Shareholders reaction to corporations’ environmentally
harmful behavior and the moderating effect of CSR.
A cross-country comparison of Brazil and The Netherlands
University of Groningen
Faculty of Economics and Business
Master Thesis International Financial Management
Student Number: s2892529 Name: Andrea Lievano Cruz
Study Program: MSc IFM Supervisor: Dr V. Purice Field Keywords: CSR, environmental harmful events, stock market reaction, event
study Abstract
1
1 Introduction
Over the past decades global awareness towards the environment has increased drastically. All over the world corporations have increasingly been scrutinized due to their irresponsible behavior towards the environment, and the footprint they leave behind. It is argued that this growing social awareness has increased the pressure under which corporations find themselves to adhere to society’s norms and values.
Known examples of this have been BP’s oil scandal in 2010 and more recently Volkswagen emissions scandal in 2015. Both companies faced not only reputational repercussions but also legal and financial ones. Volkswagen had to agree to pay over 46 billion US Dollars forClean Air Act violations in the US alone (Reuters, 2016), while BP had to pay over 42 billion US dollars in civil and criminal settlements (Forbes, 2013).
Over the years many studies have investigated the relation between corporations unethical behavior and firms performance. More recently academics from different disciplines have narrowed down their scope to the study of corporations environmental behavior. This growing literature on corporations sustainable behavior has ranged from investigating the role of environmental CSR in the corporate arena to studying why corporations adopt better sustainable practices, to how it affects their corporate performance (e.g.: Etzion, 2007; Orlitzky, Schmidt and Rynes, 2003).
More recent studies have both portrayed the positive effect of environmental CSR and the negative effect of irresponsible behavior towards the environment on stock performance. For example Flammer (2013) explored both these effects on corporations stock prices, and how these evolved over time. By drawing from the resource based view and the stakeholder theory, the author found evidence that shareholders react positively to eco-friendly initiatives and negatively to eco-harmful events. Moreover, she finds that with time shareholders’ negative reaction increases, while shareholders’ positive reaction decreases.
2
focused on Asian and eastern European countries (e.g.: Chan and Cheung, 2012 and Thanestsunthorn, 2015)
Furthermore, from the perspective of the insurance based view of CSR, some authors have indicated that CSR can significantly mitigate the negative stock market reaction to the announcement of legal actions against companies (e.g.: Godfrey, Merril and Hansen, 2009) . This evidence makes us question whether CSR could mitigate shareholders reaction to corporations environmentally harmful activities.
Given these voids in the literature, the present paper aims at contributing to the literature not only by investigating and comparing shareholders reaction to corporations eco-harmful behavior in the Netherlands and in Brazil, but also by studying the role that CSR might have in mitigating shareholders reactions to corporations eco-harmful events in Brazil and in the Netherlands. Thus, through this paper the following research questions will be addressed:
RQ1: Do shareholders punish corporations environmental harmful behavior? If so, does this effect differ when comparing Brazil and the Netherlands?
RQ2a:Does CSR moderate the relationship between corporations’ eco-harmful behavior and stock prices? If so, does the moderating effect of CSR differ when comparing Brazil and The
Netherlands?
The questions addressed in this paper have already been approached by other authors. Thus, this paper is an attempt at replicating Flammer’s results. However, this paper differs in that its focus lies on studying country effects due to institutional differences in developing and developed countries. Furthermore, unlike Flammer’s paper, this paper focuses on studying shareholders reactions during only a ten year period.
3
In keeping with our theoretical framework, we found that shareholders do tend to punish corporations’ environmental harmful behavior. However, no evidence is found supporting differences in shareholders reactions across countries due to varying institutions. Nor do we find proof of an insurance like effect of CSR in mitigating shareholders reactions to environmental harmful events. However, it is also possible that no evidence was found due to the small sample. Overall, the findings presented in this paper, seem to support Porter’s view of environmental performance as a resource: firms that engage in eco-harmful behavior are punished by investors, as they may perceive that important resources may end up being used in covering legal and clean-up costs.
The present paper is divided as follows: the next section discusses the theoretical background leading up to the hypotheses. The third section represents the data and methodology used. Lastly, the fourth and fifth sections discuss the results and conclusions respectively.
2 Literature Review
2.1 Unethical behavior and Shareholders reaction
Questionable business practices have been studied by many academics through the years. Many have found evidence that companies that behave unethically suffer a significant penalty to their stock prices. Early examples of this have been provided by authors such as Frooman (1997), who studied the market’s reaction to events of socially irresponsible and illicit behavior. Through the use of a meta analyzes of 27 events, the author found that for firms engaging in illicit and irresponsible behavior, shareholders suffer a significant loss to their wealth. More recent examples are provided by Zeidan (2013), who through the study of 128 publicly traded banks over a 20 year period, found a negative market reaction to corporations illegal behavior.
It is said that these losses are attributable to the direct costs of corporations facing legal sanctions. However, some authors ( e.g.: Karpoff & Lott, 1993) argue that these costs alone don’t explain the actual magnitude of the loss in shareholders wealth. Some academics propose that besides these direct costs, firms who engage in unethical behavior suffer from reputational loss.
4
customers tend to penalize offending firms by decreasing their purchases of products and services (Gunthorpe, 1997), investors may withhold from investing in firms (Baucus & Baucus, 1997), employee moral may suffer due to fear of personal reputation loss (Zahra, Priem, & Rasheed, 2005), and managers may spend resources, that might be put to better use, in trying to repair a company’s reputation (Langus & Motta, 2010). Overall it seems that a loss in reputation may not only lead to losses in future cash flows, but also increases in cost of capital, potential employee turnover, and forgoing profitable investments.
2.2 Shareholders sensitivity to unsustainable behavior
According to the resource based view theory, firms gain competitive advantage through the unique resources they have at their disposal. This theory has been used to illustrate why corporations engage in environmental CSR. Proponents argue that companies that foster resources in support of the environment are likely to gain competitive advantages and hence higher profits (Flammer, 2013). For example Porter (1991), argued that environmental pollution is a waste of resources, and as such is detrimental to the organization.
Jones’ (1995) instrumental stakeholder theory can further be used to argue the benefits of being environmentally responsible. Jones proposes that social responsible efforts can be useful in obtaining stakeholder support: by engaging in environmentally responsible behavior companies can improve their reputation and appeal to concerned stakeholders.
Through the institutionalization of environmental CSR, awareness towards environmental harmful behavior has been increasing. As social norms towards the environment evolved, society at large has started to look down upon companies that behave poorly towards the environment. One example of this is presented by Flammer (2013). In her study she argues that the more sustainability is institutionalized the more the negative news will have a negative effect on the firm. It seems as if acting responsibly towards the environment has become more of an ethical norm to society. And thus, actions by corporations against the environment seem to be penalized very similarly as other unethical behaviors.
