04 0.02 0.52 0.05 -0.01 -0.04 0.09 1.00
Page 41
41
Table 4. Effects of Bribery Risk on Firm Growth and Cost of Equity
Baseline
Model
Passage
Effect DiD Regressions
DV: EY (1) (2) (3) (4) (5) (6)
Passage*Bribery -0.027*** -0.027*** -0.025** (0.010) (0.010) (0.010)
BTMD*Passage*Bribery -0.038** -0.038** -0.034** (0.016) (0.016) (0.016)
Passage*Segment_CPI 0.074***
(0.028)
BTMD*Passage*Segment_CPI 0.060
(0.045)
Passage -0.010*** -0.006 -0.062*** -0.030*** -0.003 (0.004) (0.004) (0.020) (0.010) (0.004)
BriberyRisk 0.023** 0.022** 0.023** (0.011) (0.011) (0.011)
Segment_CPI -0.088**
(0.034)
BTMD 0.033*** 0.037*** 0.032*** 0.097*** 0.031*** 0.026*** (0.004) (0.006) (0.006) (0.032) (0.006) (0.006)
BTMD*Passage -0.003 0.002 -0.046 0.003 0.003 (0.006) (0.007) (0.032) (0.007) (0.006)
BTMD*Bribery 0.035** 0.035** 0.034** (0.015) (0.016) (0.015)
BTMD*Segment_CPI -0.086*
(0.044)
RiskFreeRate -0.006***
(0.002)
EY_average 0.662*** (0.088)
Constant 0.091*** 0.096*** 0.093*** 0.158*** 0.120*** 0.038*** (0.001) (0.003) (0.003) (0.024) (0.010) (0.009)
Firm FE Y Y Y Y Y Y
Observations 5,503 5,503 5,503 5,503 5,503 5,503
R-squared 0.061 0.068 0.074 0.074 0.075 0.088
Number of Firms 934 934 934 934 934 934
This table reports our main results of the effects of bribery risk on firms’ cost of equity and long-run growth for 2003-
2015 except 2009. The dependent variable is the forward earnings yield (EY) measured as earnings at t+1 divided by
market value of equity at t. Model (1) estimates the implied cost of equity and the long-run growth using all firms with
firm-fixed effects. Model (2) measures the changes in the cost of equity and growth rate after the Bribery Act. Model (3)
presents a regression for estimating the additional changes in the interested variables according to firms’ bribery risk with
firm -fixed effects. Model (4) uses the Segment_CPI score instead of the bribery risk indicator. Model (5) includes risk-
free rate which uses Bank of England’s ‘Annual Average of UK Official Bank Rate to control annual time trends. Model
(6) adds each year’s average value of EY. For convenience, Segment_CPI is scaled by 100. Robust standard errors clustered
at the firm level are presented in parentheses. ***, **, and * denote statistical significance at the 1, 5, and 10% levels
respectively.
Page 42
42
Table 5. Bid-Ask Spread Effects
DV=ln(Spread)
VARIABLES (1) (2) (3) (4)
Passage*Bribery -0.298*** -0.137
(0.100) (0.083) Passage*Segment_CPI 1.474*** 0.645***
(0.282) (0.234)
Passage -0.564*** -0.429*** -1.622*** -0.899***
(0.038) (0.035) (0.198) (0.165)
BriberyRisk 0.053 -0.033
(0.109) (0.090) Segment_CPI -0.821** -0.468
(0.413) (0.317)
LnPrice -0.107* -0.103
(0.064) (0.064)
LnMarketCap -0.495*** -0.491***
(0.065) (0.065)
Volatility -4.506** -4.301**
(1.763) (1.773)
LnVolume -0.045* -0.048*
(0.025) (0.026)
Constant -3.892*** 2.345*** -3.317*** 2.626***
(0.024) (0.701) (0.286) (0.718)
Firm FE Y Y Y Y
Observations 4,668 4,668 4,668 4,668
R-squared 0.216 0.445 0.226 0.446
Number of Firms 884 884 884 884
This table reports DiD regressions of ‘the natural logarithm of bid-ask spreads’ in Model (1)-(4) and of ‘the natural
logarithm of trading volume’ in Model (5)-(8) on the bribery risk for ±5 years around the passage of the Bribery
Act. The bid-ask spread used in the regression is measured by annual average of the daily closing ask price less the
closing bid price scaled by the midpoint of the closing ask and bid prices available from the Datastream. We use
two different measures of bribery risk, BriberyRisk (1 or 0) and Segment_CPI score. Model (1), (3), (5), and (7) do
not include control variables, whereas Model (2), (4), (6), and (8) include several control variables. All models
include firm-fixed effects. All continuous variables are winsorized. For convenience, Segment_CPI is scaled by
100. Robust standard errors clustered at the firm level are presented in parentheses. ***, **, and * denote statistical
significance at the 1, 5, and 10% levels respectively.
