Sustainability Performance, Litigation Provisions and Corporate Bond Spreads

Contents

1 Introduction

2 Literature Review
2.1 Sustainability and Cost of Debt
2.2 Sustainability and Litigation Risk

3 Data and Methodology 9
3.1 Sustainability Performance and Corporate Bond Yield Spreads
3.2 Sustainability Performance and Litigation Provisions
3.3 Mediation Analysis of Litigation Provisions
3.4 Moderation Analysis of Litigation Provisions
3.5 Event study: Volkswagen Scandal & Paris Climate Agreement

4 Results
4.1 Descriptive Statistics
4.2 Sustainability Performance and the Corporate Bond Spread
4.2.1 Moderation Analysis: ESG, Time, Bond Spreads
4.2.2 Moderation Analysis: ESG, Industries, Bond Spreads
4.2.3 Moderation Analysis: ESG, Countries, Bond Spreads
4.3 Sustainability Performance and Litigation Provisions
4.3.1 Moderation Analysis: ESG, Time, Litigation Provisions
4.3.2 Moderation Analysis: ESG, Industries, Litigation Provisions
4.3.3 Moderation Analysis: ESG, Countries, Litigation Provisions
4.4 Mediation Analysis: Litigation Provisions
4.5 Moderation Analysis: Litigation Provision
4.6 Event study: Volkswagen Scandal &Paris Climate Agreement

5 Discussion and Limitations
5.1 ESG performance and Cost of Debt
5.2 ESG performance and Litigation Provisions
5.3 Mediation and Moderation Analyses of Litigation Provisions
5.4 Limitations

6 Conclusion

References

1 Introduction

Over the last decades, there has been significant research conducted in environmental and financial economics, exploring the relation between corporations’ sustainability performance and their cost of capital. Whereas most studies indicate a negative relationship between sustainability performance and the cost of equity (El Ghoul, Guedhami, Kwok, & Mishra, 2011; Sharfman & Fernando, 2008), the relationship between sustainability performance and the cost of debt remains unclear. Results in this field are often contradicting, with fairly limited on the direction of the relationship (Baran & Zhang, 2012; Bauer & Hann, 2010; Ge & Liu, 2015; Hoepner, Oikonomou, Scholtens, & Schr¨oder, 2016; Menz, 2010; Oikonomou, Brooks, & Pavelin, 2014; Sharfman & Fernando, 2008). In this paper, I investigate this relationship by using a global dataset from 2003 to 2016, trying to answer the following research question: What is the effect of companies’ ESG performance on their yield spread? To shed light on the relation between sustainability performance, also often referred as ESG performance, and the cost of debt, I aim to look at one of the reasons that may influence the relationship between the two. The variable of interest that is included in this paper is litigation provisions, used as a proxy for litigation risk. Literature in this field often high- lights potential effects that strong and weak ESG performances can have on the potential for future litigations (Hong & Kacperczyk, 2009). Therefore, I attempt to answer the following research question: How does ESG performance affect corporations’ litigation risk?

Including litigation provisions enables the study to test for potential relationships.The first relationship that is investigated focuses on whether ESG performance affects the yield spread through an increase in litigation risk. In this causal scenario, ESG performance would affect the risk of litigation, which then in turn would affect corporations’ yield spreads. Hence, I try to answer the question: Does ESG performance affect the corporate bond yield spread through litigation risk? The second causal setting that I test, is whether the level of litigation risk in place determines the magnitude and significance of ESG performance on the yield spread. In this causal scenario, the risk of litigation determines whether sustainability performance affects the corporate bond yield spread. This scenario aims at answering the following question: Does the magnitude of the relationship between ESG performance and the yield spread change with the level of litigation risk in place?

Understanding the role of the litigation risk channel could help to shed light on the con- troversial findings that prior literature presents. As well as investigating the role of litigation risk, additional moderation analyses are conducted for time, industry, and country in which corporations are located. From an academic perspective as well as from a business perspec- tive the clarification of this relationship is of high interest. Explaining the relationship of ESG performance and the cost of debt can incentivize corporations to further engage in sus- tainability. The quantification of the relationship would help managers to justify investments in ESG improvements. Furthermore, clarifying which sub-dimensions of sustainability may be especially strongly recognized by capital markets, could help to improve fund allocations of sustainability projects.