5 H1a: Announcement of eco-harmful behavior leads to a decrease in stock prices.
2.3 Country effects
Institutional theory states that for organizations to be able to thrive, they must adhere to the rules and beliefs systems in the environment in which they operate ( DiMaggio & Powell, 1983; (Meyer & Rowan, 1977). This institutional isomorphism allows firms to gain legitimacy (Hoffman, 1999). However, according to theories of comparative political economy, it has been argued that in the absence of institutions, firms tend to behave opportunistically in order to maximize their profit even if it means behaving irresponsibly (Campbell, 2007). Furthermore, some argue that this behavior tends to vary across different countries: in his paper on why corporations behave responsibly, Campbell (2007) creates a theoretical framework based on the institutionalization theory, to argue that in countries with less normative and regulative institutions, corporations are less likely to behave socially responsible.
It is argued that regulatory and legal institutions have the power to impact corporate behavior (Henriques & Sardosky, 1996). Environmental regulations have been shown to influence corporations environmental performance, by constraining their behavior through the use of fines and penalties ( (Rivera, 2004); (Winter & May, 2001)) . We argue that corporations, that violate these regulatory institutions, may suffer from decreases in stock prices, as shareholders may perceive that the organization will suffer from operational constrains and waste of resources. Moreover, in countries with stronger regulatory institutions, noncompliance will lead to greater decrease in stock prices.
Ortiz-de-Mandojana, Aguilera-Caracuel, and Morales-Raya (2016) argue that normative institutions can also influence firms’ environmental practices. Normative institutions are defined as “social norms, values, beliefs and assumptions about human nature and behavior that are
socially shared and held by individuals in a given country” by Kostova (1999). They include “rules of thumb, standard operating procedures, occupational standards, and educational curricula” (Hoffman, 1999). As mentioned before, when organizations go against society’s norms
and values they risk their reputation and legitimacy. So, in countries with stronger normative institutions, corporations will suffer more, and thus incur in greater drop in stock prices.
6 2.3.1 Brazil vs the Netherlands
Brazil’s environmental legislation has slowly been evolving through the years: from a period of practically no existence, to a period characterized with state intervention to introduce a balance between economic growth and environmental preservation, to more recently a period of adherence to international standards of sustainability (Drummond & Barros-Platiau, 2006).
With constant accusations of deforestation and pollution against corporations, it seems like these regulative efforts have not been effective enough despite the increased efforts towards environmental protection (De Oliveira, 2002). It has been argued that ineffective regulatory intervention has been caused among other by lack of regulatory enforcement (Bechtel & Tosun, 2009) (Campbell, 2007). Moreover, several sources (Cole, 2007; Drummond & Barros-Platiau, 2006) have documented the role that corruption can play in preventing effective environmental oversight: polluting firms that bribe officials are able to avoid punishment, thus rendering regulatory pressures ineffective (Campbell, 2007).
Corruption in Brazil has been known to be a persistent social problem. In 2015, the Transparency International Corruption Index, ranked Brazil at the 76th position among other 168 countries. This index is known for measuring the perceived levels of corruption in the public sector (Transparency International, 2016). With a score of 38, Brazil ranks among nations with serious corruption problems.
Another argument explaining the lack of environmental enforcement is related to the country’s economic growth. It said that national governments who’s focus lies in rapid economic growth, tend to ignore environmental and social consequences by easing corporate regulations, in hopes of retaining local investment (Oze & Kusku, 2009). As an emerging economy, Brazil greatest asset has always been its natural resources. However, the exploitation of these resources has come at the price of Brazil’s natural environment. Although the country has slowly started to adopt new policies to protect the environment, the government keeps prioritizing the interests of business at the cost of the environment (AZO CleanTech, 2015).
7
Unlike Brazil, the Netherlands is perceived as one of the least corrupt countries in the world according to the Transparency International Corruption Index: ranking at the top ten of their list (Transparency International, 2016).
Furthermore, as a post-industrial society, the Netherlands has evolved from an production-based economy to a service-based economy. As such its values and norms are probably no longer dictated by its need to grow rapidly through industrialization. Therefore, society comes to expect more sustainable practices from organizations (Bertels & Peloza, 2008).
Based on the this institutional comparison of Brazil and the Netherlands, it seems like environmental norms in Brazil have not yet evolved in the same manner as environmental norms in the Netherlands. In general, environmental institutions in Brazil tend to lack enforcement and suffer from corruption, while Dutch institutions are heavily enforced and adopted by the society at large. As discussed before, the institutional theory indicates that firms can suffer from risks to their reputation and legitimacy due to poor environmental behavior. And this reputational loss can affect the performance of the firm due to damage to firms relationship with its stakeholders. Differences in institutionalization of sustainability suggests that in countries that showcase weak environmental institutions, shareholders will have a weaker reaction towards negative environmental news. So, we argue that shareholders, will not penalize firms in Brazil as heavily as firms in the Netherlands when engaging in eco-harmful behavior. Thus, we propose the following hypothesis:
Hypothesis 1b: The announcement of corporations eco-harmful behavior leads to a greater decrease in stock price in the Netherlands than in Brazil.
2.4 Moderating effect of CSR
The signaling theory posits that companies disclose CSR information to contribute to the improvement of their reputation. Moreover, Toms (2002) proposes that implementation, monitoring, and disclosure of environmental policies contribute significantly to the creation of environmental reputation.
8
insurance view of CSR: negative stock market reaction to the announcement of legal actions against companies is significantly mitigated for firms that participate in institutional CSR activities. Whether CSR “insurance like benefits” mitigate the effect of eco-harmful news is still not known. So, we propose the following hypothesis:
Hypothesis 2a: In the context of firm’s eco-harmful events, the decline in stock prices is smaller for firms that engage in CSR activities than for firms that do not.
However, this line of research implicitly assumes that CSR has a similar signaling effect across different institutional environments. However, given the vast differences in institutional environments, the effectiveness of CSR as a mechanism to mitigate corporations’ negative environmental news may vary across varying institutional environments.
We argue that in countries with weaker institutions, CSR will have a greater mitigating effect. In countries with weaker institutions, companies will be better able to signal to stakeholders their ability to fill in institutional voids, and thus benefit more from the reputational effect of engaging in CSR. So we propose the following hypothesis:
Hypothesis 2b: The moderating effect of CSR on the relationship between corporations eco-harmful behavior and stock prices is smaller for firms in Brazil than for firms in the Netherlands.
3 Methodology
Event study methodology has been used frequently in the finance literature to investigate market’s reaction to corporate events and to study market efficiency. As the present paper investigates the stock market’s reaction to the announcement of environmentally harmful corporate news, we employ an event study methodology. Furthermore, because we are interested in understanding how shareholders reactions are related to firms characteristics, such as country and CSR performance, we perform an OLS regression.
3.1 Data collection
9
environmental events evolves over time. To do this the author used time periods of ten years, between which she expected differences in shareholders reactions. Her results supported her claims. So, we argue that a period of ten years is short enough to exclude the possible significant changes of environmental norms over time, and long enough to be able to collect sufficient events. With this in mind we collect events in the period from 2006 to 2015.
Following Flammer’s methodology, news about environmental events were collected by using the following keywords and variations thereof: “Pollution”, “Toxic / Hazardous waste”, “Contamination”, “Spill”, “Emission”. Furthermore, in order to collect as many events as possible for both Brazil and Netherlands, these keywords were searched in three different languages: Dutch, Portuguese, and English. A draw-back of this approach, as stated by Flammer, is that the keyword search may be too narrow, and thus may exclude other events.