Page 43
43
Table 6. Anti-Bribery Management
DV: Asset4 Anti-Bribery Provisions (1) (2) (3) (4)
Passage*Bribery 0.522** 0.431**
(0.235) (0.217)
Passage*Segment_CPI -2.717*** -2.527*** (0.761) (0.746)
Passage 2.186*** 4.005*** 4.123*** 5.919*** (0.123) (0.343) (0.505) (0.566)
BriberyRisk -0.220 -0.146
(0.245) (0.227)
Segment_CPI 1.398 1.275 (0.878) (0.827)
LnMarketCap 0.152* 0.104 (0.087) (0.086)
Leverage 0.709* 0.587 (0.396) (0.392)
WCR -0.064* -0.061 (0.037) (0.038)
LnConcenOwn -0.014 -0.014 (0.023) (0.022)
BM 0.145* 0.129* (0.078) (0.073)
RiskFreeRate 0.452*** 0.478*** (0.070) (0.069)
Constant 1.025*** -3.285*** 0.0321 -3.599*** (0.069) (1.171) (0.592) (1.286)
Firm FE Y Y Y Y
Observations 2,450 2,292 2,450 2,292
R-squared 0.499 0.513 0.506 0.519
Number of Firms 267 266 267 266
This table presents DiD regressions of the firms’ anti-bribery provision score on the bribery risk around the passage
of the Bribery Act for 2003-2014. The anti-bribery provision score is constructed with six indicators related to anti-
bribery/corruption provisions, which are collected by Asset4. In our sample, the firms having the data for Asset4
anti-bribery provisions are included into the analysis. The indicators are 1) whether the company mentions public
commitment to avoid bribery and corruption at the senior management and the board level, 2) states anti-bribery
and anti-corruption in its code of conduct, 3) has internal management tools over bribery and corruption like whistle
blowing systems, or hotlines, 4) has a policy to withstand bribery and corruption in its business transactions, 5)
communicates relevant issues with employees at the organizational processes, and 6) has relevant employee
trainings. Asset4 records “Yes” or “No” for each indicator so that we assign the value of one to “Yes” and zero to
“No”. All values are aggregated and the total score ranges from zero to six. Higher score means better anti-bribery
management. We use two different measures of bribery risk, BriberyRisk (1 or 0) and Segment_CPI score. Model
(1) & (3) do not include control variables, while Model (2) & (4) do. All continuous variables are winsorized.
Robust standard errors clustered at the firm level are presented in parentheses. For convenience, Segment_CPI is
scaled by 100. ***, **, and * denote statistical significance at the 1, 5, and 10% levels respectively.