In addition to investigating the aforementioned relationships, I conduct an event study, combining the recent Volkswagen scandal and the Paris Climate Agreement of 2015, us- ing an adjusted difference-in-difference setting, to prove causality between ESG and the yield spread. This study contributes to academic research by providing insights and poten- tial explanations for contradictory findings in previous papers. Furthermore, it contributes additional evidence of a potential causal relationship between corporations’ environmental performance and its financial performance, underlining the importance of sustainability.

The remainder of the paper is structured as follows: First, key findings in literature will be reviewed in both the sustainability and cost of debt context as well as the litigation risk field. Second, the methodology will be discussed along with the sample selection and tests that are conducted. Third, results of all regressions will be presented. Fourth, implications of these results will be discussed along with limitations to the study. Finally, a conclusion will be provided summarizing all major findings of this paper.

2 Literature Review

The field of sustainability performance has gained importance over the last decades, leading to a tremendous amount of research conducted in this field. One of the major questions, which researchers have been looking to answer is, whether ESG performance affects the fi- nancial performance of a company. Along with the performance on the stock market of high sustainability companies (Eccles, Ioannou, & Serafeim, 2014), a major research field focused on is the relationship to the cost of capital. Whereas most of these efforts focus on the cost of equity, relatively little research has been conducted in the debt financing context of sus- tainability performance. Not only does the amount of research conducted differ, but also the results of these areas. On the one hand, papers investigating the relationship between ESG and the cost of equity mainly find a negative relationship (Bassen, Meyer, & Schlange, 2006; Chava, 2014; El Ghoul et al., 2011). On the other hand, papers investigating the relationship to the cost of debt capital find mixed results. This paper aims at clarifying the relationship to provide academics and practitioners a deeper understanding of the relationship between ESG performance and the cost of debt.

2.1 Sustainability and Cost of Debt

According to Klein (2015), sustainability factors need to be included into the analysis of fixed-income assets, as these have the potential to reduce both portfolio and idiosyncratic risk. In his argumentation, these may reduce the risk of receiving a lower credit rating, being prone to a higher price volatility, and higher yield spreads. Indeed, Bradley, Chen, Dallas, and Snyderwine (2007) find a difference for credit ratings related to governance factors. Fur- thermore, Derwall and Koedijk (2009) find a significant outperformance of a SR-balanced portfolio. This paper will focus on the last potential benefit mentioned by Klein (2015), investigating the effect of sustainability metrics to yield spreads. Looking at the cost of debt, the literature is split into two levels. One investigating sovereign bonds at a country level and the other focusing on a corporate level. The findings in the field of sovereign bonds are mostly similar, indicating a negative relationship between countries ESG performance and the bond spread (Capelle-Blancard, Crifo, Diaye, Scholtens, & Oueghlissi, 2016; Crifo, Diaye, & Oueghlissi, 2015; Hoepner et al., 2016). Besides the investigation of sovereign bonds, research has also investigated whether banks value sustainability in their decision of granting loans. Nandy and Lodh (2012) provide evidence that firms’ sustainability perfor- mance matters as they find that banks reward eco-friendly firms with lower bond spreads. Next to that, Kleimeier and Viehs (2016) find that companies that voluntarily disclose their carbon emission levels show lower yield spreads at bank loans. Contrary to these results, Goss and Roberts (2011) only find marginal significance of bank loan conditions related to sustainability factors. According to them, CSR performance does not matter for most of the companies analyzed. Only comparing the worst CSR firms to the best, showed a significant difference of 20bps. Due to the modest premiums found, Goss and Roberts (2011) argue that sustainability performance of companies does not matter for companies’ bank loans. Overall, results show that on country level sustainability matters for sovereign bond yield spreads. On the bank-level, sustainability performance is at least somewhat being recognized and rewarded.