In order for events to be included in the final dataset, the events had to comply with the following criteria: (1) no other significant events must occur at the time of the event, such as earning announcements; (2) the companies of interest must be publicly traded at their respective stock exchange (Brazilian companies in the Bovespa stock market and Dutch companies in the Euronext stock market); (3) stock market data must be available during the estimation period and event window. After applying these criteria, a total of 45 events were collected, comprising of 21 Dutch events, and 24 Brazilian events.
After collecting the events, historical stock prices were collected from DataStream. For each event, data was collected on 203 trading days; 2-3 days to calculate the event window and 200 days for the estimation period. Further, data was collected on market returns by using the AEX index and IBOVESPA index. When performing multi-country event studies, it is important to take into consideration differences in country-specific occasions such as holidays. Park (2014), argues that this may create a problems, as stock market data may be unavailable for those days. As suggested by Campbell et al. (2010), we employ the “trade to trade” method to deal with this issue. As Campbell et al. explain this entails “omitting missing-price days from the calculations while
accounting for the corresponding market-index returns when the stock eventually trades” .
10
3.2 Event study
Event studies are commonly used to examine stock price movements due to specific corporate events. More specifically, this type of study seeks to determine whether changes in stock prices during an event are abnormal. Whether these events have an effect on stock prices, is determined by calculating the cumulative average abnormal returns (CAAR). In order to calculate the CAARs, a series of steps are followed.
First, the abnormal daily returns (AR) for each event are calculated. This is done by subtracting the expected return from the actual return during the event window:
𝐴𝑅𝑖𝑡 = 𝑅 𝑖𝑡− 𝐸(𝑅𝑖𝑡) ( 1)
In event studies the event date (T=0) is set as the day of the news publication. However, Flammer(2013) explains, that because there is uncertainty present about the actual date of the event, academics often use an event window encompassing the days before, during and after the event. Following Flammer’s methodology a 2 day (T-1 to T) event window is employed for the analysis and a 3 day (T-1 to T1) event window is used as a robustness test.
To calculate the expected returns the literature mentions several models. The most used ones are the constant mean model and the market model. Although Brown and Warner (1985) find that the constant mean model yields similar results to more sophisticated models, the market model is argued to be better as it controls for general market movements (Brooks, 2014). Thus, we use market model to estimate the expected returns. The formal specification for this model is:
𝑅𝑖𝑡 = 𝛼𝑖+ 𝛽𝑖𝑅𝑚𝑡+ 𝑢𝑖𝑡 ( 2)
where Rit represents the return of each individual firm i on day t, αi is the intercept, βi is the
systematic risk of stock i, Rmt is the market’s daily return on day t, and uit is the risk-adjusted
residual for firm i on day t.
11
𝐸(𝑅𝑖𝑡) = 𝛼𝑖+ 𝛽𝑖𝑅𝑚𝑡 ( 3)
Second, the average abnormal return (AAR) for each day in the event window is calculated by aggregating and averaging ARit:
𝐴𝐴𝑅𝑡 = 1
𝑁∑ 𝐴𝑅𝑖𝑡
𝑁
𝑖=1 ( 4)
Lastly, the cumulative average abnormal return (CAAR) is calculated by aggregating the average abnormal returns:
𝐶𝐴𝐴𝑅𝑖 = ∑𝑡=0𝑡=−1𝐴𝐴𝑅𝑖,𝑡 ( 5)
As mentioned earlier, CAAR is a measure of how much the stock returns deviate from the expected returns during the event window. To test whether CAAR significantly deviates from the expected returns we test the following hypotheses:
H0: CAAR=0
Ha: CAAR<0
12
3.3 Regression analysis
Although Park recommends to use a logistic regression when studying the effects of country differences on the direction of abnormal returns, we decide to follow Flammer’s methodology and employ an OLS regression. An OLS regression will allow to study the effect of firm characteristics on the continuous dependent variable “abnormal returns”. Furthermore, the use of an OLS regression will also allow us to rule out the influence of alternative explanations.
To test whether country effects influence the magnitude of the abnormal returns, the following regression was estimated:
𝐶𝐴𝑅𝑖,𝑗=∝0+ 𝛽1𝐵𝑟𝑎𝑧𝑖𝑙𝑗+ 𝛽2𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑖 + 𝛽6𝐸𝑣𝑒𝑛𝑡𝑖,𝑗+ 𝛽3𝑆𝑖𝑧𝑒𝑖,𝑗+ 𝛽4𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑖𝑗+
𝛽5𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖,𝑗+ 𝜀𝑖,𝑗 ( 6)
where CARi,j represents the cumulative abnormal return for company i and event j. Brazil is a
dummy variable representing “Brazil” if 1, and “not Brazil” if 0. Further, we include industry and event fixed effects. Industry fixed effects are included to control for possible heterogeneity at the industry level that could drive the results. Following Flammer’s methodology, we also include event fixed events to control for possible heterogeneity at the event level that may drive the results. As Flammer mentioned, it is important to note that these “fixed effects only control for differences
in size of the events across categories and not within categories”. Further, the variables Size,
Performance and Leverage will be controlled for.
13
may have built reputations that protect them from the influence of negative events. However, it is possible that larger firms are more scrutinized than smaller firms and thus suffer more from reputational loss. Performance is operationalized as the return on assets (ROA), and is expressed as the ratio of net income to total assets. Better performing firms may be better prepared to cover the costs inherent to environmental harmful behavior. On the other hand, poor performing firms may drive investors to sell their stock, as they may not be able to cover the costs of their actions. The variable Leverage is defined as the ratio of the book value of total debt to the book value of total equity. High leveraged firms represent riskier investments than lower leveraged firms, as they are more exposed to default and bankruptcy risks. More leveraged firms may have more difficulty in meeting their obligations towards debt and equity holders if the costs of their environmental harmful behavior is high enough.
To test whether CSR influences abnormal returns due to eco-harmful behavior, and whether CSR moderates the relation between country and abnormal returns, the following regression is estimated:
𝐶𝐴𝑅𝑖,𝑗,𝑡=∝0+ 𝛽1𝐵𝑟𝑎𝑧𝑖𝑙𝑗+ 𝛽2𝐶𝑆𝑅𝑖,𝑗,𝑡−1+ 𝛽3𝐶𝑆𝑅𝑖,𝑗,𝑡−1∗ 𝐵𝑟𝑎𝑧𝑖𝑙𝑗+ 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑖+
𝛽5𝐸𝑣𝑒𝑛𝑡𝑖𝑗+𝛽6𝑆𝑖𝑧𝑒𝑖,𝑗+ 𝛽7𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑖,𝑗+ 𝛽8𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖,𝑗+ 𝜀𝑖,𝑗,𝑡 ( 7)
where CAR i, j, t represents the cumulative abnormal return for company i, event j, and year t. Here
we add the variable CSR and the interaction term CSR*Brazil to the model. The variable CSR was operationalized following the methodology of Cheng et al. (2013). Cheng et al. calculate CSR scores based on three key performance indicators (KPI) from DataStream ASSET4: corporate governance score, environmental performance score1, and social performance score, also called ESG scores. Cheng et al. argue that each score should be given an equal weight when calculating the overall CSR score. Flammer argues that to ensure that the CSR scores are not influenced by the events, it is necessary to lag the scores by one year. Thus, CSRt-1 scores are calculated as the
lagged average of the three KPI. The interaction term CSR*Brazil, represents the interaction between Country and CSR. In both the regression analyses robust standard errors clustered at the company level are used, to address possible heteroscedasticity across groups2.