Page 44
44
Table 7 Effects of Firm Size, Leverage and Concentrated Ownership
Firm Size Effect
(1): LnTotalAssets>median
(2): LnTotalAssets≤median
Leverage Effect
(3): Leverage>median
(4): Leverage≤median
Concentrated Ownership
(5): ConcenOwn>median
(6): ConcenOwn≤median
DV: EY (1) (2) (3) (4) (5) (6)
Passage*Bribery -0.036* -0.009 -0.043** -0.011 -0.050*** 0.003
(0.020) (0.012) (0.017) (0.013) (0.011) (0.011)
BTMD*Passage*Bribery -0.042 -0.052** -0.053** -0.025 -0.080*** 0.010
(0.030) (0.022) (0.025) (0.021) (0.018) (0.015)
Passage -0.008* -0.004 0.001 -0.016*** -0.004 -0.010*
(0.005) (0.006) (0.007) (0.005) (0.006) (0.005)
BriberyRisk 0.031 0.015 0.035** 0.008 0.047*** -0.019*
(0.020) (0.015) (0.016) (0.015) (0.013) (0.010)
BTMD 0.037*** 0.029*** 0.028*** 0.039*** 0.031*** 0.042***
(0.007) (0.009) (0.008) (0.008) (0.008) (0.006)
BTMD*Passage -0.002 0.004 0.010 -0.010 -0.002 -0.002
(0.007) (0.010) (0.011) (0.009) (0.010) (0.008)
BTMD*Bribery 0.045 0.038** 0.050** 0.020 0.074*** -0.028**
(0.029) (0.019) (0.024) (0.019) (0.014) (0.014)
Constant 0.095*** 0.093*** 0.089*** 0.101*** 0.093*** 0.097***
(0.004) (0.004) (0.005) (0.005) (0.004) (0.004)
Firm FE Y Y Y Y Y Y
Observations 2,758 2,745 2,745 2,758 2,722 2,781
R-squared 0.092 0.064 0.078 0.075 0.088 0.091
Number of Firms 435 631 628 668 701 571
This table reports the results of the impacts of firm size, leverage, industry risk, and concentrated ownership on the
relationship between bribery risk and the test variables, implied cost of equity and long-run growth. Model (1) and (2)
divide the sample according to median value of firm size (LnTotalAssets) which is measured by firms’ natural logarithm
of total assets. Model (3) uses a subsample having firm leverage (Leverage) larger than median, whereas Model (4) uses
a subsample of less than median. Model (5) and (6) use subsamples divided by median of firms’ concentrated ownership
(ConcenOwn). Robust standard errors clustered at the firm level are presented in parentheses. ***, **, and * denote
statistical significance at the 1, 5, and 10% levels respectively.
Page 45
45
Table 8. CAR Analyses
Using the 'Market
Return (Adjusted)
Model'
6 months 12 months 18 months 24 months 30 months 36 months
DV: ∑(Ri-Rm) (1) (2) (3) (4) (5) (6)
BriberyRisk 0.068 0.133** 0.124* 0.118 0.082 -0.136*
(0.055) (0.061) (0.072) (0.085) (0.087) (0.082)
Size 0.022** 0.033** 0.020 0.003 -0.020 0.003
(0.011) (0.013) (0.015) (0.018) (0.018) (0.020)
BM -0.038 -0.142*** -0.208*** -0.256*** -0.270*** -0.271***
(0.032) (0.038) (0.046) (0.051) (0.052) (0.051)
Momentum -0.663*** -0.814*** -0.859*** -1.001*** -0.941*** -0.940***
(0.088) (0.100) (0.117) (0.133) (0.127) (0.140)
Constant -0.377** -0.600*** -0.309 0.179 0.548** 0.252
(0.163) (0.175) (0.222) (0.276) (0.266) (0.297)
Industry FE Y Y Y Y Y Y
Observations 468 507 507 487 487 474
R-squared 0.189 0.218 0.196 0.209 0.183 0.183
This table presents the results of cross-sectional regressions of cumulative relative returns to market returns (Ri-
Rm) using the ‘Market Return Model’ with different event windows from 6 months (1) to 36 months (6) after the
passage of the U.K. Bribery Act. FTSEAll is used for the market return and the monthly returns for companies are
obtained using the monthly Return Index (RI) from the DataStream. The regression uses firm characteristics such
as firm size (Size), book-to-market ratio (BM), and momentum (Momentum) for firm-by-firm analysis instead of
factor analysis. The Size, BM, Momentum are winsorized. Robust standard errors are presented in parentheses. ***,