Turning to the investigation of the relationship between debt issued by companies and their sustainability performance, one finds even more mixed results. Whereas, some studies find a negative relationship among the two (Bauer & Hann, 2010; Ge & Liu, 2015; Oikonomou et al., 2014), arguing that ESG performance is being rewarded by the credit market, other studies find the opposite (Baran & Zhang, 2012; Menz, 2010; Sharfman & Fernando, 2008). According to Chava (2014) there are two main reasons why lenders would be interested in the sustainability performance of the borrower. First, the potential to be held liable as a lender for e.g. environmental damages. Second, the potential loss of reputation due to neg- ative publicity of the borrowing company. Taken together, their paper finds that the loan syndicate size reduces, as borrowers of capital avoid companies with a low sustainability performance. In line with this argumentation is the finding of Bauer and Hann (2010) that companies having environmental concerns are associated with higher bond spreads, hence a higher cost of debt, and vice versa. Similarly, findings provided by Oikonomou et al. (2014) show that credit markets reward a strong and penalize bad CSR performances, with lower and higher bond spreads, respectively. Ge and Liu (2015) confirm the findings and further- more find that these effects are pronounced for bonds that have an investment grade, that are financially healthy, and that show a weak corporate governance. Nevertheless, as mentioned previously not all studies have found the relationship between sustainability and cost of debt to be negative. In the research of environmental risk management and the cost of capital from Sharfman and Fernando (2008), a negative relationship is found between sustainability performance and the cost of equity, but a positive one for the cost of debt. Next to that, the study from Menz (2010) investigates the same relation on the corporate bond level and finds that socially responsible investors are exposed to a higher risk premium, hence a higher cost of debt capital. A potential explanation for being penalized for strong ESG performance is discussed by Kru¨ger (2015), who elaborates on the value maximization of sustainability practices. In studies that indicate a positive relationship between ESG performance and the yield spread, sustainability may be viewed as overspending, thus more as a cost rather than a benefit for bondholders. Contrary to this perspective, the other studies, finding a nega- tive relationship, indicate that sustainability benefits outweigh the costs. This paper will investigate the relationship between ESG performance and yield spreads. Due to a slightly larger number of prior studies that indicate a negative relationship between sustainability performance and the yield spread, this paper expects the relationship to be negative.

Hypothesis 1: ESG performance is negatively associated to the corporate bond spread.

Besides investigations of sustainability performance on an aggregate level, the recent study by Hoepner et al. (2016) breaks down this analysis even further into sustainability sub-scores of country’s institutional frameworks. The study finds that most of the effect be- tween sustainability performance and yield spreads is shaped by environmental and societal performance scores. In line with these findings, I expect to find a stronger significance of the environmental and social factors in the analysis of sub-dimensions.

Hypothesis 2: The relationship between ESG performance and the corporate bond spread is mainly driven by environmental and societal factors.

In face of increasing awareness of sustainability aspects and an increased number of envi- ronmental scandals during recent years, I also expect the relationship gain importance over time. Due to the punishing effects that could be observed by bad sustainability performance, I expect that capital markets increasingly reward sustainability performance as a risk hedge (Hong & Kacperczyk, 2009; Bauer & Hann, 2010), thus a negative trend should be observ- able in the relationship between ESG and the yield spread.

Hypothesis 3: The relationship between ESG performance and the corporate bond spread gained significance and negativity in magnitude overtime.

2.2 Sustainability and Litigation Risk

Besides the direct investigation of the relationship between sustainability performance and the cost of capital, other factors have been investigated that also may influence the yield spread. Studies have been conducted to test the relationship between sustainability measures and the credit rating of the company (Attig, El Ghoul, Guedhami, & Suh, 2013; Bradley et al., 2007) and the overall firm risk (Jo & Na, 2012). The paper of Attig et al. (2013) finds that strong social performance is associated to a higher credit rating. This finding is especially driven by the stakeholder management of the company. Mostly in line with Attig et al. (2013) are the results of Bradley et al. (2007), who find that credit ratings are negatively related to measures of antitakeover protections for firms with an investment grade rating. Nevertheless, this finding indicates a positive relationship for firms with a speculative rating. These findings highlight that, in presence of information asymmetry, antitakeover protections are perceived as negative for companies. This finding can be explained by the chance of management entrenchment and connected agency problems. Next to the inves- tigation of the credit rating relationship, the relationship of sustainability performance of cash flow volatility has been investigated. Furthermore, Jo and Na (2012) find that CSR engagements are negatively associated to firm risk, rejecting the window dressing hypothesis in favor of the risk reduction hypothesis. To my knowledge, no elaborate investigation has been undertaken regarding the relationship between sustainability and litigation risk, even though prior studies identified litigation risk to have a potentially significant impact.

According to the study of Hong and Kacperczyk (2009), companies that operate in so- called sin-industries, which include tobacco, alcohol, and gaming, are being neglected by norm-constrained investors, due to a potentially higher risk of litigations. They highlight that litigation risk is being increased by social norms. The link between sustainability per- formance and litigation risk is also being mentioned in the paper of Bhagat, Bizjak, and Coles (1998) who found that shareholders significantly lost wealth after environmental law- suits. Besides the two studies, Bauer and Hann (2010) mention environmental violations or allegations as potential reasons for why the sustainability performance affects the cost of debt negatively. In line with the argumentation of Hong and Kacperczyk (2009) I expect to find a negative relationship between sustainability measures and the litigation risk.