1 Regression analysis was repeated using environmental performance scores instead of CSR scores. Results obtained
gave similar results.
14
4 Results
4.1 Shareholders Reaction to Environmental Issues
The analyses start with a simple event study to test the hypothesis whether shareholders react negatively to the announcement of eco-harmful behavior. For this analysis, the whole sample of 45 events together is examined. Shapiro-Wilk normality tests were performed on both the CARs for the (-1,0) and (-1,1) event windows. The tests, presented in appendix B, indicate a non-normal distribution for (-1,0) event window, and a normal distribution for the (-1,1) event window. Figure 1 presents a scatterplot of the cumulative abnormal returns during the (-1,0) event window.
Figure 1: scatterplot cumulative abnormal returns for eco-harmful events in the (-1,0) event window
Figure 1 show that the cumulative abnormal returns in the (-1,0) event window consists of 24 negative CARs and 21 positive CARs. Furthermore, the scatterplot indicates that the positive CARs are clustered nearer to 0 than the negative CARs. The figure also hints towards the presence of 1 outlier: when looking into the data this outlier belongs to the MMX event in 2008, when the company was publicly accused and fined for illegal logging activities.
Table 1 presents the results from the parametric standard test statistic and the non-parametric generalized sign test. The second column in the table shows significantly negative CAARs of -.01 and -.006 for the (-1,0) and (-1,1) event windows respectively, both significant at the 5% level. However, because there are issues of non-normality for the (-1,0) event window, and the sample is small, the analysis is repeated using the non-parametric generalized sign test. The last column shows the z-scores from the generalized sign test. The results for the generalized sign test indicate
15
that the CAAR are only statistically significant (p-value< 0.05) for the (-1,0) event window. However, this is not the case for the (-1,1) window. These results are supportive of hypothesis 1a that shareholders tend to react negatively to eco-harmful corporate news. More specifically, when looking at the two day event window, stock returns of firms that exhibit eco-harmful behavior seem to drop by 1% (CAAR=-0.01). Furthermore, the results are consistent with Flammer (2013), who in accordance with the resource based view found declines in share prices related to the announcement of corporations eco-harmful behavior.
Table 1:event study
Cumulative average abnormal returns
Event window
CAAR Z-value
Parametric test
Z-value Generalized Sign test
(-1,0) -0.010 -6.98** 5.77**
(-1,1) -0.006 -3.36** 1.24
Note: This table presents the cumulative average abnormal returns for the event windows (-1,0) and (-1,1), with the corresponding z-value between parentheses. The symbols **, and *** denote statistical significance of the CAARs at 0.05, and 0.01 levels respectively
4.2 Descriptive statistics
Table 2 presents a summary of the descriptive statistics of the key variables used in the regression analysis. The table reports the summary statistics for Brazil and the Netherlands separately and together. For a more expanded view of the descriptive statistics, see appendix C.
Table 2: descriptive statistics Descriptive statistics
Variable Brazil Netherlands All
Mean Std. dev. N Mean Std. dev. N Mean Std. dev. N
CAR -0.016 0.039 23 -0.004 0.022 21 -0.010 0.032 44
Size 18.374 1.717 23 16.913 1.809 21 17.677 1.891 44
Leverage 0.7134 4.878 23 0.753 0.759 21 0.732 3.528 44
Performance 0.007 0.086 23 0.078 0.053 21 0.041 0.079 44 CSR 62.915 15.058 21 77.695 14.232 19 69.935 16.298 40
Note: This table presents the descriptive statistics for the main variables according to country. “CAR” is the cumulative abnormal return for the (-1,0) event window, “Size” is the natural logarithm of total assets, “Performance” is the ratio of net income to total assets, “Leverage” is the ratio of book value of debt to book value of equity, “CSR” is the average of the ESG scores.
16
firms tend to be slightly larger; perform worse; have a slightly lower debt-to equity ratio; and have a lower CSR score than the sample of Dutch firms.
4.3 Univariate results
Table 3 details the correlation coefficients between the key variables and their significance. From the table it is observable that only the variables Performance and the interaction “CSR*Brazil” exhibit a correlation with CAR, both significant at the 10% level. This goes against the expectations that country effects are responsible for differences in CAR. Also, there seems to be no indication that CSR may have insurance like effects. A very high correlation between the interaction “CSR*Brazil” and the dummy “Brazil” is also observed. This indicates a serious problem of multicollinearity. Brooks (2014) mentions 4 strategies on how to deal with this problem: increasing sample size, dropping one of the collinear variables, transforming the variables, or ignoring the problem. In the second part of the regression analysis, it is decided to drop one of the collinear variables, and then repeat the test including the previously dropped variable, while dropping the other previously included collinear variable.
Table 3:correlation matrix Correlation Matrix Variable 1 2 3 4 5 6 7 1. CAR 2. Brazil -.203 3. Size -.133 .448*** 4. Performance .277* -.641*** -.354** 5. Leverage .032 .348** .132 -.403*** 6. CSR .023 -.458*** .353** .293* -.331** 7. CSR*Brazil -.265* .947*** .578*** -.596*** .241 -.222
8. Chemic & pharmaceutical b -.028 -.350** -.346** .173 -.179 .212 -.332**
9. Construction & warehousing b .031 -.442*** -.504*** .529*** -.134 .089 -.418***
10. Transportation b .192 -.168 -.070 -.050 -.055 .085 -.159
11. Banking b .023 -.011 .377** -.229 .568*** .122 .005
12. Manufacturing b .130 -.299* -.179 -.013 -.136 -.052 -.284*
13. Petroleum, steel & mining b -.147 .756*** .530*** -.367** .045 -.2486 .709***
14. Pollution c -.223 -.045 -.222 .085 -.284* -.094 -.085 15. Oil spill c .0814 -.246 -.044 .266* -.193 .069 -.193 16. Violation c .088 -.095 .058 -.011 -.206 .175 -.029 17. Deforestation c .018 -.017 .298* -.039 .342** .188 .015 18. Toxic waste c .086 -.011 -.069 -.255 .182 -.039 -.074 19. Emission c .086 .399** .087 -.166 .367** -.228 .357**
Note: This table presents the Pearson correlations for the regression variables. “CAR” is the cumulative abnormal return for the (-1,0) event window, “Brazil” is the dummy variable for Brazil and the Netherlands, “Size” is the natural logarithm of total assets, “Performance” is the ratio of net income to total assets, “Leverage” is the ratio of book value of debt to book value of equity, “CSR” is the average of the ESG scores, and “CSR*Brazil” is the interaction between country and CSR.