**, and * denote statistical significance at the 1, 5, and 10% levels respectively.
Using the 'Market
Return (Adjusted)
Model'
6 months 12 months 18 months 24 months 30 months 36 months
DV: ∑(Ri-Rm) (1) (2) (3) (4) (5) (6)
Segment_CPI -0.374** -0.573*** -0.640*** -0.619** -0.489* 0.277
(0.172) (0.184) (0.219) (0.260) (0.256) (0.270)
Size 0.018 0.030** 0.016 -0.002 -0.024 0.001
(0.011) (0.013) (0.015) (0.018) (0.018) (0.020)
BM -0.032 -0.135*** -0.200*** -0.252*** -0.267*** -0.272***
(0.032) (0.038) (0.046) (0.051) (0.052) (0.051)
Momentum -0.649*** -0.809*** -0.852*** -0.991*** -0.933*** -0.942***
(0.088) (0.099) (0.116) (0.132) (0.126) (0.139)
Constant -0.088 -0.196 0.133 0.635* 0.906*** 0.046
(0.218) (0.219) (0.274) (0.346) (0.330) (0.352)
Industry FE Y Y Y Y Y Y
Observations 468 507 507 487 487 474
R-squared 0.196 0.228 0.207 0.216 0.188 0.181
The table presents the results of the same cross-sectional regressions with Segment_CPI score instead of
BriberyRisk indicator, which is used in the first table. For convenience, Segment_CPI is scaled by 100.
Page 46
46
Using the 'Market
Model'
(Beta-Adjusted)
6 months 12 months 18 months 24 months 30 months 36 months
DV: CAR (1) (2) (3) (4) (5) (6)
BriberyRisk 0.045 0.080 0.060 0.112 0.070 -0.231*
(0.063) (0.073) (0.093) (0.112) (0.123) (0.131)
Size -0.021* -0.037** -0.059*** -0.089*** -0.100*** -0.095***
(0.012) (0.016) (0.019) (0.024) (0.026) (0.030)
BM 0.007 -0.113** -0.173*** -0.186*** -0.183** -0.170**
(0.038) (0.049) (0.059) (0.069) (0.072) (0.079)
Momentum -0.752*** -1.048*** -1.345*** -1.552*** -1.735*** -1.664***
(0.100) (0.127) (0.157) (0.180) (0.183) (0.199)
Constant 0.113 0.045 0.334 1.141*** 1.196*** 0.954**
(0.183) (0.223) (0.288) (0.358) (0.367) (0.423)
Industry FE Y Y Y Y Y Y
Observations 416 451 451 434 434 422
R-squared 0.213 0.226 0.244 0.262 0.278 0.248
The third table presents the results of cross-sectional regressions of cumulative abnormal returns (CARs) using the
‘(Beta-Adjusted) Market Model’ which regresses the stocks’ monthly returns on the monthly market returns (FTSEAll)
over three years to obtain the expected returns. The event windows range from 6 months (1) to 36 months (6) after the
Bribery Act. Robust standard errors are presented in parentheses. ***, **, and * denote statistical significance at the
1, 5, and 10% levels respectively.
Figure 1. 3-Year Performance Between Treated and Control Groups Using Equally Weighted
Returns Indexes
This figure presents return indexes between treated (high bribery-risk) and control (low bribery-risk) groups which
equally weighted firms’ monthly relative returns to market returns (FTSEALL). The starting point is at the passage
date of the Bribery Act, 25 March 2009.
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36
Market-Adjusted Monthly Returns of Treated and Control Groups
BriberyRisk=1 (high bribery) BriberyRisk=0 (low bribery)
Page 47
47
Appendix A. Estimation of implied cost of equity and expected growth rate
One of the prevalent methods of estimating cost of capital in the financial economics literature is the
discounted residual income valuation model (Claus & Thomas, 2001; Gebhardt et al., 2001). This
estimation model obtains the implied cost of capital estimate by equating the present value of the expected
future payoffs to the current market value of equity. In this process, analysts’ earnings forecasts are used as
the market’s expectation of future cash flows, suffering from subjective bias and timeliness problems of the
forecasts related to the accuracy of estimating the implied cost of capital. To avoid these issues, we use a
simple revision of the standard discounted residual income model which simultaneously estimates the
implied cost of equity and the expected growth rate (Easton et al., 2002). With this measure, we can also
avoid the need for making invalid assumptions about the expected growth rate.