Hypothesis 4: ESG performance is negatively associated to litigation risk.

Furthermore, as highlighted in previous studies, I expect the environmental sub-score (Bhagat et al., 1998) and the social sub-score (Hong & Kacperczyk, 2009), to account for most of the effect.

Hypothesis 5: The relationship between ESG performance and litigation risk is mainly driven by environmental and societal factors.

Besides, considering environmental scandals during the last decade, such as Deepwater Horizon, the Fukushima catastrophe, and the Volkswagen scandal, I expect the effect of ESG performance on litigation risk to have increased over years.

Hypothesis 6: The relationship between ESG performance and litigation risk is expected to gain significance and negativity in magnitude overtime.

Similar to the study of Bassen et al. (2006), who investigates the role of company risk in the setting between sustainability performance and financial performance, this paper investi- gates the role of litigation risk in this given setting. Nevertheless, the exact role of litigation risk remains uncertain, whether it may be a mediating role or a moderating role.

The question answered in the mediating analysis is how one variable affects another one (Hayes, 2013). In this case, I investigate if sustainability performance affects the bond spread through litigation risk. In this mediation setting, sustainability performance is expected to be negatively associated to litigation risk. Lower risk is then expected to be recognized by debt capital markets in form of lower yield spreads.

Hypothesis 7: Sustainability performances affect the corporate bond spread through the risk of future litigations.

The question investigated by the moderation analysis is whether the magnitude of an effect is influenced by a specific variable, hence if the relation is different when the moder- ation variable is in place (Hayes, 2013). In this causal setting, sustainability performance is expected to have a stronger effect on the yield spreads in a setting where the level of litigation risk is high.

Hypothesis 8: The effect of sustainability performance on the corporate bond spread increases in an environment with higher litigation risk.

3 Data and Methodology

To investigate the relation between sustainability performance, ligation risk, and the cost of debt, a unique global dataset from 2003 to 2016 is created combining data from Thomp- son Reuters Datastream and from FactSet. The sustainability performance in this study is measured with the ASSET4 scores available on Datastream on aggregate level, in form of the overall sustainability score (ESG Score), but also on a more detailed level with sub- dimensions. These include the environmental, social, governmental and economic score. Each of the sub-dimensions depicts a different dimension on which the sustainability per- formance of a company is evaluated. Datastream only facilitates information on ”Litigation Expenses and Provisions” and ”Litigation Provisions”. As the first of the two is both a forward-looking (the provisions) as well as a backward-looking (the expenses) measure, liti- gation provisions, which is only a forward-looking measure, is chosen as proxy for litigation risk in this study. I expect it to better display the potential of future risks. According to Tippett (2016) provisions provide a strong and significant proxy and indicator for litigation risk. From an accounting-perspective, litigation provisions mark the potential amount of fu- ture expenses that are likely to be incurred when the company is involved in lawsuits. They are ”measured at the best estimate (including risks and uncertainties) of the expenditure re- quired to settle the present obligation, and reflects the present value of expenditures required to settle the obligation where the time value of money is material” [International Accounting Standard 37]. Funds that are being flagged as provisions will not be spent, reallocated, or invested, thus they significantly impact the businesses’ operations. As it is the businesses best interest to manage funds as efficiently as possible, I expect the litigation provisions vari- able to capture all expected future lawsuit expenses. As the data for litigation provisions is available on a yearly interval only, all other variables of the study are downloaded in yearly intervals as well.

To draw conclusions about the cost of debt, all bonds that have been issued by companies between 2003 and 2016 are downloaded from FactSet. Subsequently, the yield spread of the respective bond is matched with Datastream data, using the ”Yield Spread Over Benchmark” variable. The controls included can be separated into two categories. First, company-specific variables and, second, bond-specific control variables. Company-specific variables include controls for returns, size, liquidity, and leverage using the ROA and EBIT, Sales, the EBIT- Interest-Coverage, and the Debt-to-Equity ratio, respectively. Other company-controls were excluded beforehand due to high correlations among the measures. Bond-specific attributes include the Modified Duration, the Coupon, and the S&P rating of the bond. Furthermore, I control for differences in time as well as the companies’ industry and the geographical location.