17
4.4 Shareholders’ Reaction to Environmental Issues: Brazil Vs the Netherlands
In this section the hypothesis that “the announcement of corporations eco-harmful behavior leads
to a greater decrease in stock price in the Netherlands than in Brazil” is tested. As mentioned in
the methodology, an OLS regression is implemented to test this hypothesis and to rule out confounding effects. When performing the OLS regression, a test of normality of the residuals was performed. The results in appendix D, indicate that there were no violations of this assumption. Furthermore, as mentioned in the methodology, white clustered robust standard errors where used in order to account for possible heteroscedasticity across firms.
Table 4: OLS regression of Cumulative abnormal returns on Country effects Regression Analysis of Cumulative Abnormal return (-1,0)
Variable Model 1a Model 1b Model 1c
CAR Std. error CAR Std. error CAR Std. error
Brazil .002 (.007) .004 (.007) .008 (.013) Size -.001 (.002) -.003 (.003) -.003 (.003) Performance .187*** (.057) .239*** (.081) .274*** (.098) Leverage .004*** (.001) .003** (.001) .002 (.002) Chemic & pharmaceutical - - -.009 (.015) -.008 (.015) Construction & warehousing - - -.014 (.010) -.0106 (.007) Transportation - - .029*** (.007) .035** (.014) Banking - - .009 (.014) .022 (.021) Manufacturing - - .012 (.009) .010 (.009) Pollution - - - - -.007 (.008) Oil spill - - - - -.006 (.015) Violation - - - - .004 (.011) Deforestation - - - - -.009 (.017) Toxic waste - - - - .01 (.012) Constant -.008 (.039) .026 (.048) .033 (.049) Observations 44 44 44 Adjusted R2 .578 .639 .656 F-statistic 13.363 6.687 3.954
Note: This table provides estimated coefficients from regressing the CAR variable on country; control variables firm size, performance, and leverage; and event and industry fixed effects. The notations ***,**, and* denote
statistical significance at the 1%, 5%, and 10% levels respectively.
18
The results indicate no statistical significant effect of the independent variable “Brazil” on the dependent variable “CAR” across the three models. So, there seems to be no evidence supporting the proposed hypothesis: when comparing Brazil and The Netherlands, the market does not seem to react differently when corporations harm the environment. This contradicts the expectation that due to institutional differences, shareholders will react differently towards eco-harmful news.
Across the three models, when looking at the influence of the control variables on CAR, we see that “Performance” has a significant effect on “CAR”. Performance has a positive coefficient of .187 in the first model, a coefficient of .239 in the second model, and a coefficient of .274 in the third model, all three at a significant level of 1%. This means that better performing firms tend to have higher abnormal returns than worse performing firms. This conforms with the expectations that poor performing firms may drive investors into selling their stock, as the legal and clean-up costs related to eco harmful behavior may be a heavier burden to them. This supports Porter’s view that profitability and pollution reduction might not be mutually exclusive, given that pollution is a waste of resources and thus affects a firms profitability.
In the two first models, “Leverage” seems to have a very small, but significant positive influence on “CAR”: with a coefficient of .003 in model 1a and a coefficient of .004 in model 1b, at the significant level of 1% and 5% respectively, it seems that higher leveraged firms tend to be punished less harshly by investors, than less leveraged firms. However, this influence disappears when controlling for event fixed effects, which indicates that the effects of leverage on CAR, may be due to heterogeneity at the event level.
Lastly, when looking at the industry and event fixed effects, we see that only “Transportation” has a significant positive (βmodel 1a=.029, p-value<.01; βmodel 1b=.035, p-value<.05) effect on CAR. This
means that compared to other industries, firms in the Transportation industry seem to be punished less severely than firms in other industries.
4.5 CSR & Shareholders’ Reaction to Environmental Issues: Brazil Vs the Netherlands
19
was performed. The results in appendix D, indicate that there was violation of normality of the residuals, for model 2a, 2b, 2d and 2e. Furthermore, as mentioned in the methodology, white robust clustered standard errors where used in order to account for possible heteroscedasticity across firms.
Table 5:OLS Regression of Cumulative abnormal returns on CSR and CSR*Brazil Regression Analysis of Cumulative Abnormal return (-1,0) Variable
Model 2a Model 2b Model 2c Model 2d Model 2e Model 2f
CAR Std. error CAR Std. error CAR Std. error CAR Std. error CAR Std. error CAR Std. error Brazil -.005 (.008) .007 (.009) .0108 (.012) - - - - Size .000 (.002) -.004 (.005) -.004 (.004) .001 (.002) -.003 (.005) -.003 (.004) Performance .128 (.001) .220** (.095) .252*** (.081) .113 (.085) .189* (.095) .212** (.083) Leverage .002* (.091) .002 (.002) -.0004 (.003) .002* (.001) .003 (.002) .000 (.003) CSR .000 (.000) .000 (.000) .000 (.000) .000 (.000) .000 (.000) .000 (.000) CSR*Brazil - - - .000 (.000) .000 (.000) .000 (.000) Chemic & pharmaceutical - - -.016 (.020) -.009 (.016) - - -.017 (.021) -.011 (.017) Construction & warehousing - - -.021 (.013) -.016 * (.009) - - -.021 (.014) -.017 (.011) Transportation - - .024*** (.006) .0355*** (.010) - - .020*** (.006) .029*** (.008) Banking - - .014 (.022) .0285 (.025) - - .007 (.021) .020 (.024) Manufacturing - - .008 (.006) .004 (.007) - - .004 (.007) -.001 (.007) Pollution - - - - -.012 (.010) - - - - -.014 (.009) Oil spill - - - - -.003 (.012) - - - - -.006 (.012) Violation - - - - .000 (.010) - - - - -.001 (.009) Deforestation - - - - -.008 (.019) - - - - -.007 (.018) Toxic waste - - - - .011 (.010) - - - - .009 (.010) Constant -.013 (.024) .047 (.065) .055 (.056) -.025 (.027) .038 (.073) .049 (.064) Observations 40 40 40 40 40 40 Adjusted R2 .107 .219 .297 .121 .216 .287 F-statistic .811 .817 .783 .945 .798 .643
Note: This table provides estimated coefficients from regressing the CAR variable on Brazil, CSR and CSR*Brazil; control variables firm size, performance, and leverage; and event and industry fixed effects. The notations ***,**, and * denote
statistical significance at the 1%, 5%, and 10% levels respectively.
20
firm size, leverage, and performance; the fifth model adds industry fixed effects to the regression, and the sixth model adds event fixed effects to the regression.
When looking at model 2a to 2c, “Brazil” does not have any significant influence on CAR, again indicating that Brazilian and Dutch shareholders do not differ in their reaction to eco-harmful events. Across all models, “CSR” does not have a significant effect on CAR. Indicating that CSR does not moderate shareholders reaction to environmentally harmful corporate news. This contradicts Godfrey et al. (2009), who proposed that when firms engage in CSR activities, they gain positive attributions from stakeholders, who then soften their negative reactions to negative corporate news.