For the simultaneous estimation of the cost of equity r and the rate of long-term growth g, we can
start from the standard residual income valuation model. It assumes market value of equity as book value
of equity plus present value of future expected residual income following as:
Vt = Bt + ∑(Et+1 - r Bt)
(1 + r)t
∞
t
where Vt is market value of equity in year t; Bt is book value of equity; Et+1 is future earnings; r is future
discount rate. When we assume that varying future discount rates and growth rate of future earnings are
approximated by constant equivalents, r and g, the previous model can be modified as follows:
𝑉𝑡 = 𝐵𝑡 + (𝐸𝑡+1 − 𝑟𝐵𝑡)
(𝑟 − 𝑔)
After some straightforward algebra, we can obtain our estimation model as in the following:
𝐸𝑡+1
𝑉𝑡= 𝑟 + 𝑔 (
𝐵𝑡 − 𝑉𝑡
𝑉𝑡)
Then, we estimate it in a panel regression model as:
Page 48
48
𝐸𝑌𝑖,𝑡 = 𝑟 + 𝑔 𝐵𝑇𝑀𝐷𝑖,𝑡 + ɛ𝑖,𝑡
where 𝐸𝑌𝑖,𝑡 is forward earning yield of i firm at year t, which is measured as earnings at t+1 divided by
market value of equity at t. Instead of analysts’ earnings forecasts, actual forward earnings are used in our
estimation. 𝐵𝑇𝑀𝐷𝑖,𝑡 is book-to-market discount and ɛ𝑖,𝑡 is error term.
Page 49
49
Appendix B. Variable Definitions
Variables Definitions
EY Forward earning yield measured as EPS at t+1, which is obtained from net income
available to common equity (Worldscope code WC01751) scaled by shares outstanding
(WC05301), divided by year-end market price of equity (WC05001) at t
BTMD Book-to-market discount measured from book value of equity minus market value, then
scaled by market value; Alternatively, this value is obtained from book value of equity
(WC03501) divided by market capitalization (WC08001), then minus one
Passage Indicator variable for the passage of the UK Bribery Act, equal 1 if the year is post-
legislation period (after 2009) and 0 otherwise
BriberyRisk Indicator variable for bribery risk, equal 1 for the test firms with high bribery risk and
0 otherwise
Segment_CPI Indicator variable for bribery risk obtained by summing up a firm's sales ratio of each
geographic segment multiplied by TI's CPI (corruption perception index) score for the
geographic region. When the company reports the segment as combined continents, the
average of CPI scores is used. Then, the bribery risk is assigned to 0 for the total score
(Segment_CPI) larger than 55 and 1 otherwise.
(The geographic segment sales are obtained from WC19601, WC19611 to WC19691
and the geographic segment regions are obtained from WC19600, WC19610 to
WC19690.)