To generate a final sample, all companies and observations are excluded that did not provide all the variables mentioned above, leaving the study with a final dataset of 4,797 observations from 175 companies ranging from 2003 to 2016. The variable that limited the sample size the strongest, due to its limited availability, is litigation provisions. To answer the research questions, several regression are conducted. As the analyses conducted uses time- series data, all regressions in this paper are corrected for time fixed effects. Additionaly, not only cultural differences may play a significant role in the awareness and importance of sustainability, but also the industry in which a company operates. Therefore, industry and country fixed effects are included, as the yield spread and litigation provisions are expected to significantly change along these two dimensions. The resulting panel least squares (PLS) regressions are further clustered to adjust the standard errors for their origin company.

3.1 Sustainability Performance and Corporate Bond Yield Spreads

Before the role of litigation risk between sustainability performance and the cost of debt can be tested, several analyses are conducted beforehand. First, I analyze the general relation between the ESG score and the corporate bond yield spread in a PLS, to observe whether, and if yes how, corporate sustainability performance influences the cost of debt. After testing the significance of the overall performance, the same analysis is conducted for the ASSET4 ESG sub-dimensional scores. To gain an even deeper understanding of the relationship between sustainability performance and corporate bond yield spreads, I further investigate the role dimensions of time, country, and industry by including interaction terms. Given a significant relationship between sustainability performances and yield spreads, moderation tests can be conducted with additional variables to observe whether the potential moderation variable affects the magnitude of the previously established relationship. Tables 5 to Table 7 show the results for the moderation analyses, including the interaction terms ESG Score*Year, ESG Score*Country, and ESG Score*Industry.

Abbildung in dieser Leseprobe nicht enthalten

In these equations the YieldSpread relates to the ”Yield Spread over Benchmark” variable. The ESGSc o r e relates to the respective sustainability performance measure. Controls in equations 1 to 4 include company characteristics, such as the ROA, Debt/Equity, Sales and EBIT Interest Coverage, bond characteristics, including Modified Duration, Coupon, and S&P Rating, as well as fixed effects for time, industry, and country. Furthermore, i represents the respective company observations at time t.

3.2 Sustainability Performance and Litigation Provisions

After providing an elaborate overview of the initial relationship, I investigate the role of lit- igation risk. Before a mediation analysis of litigation provisions in the context of ESG score and bond spreads can be conducted, the affiliation between the ESG score and its sub-score to litigation provisions must be investigated and proven to be significant. Due to the high spread of the litigation provisions variable and its non-normality, a logarithmic transforma- tion is applied to this variable. Like the first set of regressions, the relationship is tested using a PLS regression. Furthermore, more in-depth analyses are conducted here as well, including a moderation analysis of time, country, and industry, using the same interaction terms as in the previous section.

Abbildung in dieser Leseprobe nicht enthalten

In the equations above the ln(LitigationProvisions) is the litigation risk proxy downloaded from Datastream. As in equations 1 to 4, the ESGSc o r e relates to the sustainability perfor- mance measure. For the equations 5 to 8 Controls relates to only company-characteristics controls and fixed effects for time, country and industry. Furthermore, the variables Y e a r, Industry, and Country represent dummy variables taking the value of 1 for the respective year, industry, and country, and otherwise 0.

3.3 Mediation Analysis of Litigation Provisions

After proving a significant connection between sustainability performance and both the yield spread and litigation provisions, I use the litigation provisions variable to test for a poten- tial mediating effect. A meditation is present when some of the total effect between two variables, in this case sustainability performance and yield spread, is running through the mediating variable. From a causality perspective, a mediation process answers the question- ing of how the independent variable affects the dependent variable. In the given context the question that is being targeted is: Does ESG performance affect the corporate bond yield spread through litigation risk? The mediation analyses of this paper include the Sobel-test (Sobel, 1982) and bootstrapping. In general, both tests analyze whether the indirect effect, which is the effect of sustainability performance on litigation provisions, a, multiplied with the effect of the litigation provisions on the yield spread, b, is significantly different from zero.

Indirect Effect: ESG performance → Litigation Provisions → Yield Spreads

To test this, the Sobel test uses the slopes and standard errors of the regressions between sustainability performance and litigation provisions as well as litigation provisions and the yield spread. It uses the null hypothesis T a ∗ T b = 0, which is being tested by the alternative hypothesis being T a ∗ T b / = 0. The ratio to calculate the respective z-value, is displayed in equation 9, using the standard error. A remark regarding the test is that it assumes nor- mality and tends to be relatively conservative, hence having a relatively high Type II error.

Abbildung in dieser Leseprobe nicht enthalten

Where a represents the effect of sustainability on litigation provisions and b describes the coefficient of the litigation provision variable on the bond spread. The variable s 2 represents the variance and se ab the pooled standard error for the regressions.