Likewise, the interaction between CSR and Brazil (CSR*Brazil) does not seem to have any influence on CAR, suggesting that CSR’s influence on shareholders reaction does not differ when comparing Brazil to The Netherlands. This goes against the notion that in countries with weaker institutions, companies will be better able to signal to stakeholders their ability to fill in institutional voids, and thus benefit more from the reputational effect of engaging in CSR.
21
In the model 2a and 2d of the regression, “Leverage” has a very small, but significant positive influence on “CAR”. With a coefficient of .002 in model 1 and a coefficient of .002 in model 4, both at the significant level of 10%, it seems that higher leveraged firms tend to be punished less harshly by investors. However, just as in the previous analysis, this influence disappears when controlling for industry and event fixed effects. Which indicates that the effects of leverage on CAR, may be due to heterogeneity at the industry and event level.
Lastly, when looking at the industry and event fixed effects, we see that only “Transportation” have a significant positive (βmodel 2b=.024, p-value<.01; βmodel 2c=.0355, p-value<.01; βmodel 2e=.020,
p-value<.01; βmodel 2f=.029, p-value<.01) effect on CAR. This means that compared to other
industries, firms in the Transportation industry seem to be punished less severely than firms in other industries. However, it is difficult to generalize this to all companies in the transportation industry, as this category only contained one event: In July 2008 KLM was accused of contaminating water around the airport, when one of its hangars leaked extinguishing agent. Also when looking at model 2c, “Construction and Warehousing” has a significant negative influence on CAR (βmodel 2c=-.016, p-value<.1), meaning that firms in the construction and warehousing
industries tend to be punished more severely than other firms. However, this effect disappears across all other models. It is possible that the sample size is responsible for this also.
4.6 Robustness Test
In this section a series of robustness checks on the regression analyses are performed. First we present bootstrap test of the results, given that some of the models violated the assumption of normality of the residuals. Second, we perform the test excluding outliers in order to see whether outliers drove the results. We decide not to perform a robustness test of the regression using the 3 day event window given that the non-parametric results were not significant for this event window.
4.6.1 Normality
We attempt to address the problem of non-normality of residuals of model 2a, 2b , 2d, and 2e from the second regression analysis, by bootstrapping the coefficients and standard errors of the models. According to Brooks (2014) “bootstrapping is a technique for constructing standard errors and
22
properties of the estimators (Brooks,2014). Brooks(2014) further mentions that this method is especially useful when dealing with small samples.
With the help of Eviews statistics program, the bootstrap analysis was performed by resampling the coefficient and standard errors 10,000 times. Table 6 displays the results from the bootstrap analysis. From the results, it can be seen that the dummy variable “Brazil” significantly influences CAR negatively (β=-.012; p-value<.1). Meaning that Brazilian firms tend to have 1.2% lower CARs than the Dutch firms. However, this influence disappears when controlling for industry fixed effects.
Further, “Performance” has a positive significant effect (βmodel2=.206, p-value<.01; βmodel5=.150,
p<.05) on CAR across models 2 and 5 after controlling for industry fixed effects. These results are in accordance with the OLS regression analysis discussed before.
Lastly, the bootstrap analysis shows, that “Transportation” also has a significant positive effect on CAR (βmodel2=.024, p-value<.01; βmodel5=.021, p<.01), meaning that firms in the transportation industry have higher abnormal returns than firms in other industries.
Table 6:Bootstrapping standard errors and coefficients-OLS Regression of Cumulative abnormal returns on CSR and CSR*Brazil Regression Analysis of Cumulative Abnormal return (-1,0)
Variable Model 2a Model 2b Model 2d Model 2e
CAR Std. error CAR Std. error CAR Std. error CAR Std. error
Brazil -.012* (.007) .003 (.009) - - - - Size .001 (.002) -.004 (.004) .003 (.002) -.001 (.004) Performance .068 (.044) .206*** (.058) .048 (.043) .150** (.056) Leverage .002 (.001) .003 (.002) .002 (.001) .003 (.002) CSR .000 (.000) .000 (.000) .000 (.000) .000 (.000) CSR*Brazil - - - - .000** (.014) .000 (.000)
Chemic & pharmaceutical - - -.014 (.017) - - -.012 (.018)
Construction & warehousing - - -.019 (.011) - - -.014 (.012)
Transportation - - .024*** (.006) - - .021*** (.005) Banking - - .009 (.013) - - -.001 (.013) Manufacturing - - .007 (.007) - - .004 (.006) Constant -.014 (.026) .039 (.054) -.031 (.026) .008 (.061) Observations 40 40 40 40 R2 .106 .219 .122 .2159 F-statistic .811 .817 .945 .799
Note: This table provides bootstrapped estimated coefficients and standard errors from the regression of the CAR variable on country, CSR and CSR*Brazil; control variables firm size, performance, and leverage; and event and industry fixed effects.
23 4.6.2 Outliers
When looking at the descriptive statistics and at the distribution of CARs we see the presence of outliers in the variables “CAR” and “Leverage”. Therefore, we repeat the regression analysis with winsorized versions of the variables “CAR” and “Leverage”. We chose to winsorize the variables at the 97.5% and 2.5%, instead of removing the outliers, because the sample being used is very small, and trimming it would create an even smaller sample which could cause greater problems of normality and reduce the power of the tests performed. The results from the OLS analyses with winsorized “CAR” and “Leverage” can be seen in appendices E and F. Overall, these results are quite similar to the main results and to the robustness test using bootstrap of coefficients and standard errors: after controlling for industry and event fixed effects, only performance and transportation have a significant positive effect on the cumulative abnormal returns, while “Leverage” and “CSR*Brazil” effects disappear.
Furthermore, these regressions were also submitted to normality tests of the residuals, these can be seen in appendix G. The results of these tests indicated there were violations of normality for model 1a.There were also violations of normality for model 2a,2b, 2d and 2e. So, we repeat the analyses for these models using bootstrapped coefficients and standard errors. The results can be seen in appendices H and I. Again the results are quite similar to the previous analysis:
From model one in appendix H, we see that country does not have a significant influence on CAR. On the other hand, the variables “Performance” and “Leverage” have a significant positive influence on CAR (βperformance=.138, p-value<.00;1 βleverage= .004, p-value<.01).
In appendix I, we see that in model 1a “Brazil” has a significant negative effect on CAR (β=-.013, p-value< .1) However, this influence disappears when controlling for industry and event fixed effects. As in the previous analyses, Performance and Transportation have a significantly positive influence on CAR after controlling for industry fixed effects (βperformance=.153, p-value<.00;1
βtransportation= .022, p-value<.01).
24
5
Conclusions
The purpose of this study was to investigate shareholders reaction to corporations eco-harmful behavior in The Netherlands and in Brazil, and to study the role that CSR might have in mitigating shareholders reactions to corporations eco-harmful events in Brazil and in The Netherlands. The questions addressed in this paper have already been addressed by other academics. For example Flammer (2013) studied shareholders’ sensitivity to corporations environmental footprint. The author found evidence that shareholders react negatively towards eco-harmful behavior, and that environmental CSR has insurance like effect that mitigates shareholders reaction to eco-harmful events. However, this type of study has mostly been addressed in single country settings.