FirmSize Natural logarithm of firm's market capitalization (LnMarketCap, WC08001) or
Natural logarithm of firm’s total assets (LnTotalAssets, WC02999)
ConcenOwn Concentrated ownership measured by a percentage of closely-held equity of
shareholders at least 5 percent of equity ownership within the firm (WC08021)
ROA Return on asset measured by net income divided by total assets (WC08326)
ROE Return on equity measured by net income divided by shareholders’ equity (WC08301)
SalesGrowth Sales growth rate measured as a difference between current and past sales (Datastream
code DWSL) scaled by past sales
WCR Working capital ratio measured as a ratio of current asset (WC02201) to current liability
(WC03101)
Leverage Total debt (WC03255) scaled by total assets (WC02999)
Return Market return measured by a ratio of a difference between adjusted stock prices
(DataStream code P#T) at the calendar year end of t and t-1 to adjusted price at t-1
Page 50
50
CFO Cash flow from operation calculated by cash flow from operations (WC04860) scaled
by total assets (WC02999)
BM Book-to-market ratio measured by book value of common equity (WC03501) scaled
by market value of equity (WC08001)
Momentum Ratio of adjusted price (DataStream code P#T) at t to adjusted price at t-1
Page 51
51
Appendix C. Robustness Tests
Table C.1 Effects of Bribery Risk on Firm Growth and Cost of Equity: dynamic effects
DV: EY Extended Regression Model (1)
Constant 0.098***
BTMD 0.049***
(0.006)
(0.008)
Act1*Bribery 0.025 BTMD*Act1 -0.027** BTMD*Act1*Bribery 0.046*
(0.015)
(0.013)
(0.025)
Act2 -0.013* Act2*Bribery 0.026* BTMD*Act2 -0.030** BTMD*Act2*Bribery 0.026
(0.008)
(0.016)
(0.014)
(0.023)
Act3 -0.005 Act3*Bribery -0.010 BTMD*Act3 -0.019 BTMD*Act3*Bribery -0.010
(0.007)
(0.017)
(0.014)
(0.024)
Act4 0.001 Act4*Bribery 7.04e-05 BTMD*Act4 -0.007 BTMD*Act4*Bribery 0.002
(0.008)
(0.016)
(0.014)
(0.023)
Act5 -0.005 Act5*Bribery 0.002 BTMD*Act5 -0.010 BTMD*Act5*Bribery -0.005
(0.008)
(0.024)
(0.013)
(0.033)
Act6 -0.003 Act6*Bribery 0.029** BTMD*Act6 -0.021** BTMD*Act6*Bribery 0.034*
(0.007)
(0.012)
(0.009)
(0.019)
Act7 -0.003 Act7*Bribery 0.010 BTMD*Act7 -0.032*** BTMD*Act7*Bribery 0.029**
(0.007)
(0.010)
(0.009)
(0.014)
Act8 -0.008 Act8*Bribery 0.012 BTMD*Act8 -0.027*** BTMD*Act8*Bribery 0.027*
(0.008)
(0.010)
(0.009)
(0.016)
Act9 -0.009 Act9*Bribery -0.009 BTMD*Act9 -0.014 BTMD*Act9*Bribery -0.023* (0.008)
(0.00881)
(0.009)
(0.014)
Act10 -0.015** Act10*Bribery -0.009 BTMD*Act10 -0.013 BTMD*Act10*Bribery -0.004 (0.007)
(0.010)
(0.009)
(0.013)
Act11 -0.018** Act11*Bribery -0.027** BTMD*Act11 -0.010 BTMD*Act11*Bribery -0.040**
(0.008)
(0.011)
(0.010)
(0.016)
Act12 -0.012 Act12*Bribery -0.005
BTMD*Act12*Bribery -0.011 (0.007)
(0.009)
(0.009)
Firm FE Y
Observations 5,503
Number of
firms
934
R-squared 0.104
This table reports the dynamic effects of bribery risk on the cost of equity and long-run growth from 2003 to 2015 except 2009.
We decompose the passage of the Bribery Act into separate time period. Our main DiD analysis excludes 2009. The dynamic
DiD specification is as follows:
EYit = r + year-dummies + firm-fixed effects + ∑ 𝑟′′𝑠 𝐴𝑐𝑡𝑠12𝑠=1 𝐵𝑟𝑖𝑏𝑒𝑟𝑦𝑅𝑖𝑠𝑘𝑠
+gBTMDit + ∑ 𝑔′𝑠 𝐴𝑐𝑡𝑠 𝐵𝑇𝑀𝐷𝑠12𝑠=1 + ∑ 𝑔′′𝑠𝐵𝑇𝑀𝐷𝑖𝑠𝐴𝑐𝑡𝑠𝐵𝑟𝑖𝑏𝑒𝑟𝑦𝑅𝑖𝑠𝑘𝑖𝑠
12𝑠=1 + ɛit
Act1 equals 1 if the year is 2003 and 0 otherwise; Act2 equals 1 if the year is 2004 and 0 otherwise; and so on. Robust standard
errors clustered at the firm level are presented in parentheses. ***, **, and * denote statistical significance at the 1, 5, and 10%
levels respectively.