Even though, Hayes and Scharkow (2013) find that more than 90% of mediation tests agree with each other, I use bootstrapping, which does not assume the sample data to be normally distributed as the Sobel test, to increase the robustness of my results. It de- scribes a computer-intensive analysis technique that can be applied to non-normal data, as it simulates datasets by using resampling with replacement (Hesterberg, Monaghan, Moore, Clipson, & Epstein, 2003). As a result, the tests are conducted in each of the generated simulated datasets. In this study for each mediation analysis 1,000 bootstraps were gener- ated to create a 95% confidence interval of whether the effect is different from zero. Each of the bootstraps calculated the indirect effect (a*b) going through litigation provisions. After- wards, the 1,000 indirect effects were ranked and the 95% confidence interval was generated. If the resulting CI excludes the value of zero, significant mediation is found. According to Hayes (2013), bootstrapping presents the state-of-the-art test for testing a potential medi- ating effect.

3.4 Moderation Analysis of Litigation Provisions

The moderation analysis is conducted by adding an interaction term of sustainability per- formance and litigation provisions to the PLS. If the interaction term is found to provide a significant p-value (p < 0.05), one can assume this variable to be a moderator. A moder- ation in this case implies that litigation provisions affect the magnitude of the relationship between sustainability performance and bond spreads. Hence, from a causality perspec- tive the sustainability performance has a differently strong effect on the yield spread when litigation risk is in place. This section targets the question: Does the magnitude of the rela- tionship between ESG performance and the yield spread change when litigation risk is higher?

Abbildung in dieser Leseprobe nicht enthalten

In the equations above, Controls relate to company-specific, bond-specific, as well as fixed effects controls for time, industry and country. The variables EnvScore, SocScore, GovScore, EcoScore describe the sustainability sub-scores of environmental, social, governmental, and ecological performance, respectively.

3.5 Event study: Volkswagen Scandal & Paris Climate Agreement

To provide an indication of a possible causal relationship of sustainability performance on the corporate bond yield spread, I use two events in the end of 2015 in combination.

The first of the two events is the Volkswagen scandal, which started in September 2015 when the EPA firstly filed a ”Notice of Violation” due to violations against the Clean Air Act. During subsequent months, the stock price dropped by roughly 30% as both media attention and investigations increased. It turned out that Volkswagen conducted environ- mental fraud by intentionally manipulating the carbon emission measures of their vehicles. This exogenous shock did not only increase the awareness of environmental performance, but it highlighted the significant downside risks that companies have with a bad environmental performance have.

The second event that took place in December 2015 is the sealing of the Paris Climate Accord, which represents a global effort to reduce greenhouse gas emissions to keep global warming below two degrees Celsius. After unsuccessful negotiations in Durban 2012 and in Doha 2013, finally in 2015 all 196 participating countries could agree on the common effort.

This agreement is perceived as a strong signal in favor of environmental practices and its relevance in upcoming years. In combination, both the Volkswagen scandal and the Paris Climate Agreement taking place during the later months of 2015, provide a strong signaling that:

1. Environmental performance is of significance for businesses’ success.
2. Bad environmental performance is being significantly punished by capital markets as well as by governments.
3. Environmental concerns will gain even more importance in the future.

Because of these events, I expect a higher punishment of capital markets for firms that underperform on environmental factors, due to an increased risk profile. Hence, the expec- tation is an increase in the yield spread for companies with a low environmental score.

To conduct this analysis, an adjusted difference-in-difference approach is employed using a treatment dummy, an interaction dummy and a time-dummy. In a normal difference- in-difference model two identical groups are observed of which one receives a treatment. Therefore, causality can be explained with the difference among the two groups after the treatment is given. In this event study, the setting is slightly different. I split corporations into two groups with one different characteristic: the environmental performance. Both groups are observed before and after a ”global treatment / event” to analyze how the rela- tion between the two groups has changed after the event.

The ”treatment group” in our case are companies that have either a ”Poor Environmen- tal Performance” or a ”Below Average Environmental Performance”- ASSET4 score, hence corporations that score 41.6 points or less. The second group is made of all corporations in the sample with a higher environmental score than 41.6. The difference-in-difference inter- action term is built by multiplying the treatment dummy with the time dummy, including all data points after 2015. If significant, this interaction term will provide the change of bond spreads for companies with a low environmental performance. The expectation for the interaction term is to be positive and significant. With this event study, I aim to provide an indication that sustainability performances affect the bond spread and not vice versa.

[...]