The resource based view, institutional and the stakeholder theories, were used to propose that shareholders are likely to respond negatively companies’ environmental behavior. Given the differences in institutionalization of sustainability between Brazil and The Netherlands, we drew from the theory of institutionalization to propose the hypothesis that shareholders in Brazil are less sensitive to eco-harmful news than shareholders in The Netherlands. Moreover, the signaling theory and insurance based view of CSR formed the basis of our argumentation to propose the hypothesis that CSR moderates shareholders’ sensitivity to eco-harmful news. Again, we drew from the institutionalization theory to propose that, in countries with weaker institutions such as Brazil, CSR would have a greater mitigating effect on shareholders reactions towards environmentally harmful behavior.
The methodology of this study was divided into two parts. The first part focused on analyzing whether shareholders were sensitive to corporations eco-harmful behavior. This was tested by means of an event study. For this purpose data on 45 eco-harmful events were collected from Brazil and TheNetherlands. The second part of the methodology was focused on looking at the country and CSR effects. By means of an OLS regression, it was tested whether differences in countries accounted for the variability in shareholders reactions. Furthermore, an OLS regression was also implemented to test whether CSR mitigated shareholders reactions, and whether this effect was different across countries.
25
behavior. This result is not surprising, as other academics have also found evidence of stock prices declines related to the announcement of environmentally harmful behavior by firms. As mentioned earlier, in her study, Flammer (2013) found declines in share prices related to the announcement of corporations eco-harmful behavior.
The results from OLS regressions indicated that there was no statistically significant evidence to support the other three proposed hypothesis at the beginning of this paper. It is possible that country differences and CSR do not play a role on how shareholders react to negative environmental news. However, it is also possible that no evidence was found due to the small sample.
Despite finding no support for some of the proposed hypotheses, the results indicated that better performing firms tend to have higher cumulative abnormal returns than worse performing firms. It could be argued that the differences observed in cumulative abnormal returns, when comparing Brazil to The Netherlands, were due to firm performances: Brazilian firms in general where characterized with lower performance and lower cumulative abnormal returns when compared to the Dutch firms. Additionally, when controlling for industry and event fixed effects, the results indicated that firms in the transportation industry tend to be punished less than firms in other industries.
Overall, the findings presented in this paper, seem to support Porter’s view of environmental performance as a resource: firms that engage in eco-harmful behavior are punished by investors, as they may perceive that important resources may end up being used in covering legal and clean up fees. Moreover, firms with poor performance (lower return on assets), suffer even more from shareholders reaction, as they may perceive that the performance of these firms may further drop as the costs of eco-harmful behavior increases.
5.1 Limitations
26
Moreover, it is possible that the lack of empirical evidence was due to the sample size, given that small samples tend to undermine the power of statistical analyses.
Second, the data collection method of events, may have introduced selection bias. As each article had to be read and judged. The selection of events may vary depending on the researcher interpretation of the definition of environmentally harmful events. This may also have contributed to a reduction in sample size.
Third, one key drawback was the presence of multicollinearity. Problems of multicollinearity are common when dealing with interaction variables, as is the case in this study. To deal with this problem, the collinear variables were dropped from some of the models. However, this method may introduce omitted variable bias. Brooks (2014) explains that omitting variables may cause the estimated coefficients on the remaining variables to be biased. To solve this problem of multicollinearity, we emphasize the need to increase the sample size in future research.
5.2 Future research
Future research could focus on differences across different geographic regions. Comparing sets of geographic regions, rather than single countries, could provide an opportunity to study how sets of cultural differences play a role in shareholders reaction to environmentally harmful events. For this purpose future researchers could use Hofstede’s cultural dimensions to compare shareholders reactions across different geographic regions. Moreover, studying a broader set of countries would also provide the opportunity to study a broader set of firms, and thus help solve the problem of small sample sizes.
Another future avenue would be to study differences across industries. This would provide an interesting a path, as companies’ institutional setting also depends on their industry. Through differences in industry regulation and self-regulation, and peer pressure, shareholders may have different incentives to react to environmental eco-harmful news.
27
6 References
AZO CleanTech. (2015, June 19). How clean is your country? (B. Smith, Editor) Retrieved August 15, 2016, from AZO CleanTech: http://www.azocleantech.com/article.aspx?ArticleID=547
Baucus, M., & Baucus, D. (1997). Paying the piper: An empirical Examination of longer-term financial consequences of illegal corporate behavior. Academy of management journal, 40(1), 129-151.
doi:10.2307/257023
Bechtel, M., & Tosun, J. (2009). Changing economic openness for environmental policy convergence: when can trade agreements induce convergence of environmental regulation? International Studies Quarterly, 53(4), 931–953.
Bertels, S., & Peloza, J. (2008). Running just to stand still? Managing CSR reputation in an era of ratcheting expectations. Corporate Reputation Review, 11(1), 56-72.
Brooks, C. (2014). Introductory econometrics for finance (3rd ed.). United Kingdom: Cambridge University Press.
Brown, S., & Warner, J. (1985). Using daily stock returns: the case of event studies. Journal of Financial Economics, 14(1), 3-31.
Campbell. (2007). Why would corporations behave in socially responsible ways? An institutional theory of corporate social responsibility. Academy of Management Review, 32(3), 946-967.
Campbell, C. J., Cowan, A. R., & Salotti, V. (2010). Multi-country event-study methods. Journal of banking & finance, 34(12), 3078-3090.
Chan, A., & Cheung, H. (2012). Cultural dimensions, ethical sensitivity and corporate governance. Journal of Business ethics(110), 45-59. doi:DOI 10.1007/s10551-011-1146-9
Cheng, B., Ioannou, I., & Serafeim, G. (2013). Corporate socialresponsibility and access to finance. Strategic Management Journal, 35, 1-23.
Cole. (2007). Corruption, income and the environment: an empirical analysis. Ecological Economics, 62(3-4), 637–647.
De Oliveira, J. (2002). Implementing environmental policies in developing countries through
decentralization: the case of protected areas in Bahia, Brazil. World Development, 30(10), 1713–1736. DiMaggio, P., & Powell, W. (1983.). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48, 147-60.
Drummond, & Barros-Platiau. (2006). Brazilian environmental laws and policies 1934–2002: a critical overview. Law Policy, 28(1), 83-108.
Etzion, D. (2007). Research on Organizations and the Natural Environment, 1992-Present: review. Journal of management(33), 637-664. doi:DOI: 10.1177/0149206307302553
Flammer, C. (2013). Corporate Social Responsibility and Stock Prices: The Environmental Awareness of Shareholders. Academy of Management Journal, 3(56), 758-781. Retrieved from
28 Forbes. (2013, February 5). BP Fighting A Two Front War As Macondo Continues To Bite And
Production Drops. (A. Fontevecchia, Editor) Retrieved August 15, 2016, from Forbes:
http://www.forbes.com/sites/afontevecchia/2013/02/05/bp-fighting-a-two-front-war-as-macondo-continues-to-bite-and-production-drops/#2eafddd02b87
Frooman, J. (1997). Socially irresponsible and illegal behavior and shareholder wealth: metha-analysis of event studies. Business Society, 36(3), 221-249. doi:10.1177/000765039703600302
Godfrey, P. C., Merril, C. B., & Hansen, J. M. (2009). The relationship between corporate social responsibility and shareholder value: an empirical test of the risk management hypothesis. Strategic Management Journal, 30(4), 425-445. doi:10.1002/smj.750
Gunthorpe, D. (1997). Business ethics: A quantitative analysis of the impact of unethical behavior by publicly traded corporations. Journal of business ethics, 16(5), 537-543. doi: 10.1023/A:1017985519237 Henriques, I., & Sardosky, P. (1996). The determinants of an environmentally responsive firm: an empirical approach. Journal of Environmental Economics and Management, 30(3), 381-395.