Page 52
52
Table C.2 Placebo Test Results
2003-2004 (pre) and
2005-2008 (post-shock)
2003-2005 (pre) and
2006-2008 (post-shock)
2003-2006 (pre) and
2007-2008 (post-shock)
DV: EY (1) (2) (3)
Passage*Bribery -0.0141 -0.0065 0.0007
(0.0131) (0.0139) (0.0140)
BTMD*Passage*Bribery -0.0181 -0.0119 -0.0022
(0.0196) (0.0203) (0.0207)
Passage 0.0003 -0.0003 -0.0058
(0.0055) (0.0062) (0.0061)
BTMD 0.0361*** 0.0386*** 0.0468***
(0.0106) (0.0112) (0.0108)
BTMD*Passage 0.0076 0.00507 -0.0059
(0.0093) (0.0103) (0.0099)
BTMD*Bribery 0.0520** 0.0469** 0.0393*
(0.0205) (0.0213) (0.0204)
Constant 0.1020*** 0.1010*** 0.1040***
(0.0046) (0.0049) (0.0047)
Firm FE Y Y Y
Observations 2,724 2,724 2,724
R-squared 0.116 0.115 0.115
Number of Firms 734 734 734
This table presents the DiD regression results using artificial event periods instead of 2009, to test whether
unobservable shocks that are not related to the U.K. Bribery Act drive our results. 2005, 2006, and 2007 are used
as an artificial event period in Model (1), (2), and (3) respectively. Robust standard errors clustered at the firm
level are presented in parentheses. ***, **, and * denote statistical significance at the 1, 5, and 10% levels
respectively.
Page 53
53
Table C.3 DiD Regression Result with the PSM-Sample
DV: EY (1) (2)
Passage*Bribery -0.024*
(0.013) BTMD*Passage*Bribery -0.048**
(0.020) Passage*Segment_CPI 0.098**
(0.040)
BTMD*Passage*Segment_CPI 0.133**
(0.054)
Passage -0.004 -0.072***
(0.009) (0.026)
BriberyRisk 0.026*
(0.015) Segment_CPI -0.085*
(0.049)
BTMD 0.023 0.093***
(0.014) (0.034)
BTMD*Passage 0.020 -0.079**
(0.015) (0.036)
BTMD*Bribery 0.043**
(0.020) BTMD*Segment_CPI -0.089*
(0.053)
Constant 0.091*** 0.152***
(0.010) (0.030)
Firm FE Y Y
Observations 994 994
R-squared 0.145 0.139
Number of Firms 137 137
This table reports the DiD regression results with the PSM sample. The PSM method pairs the
treated and the control units that are similar in terms of their observable characteristics. We
implement this procedure by using firms’ propensity scores obtained from the logit regression and
matching them with a nearest neighborhood technique with replacement. After the PSM, we
conduct the DiD regression with the balanced PSM-sample. Model (1) uses BriberyRisk dummy
variable whereas Model (2) uses Segment_CPI score as an indicator for bribery risk. For
convenience, Segment_CPI is scaled by 100. Robust standard errors clustered at the firm level are
presented in parentheses. ***, **, and * denote statistical significance at the 1, 5, and 10% levels
respectively.
Page 54
54
Figure C.4 Sensitivity Analyses of Segment_CPI Score on Cost of Equity r’’ and Growth Rate g’’
4.00% 7.81% 11.96% 17.06% 23.65% 43.47%
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
40 45 50 55 60 70
Sensitivity Analysis of Segment_CPI Score on r''
proportion (BriberyRisk=1) r''
95% C.I._low 95% C.I._high
Page 55
55
The first figure presents the median of the Segment_CPI scores of our sample and the distribution of firms’
Segment_CPI scores. The latter two are the results of sensitivity analyses of Segment_CPI score on the estimated cost
of equity (r’’) and the long-term growth rate (g’’) when using different measures for dividing the treatment and the
control firms. The histogram shows the proportion of treatment firms using different values of Segment_CPI ranging
from 40 to 70 for the measure of bribery risk.
-0.14
-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
0.02
0.04
40 45 50 55 60 70
Sensitivity Analysis of Segment_CPI Score on g''
g'' 95% C.I._low 95% C.I._high