Ho, F., Wang, H., & Vitell, S. (2012). A Global analysis of corporate social performance: the effect of cultural and geographic envirnment. Journal of business ethics, 4(107), 423-433.
Hoffman, A. (1999). Institutional evolution and change: environmentalism and the U.S. Chemical Industry. The Academy of Management Journal, 42(4), 351-371.
Jones, T. M. (1995). Instrumental stakeholder theory: a synthesis of ethics and economics. Academy of management review, 20, 404-437.
Jordan, A. (2013). Environmental Policy in the European Union: Contexts, Actors and Policy Dynamics. London: Routledge.
Karpoff, J., & Lott, J. (1993). The reputational penalty firms bear from committing criminal fraud. Journal of law and economics, 36(2), 757-802. Retrieved from http://dx.doi.org/10.1086/467297 Keijzers, G. (2000). The evolution of Dutch environmental policy: The changing ecological arena from 1970–2000 and beyond. Journal of Cleaner Production, 8(3), 179-200.
Kostova. (1999). Transnational transfer of strategic organizational practices: a contextual perspective. Academy of Management Review, 24(2), 308-324.
Langus, G., & Motta, M. (2010). The effect of EU antitrust investigations and fines on the firms valuation. EUI Working paper.
Meyer, J., & Rowan, B. (1977). ‘Institutionalized organizations: Formal structure as myth and ceremony’. American journal of sociology, 83, 340–363.
Orlitzky, M., Schmidt, F., & Rynes, S. (2003). Corporate Social and Financial Performance: A Meta Analysis. Organization Studies, 403-411.
Ortiz-de-Madojana, Aguilera-Caracuel, & Morales-Raya. (2016). Corporate governance and
environmental sustainability: the moderating role of the national institutional context. Corporate Social Responsibility and Environmental Management, 23, 150-164.
29 Park, N. (2014). A guide to using event study methods in multi-country settings. Strategic management journal, 25, 655-688. doi:10.1002/smj.399
Porter, M. E. (1991). America's green strategy. Scientific America, 264(4).
Reuters. (2016, January 8). VW blasted for shielding emissions documents from U.S. probe. (D. Shepardson, Editor) Retrieved August 15, 2016, from Reuters: http://www.reuters.com/article/us-volkswagen-emissions-emails-idUSKBN0UM1WZ20160108
Ringov, D., & Zollo, M. (2007). Corporate responsibility from a socio-institutional perspective: The impact of national culture on corporate social performance. Corporate Governance: The international journal of business in society, 7(4), 476-485. doi:10.1108/14720700710820551
Rivera. (2004). Institutional pressures and voluntary environmental behavior in developing countries: evidence from the Costa Rican hotel industry. Society & Natural Resources, 17(9), 779-797.
Thanestsunthorn, N. (2015). The impact of national culture on corporate social responsibility: evidence from cross regional comparison. Asian journal of business ethics, 4, 35-56. doi:DOI 10.1007/s13520-015-0042-2
Toms, J. (2002). Firm resources, quality signals and the determinants of corporate environmental reputation: some UK evidence. The British Accounting Review, 34(3), 257-282.
Transparency International. (2016, January). Corruption perception index 2015. Retrieved January 6, 2017, from Transparency International: http://www.transparency.org/cpi2015#results-table
Winter, S., & May, P. (2001). Motivation for compliance with environmental regulations. Journal of Policy Analysis Management, 20, 675-698.
Zahra, S., Priem, R., & Rasheed, A. (2005). The antecedents and consequences of top management fraud. Journal of management, 31(6), 803-828. doi:10.1177/0149206305279598
30
7 Appendix A-Events
Environmental harmful events
Country Company Event Date Source
Brazil Banco do Brasil Deforestation 1-Apr-11 Agence France Presse Brazil Banco da Amazonia Deforestation 1-Apr-11 Agence France Presse
Brazil Braskem Pollution 17-Jun-11 ICI Chemical News
Brazil Braskem Emission 5-Jun-14 ICI Chemical News
Brazil CSN Pollution 21-Aug-13 Esmerk
Brazil CSN Emission 16-Sep-11 Esmerk
Brazil CSN Emission 1-Jul-09 Business News Americas
Brazil CSN Emission 4-Aug-10 O Globo
Brazil CSN Toxic waste 16-Sep-15 Business News Americas
Brazil MMX Pollution 2-Jun-11 Esmerk
Brazil MMX Deforestation 22-Jul-08 Business News Americas
Brazil Petrobras Pollution 28-Nov-14 Esmerk
Brazil Petrobras Emission 8-Feb-13 Platts Oilgram News
Brazil Petrobras Oil spill 6-Feb-12 Oil & Gas News
Brazil Petrobras Pollution 2-Dec-10 Esmerk
Brazil Petrobras Emission 1-Sep-09 Esmerk
Brazil Petrobras Pollution 25-Sep-07 Associated Press International Brazil Petrobras Violation of environmental
regulations 6-Mar-06 Aberdeen Press and Journal
Brazil Usiminas Pollution 1-Sep-09 Esmerk
Brazil Vale SA Pollution 5-Nov-15 Associated Press International
Brazil Vale SA Violation of environmental
regulations 26-Nov-13 Business News Americas
Brazil Vale SA Pollution 28-Feb-11 Inter Press Service
Brazil Vale SA Deforestation 11-Jul-08 Business News Americas
Brazil BRF Pollution 12-Jul-06 Aberdeen Press and Journal
NL AkzoNobel Pollution 23-Feb-13 Algemeen Nederlands Persbureau
NL Boskalis Oil spill 30-Sep-08 Rotterdams Dagblad
NL Boskalis Pollution 5-Feb-13 NRC Handelsblad
NL DSM Violation of environmental
regulations 7-Feb-10 De Volkskrant
NL DSM Pollution 19-Sep-08 Algemeen Nederlands Persbureau
NL DSM Pollution 12-May-11 Dagblad de Limburger
NL Heineken Pollution 12-Mar-07 AFX International Focus
NL Heineken Violation of environmental
regulations 19-Apr-11 UK Government News
NL ING Deforestation 22-Nov-13 Het Financieele Dagblad
NL KLM Pollution 21-Jul-08 Het Parool
NL Philips Toxic waste 10-Jun-08 Thomson Financial News Super
Focus
NL Royal Vopak Pollution 17-Dec-12 SeeNews Netherlands
