Master Thesis The impact of greenhouse gas emissions on financial performance and the moderating effect of foreign ownership

(1)

The impact of greenhouse gas emissions on financial performance and

the moderating effect of foreign ownership

Abstract

The purpose of this paper is to investigate the potential impact of carbon emissions on financial performance, measured by firm value and cost of capital, and the moderating effect of foreign ownership. Thereby the paper addresses the question whether firms can financially benefit from lowering their greenhouse gas (GHG) emissions. Using a sample of 1,416 firm-year observations from FTSE350 firms over the years 2008-2016, the results show that firms emitting less carbon than the median within their industry are valued 0.270 times higher by the market than their peers who emit the same or more than that industry median. However, low GHG intensity firms do not benefit from a lower cost of capital, nor are the main relationships influenced by the amount of shares held by foreign investors. Furthermore, using the event study methodology with 2,090 hand-collected events, no short-term value implications are found, as stock prices do not react to firms’ annual GHG emission announcement.

Keywords: GHG emission, Firm value, Cost of capital, Foreign holdings,

Master Thesis

Fokko Posthumus (S3271161) MSc International Financial Management

Faculty of Business and Economics University of Groningen Supervisor: dr. S. Homroy

(2)

2

1. Introduction

With growing extreme weather events and natural disasters being associated with greenhouse gas emissions, the worldwide effort to oppose pollution has attained enormous attention in recent years. In December 2015, 196 countries engaged in an universal climate deal to restrain global warming and reduce global emissions (NPR, 2015). Where the Paris Agreements requires each country to exert its utmost effort by planning and reporting its own contributions, the United Kingdom (UK) was the first country to introduce mandatory disclosure on greenhouse gas (GHG) emissions. Their Companies Act 2006 was enacted in July 2013 and obligates all publicly listed firms to report comprehensive data on GHG emissions in their annual reports (Krüger, 2016). In addition to governments taking responsibility for the environment, investors increasingly demand non-financial information transparency. Greater awareness among investors will further endorse climate change mitigation policies and significantly cut down carbon footprints (Andersson, Bolton and Samama, 2016). As a result, climate change has turned into an important business issue for companies (Busch and Hoffmann, 2011). Yet, partly because of data which is only recently becoming increasingly available, little is known about the financial consequences of mandatory carbon disclosure and the importance of GHG emissions for shareholders to make investment decisions. The purpose of this paper is to investigate the potential impact of GHG emissions on financial performance, reflected in firm value and the firm’s ability to raise capital.

(4)

4 aimed at financial information (See Verrecchia, 2001; Leuz and Wysocki, 2016) and the majority of research on GHG emission focuses on voluntary disclosure (see Matsumura et al., 2014; Lewandowski, 2017). Therefore, the UK provides an exclusive environment to study mandatory GHG emission disclosure effects on firms.

Moreover, different owners may have different objectives and decision-making horizons (Hoskisson et al., 2002). Assuming that corporate social responsibility (CSR)1_{participation is} a result of decisions made by the corporate managers under pressure from the shareholders (Oh, Chang and Martynov, 2011), it is worthwhile to examine the effect of different ownership structures on corporate decisions regarding CSR. According to Johnson and Greening (1999), outside director representation is positively associated with the environment dimension of CSR, because the number of outside directors on the board increases diversity and stakeholder orientation, as a result diverse boards are more inclined to comply with environmental standards and avoid negative media exposure. More specifically, recent literature finds that, especially among foreign investors, the effects of information asymmetry are amplified (Leuz, Lins, & Warnock, 2010). According to Haniffa and Cooke (2005), Malaysian firms dominated by foreign shareholders engage in more corporate social disclosure as a legitimation strategy to acquire capital and satisfy ethical investors. When foreign investors seek legitimacy in the UK by being environmental responsive, they may influence CSR decision making within the firm and therefore the relationship between GHG emission and its economic consequences. To my knowledge, such an investigation has not yet been conducted.

In conclusion, climate change has become an important issue for businesses, but research on the effects of carbon emissions and its mandatory disclosure on firms’ financial performance is lacking. With environmental problems that are increasingly global and markets that incorporate non-financial information disclosures, the issue that rises is how GHG emissions specifically impact firm value and its ability to raise capital. Moreover, this relationship might be influenced by the relative proportion of foreign shareholders within the firm. To address these subjects, this paper is divided into two parts. In the first series of tests, unbalanced panel regression analysis are used to examine the impact of a firm’s GHG emission relative to their industry mean on company value and cost of capital. Subsequently, using daily stock price data and the event study methodology, I explore whether there is a value effect on the day that firms

(5)

5 announce their emission and how this depends on the corresponding GHG emission value. The outcomes help to understand whether or not firms can financially benefit from lowering and disclosing their GHG emissions, which is of great concern for firms’ strategic decision-making. Moreover, studying British firms enables an examination of how firms respond to mandatory environmental regulations, which is of great importance for policy makers and adds to the debate of government intervention.

(6)

6

2. Literature view

The extent and magnitude of economic consequences of climate change is surrounded by complicated, scientific uncertainties. However, it has long been acknowledged that climate can impact a country’s economic performance to a great extent (Dell et al., 2014). Consistently, studies report that rising temperatures raise economic risk, reduce economic performance and reduce labor productivity (Bansal et al., 2015; Dell et al., 2009; Verisk Maplecroft, 2015). Pollution is a negative externality and recent discoveries indicate that the decay rate of GHG in the atmosphere changes as temperature levels increase. In essence, climate change depicts the trade-off between human wellbeing and the constraints placed on the environment. This paper addresses the need of evidence of an impact at the firm level, which is in its infancy. The U.K. the first country that introduced mandatory disclosure on greenhouse gas (GHG) emissions for publicly listed firms. The Companies Act 2006 became operational in 2013, which lends itself for a comparison between pre and post mandatory regulation effects. Before 2013, in a voluntary disclosure environment, information asymmetry was high and firms could use voluntary disclosure as a marketing instrument. As of 2013, disclosure had no information value, as every company needs to disclosure. In this section, prior literature is discussed, relations are explained and placed into context and hypotheses are formulated.

2.1 GHG emissions and firm value

(7)

7 What makes this paper unique compared to previous studies, is the focus on absolute GHG emissions and its mandatory disclosure. Because the UK was the first country to introduce mandatory disclosure on GHG emissions and The Companies Act 2006 required all listed companies to report their annual GHG emissions as of 2013, it offers an exclusive setting to examine pre and post mandatory regulation effects. Before 2013, in a voluntary disclosure environment, information asymmetry was high and firms could use voluntary disclosure as a marketing instrument. As of 2013, disclosure had no information value, as every company needs to disclosure. Environmental regulation increases transparency which alleviates information asymmetry on the market. To my knowledge there is only one, yet unpublished, study, by Krüger (2016), that examines the effect of mandatory disclosure in the UK on firm value.

Several recent studies, in which single-country and international samples are used, have directly examined the valuation impact of carbon emissions and their voluntary disclosure. Chapple, Clarkson and Gold (2013) conclude that carbon intensive Australian firms experience lower market values. Furthermore, higher levels of carbon emissions negatively affect firm value in the US and the market further penalizes non-disclosing firms, which results in an even lower firm value (Matsumura et al., 2014). In particular, Lewandowski (2017) shows that for firms with superior carbon performance, annual carbon emission and market value tend to be positively associated, but this association is negative for firms with inferior carbon performance. The two components of carbon performance are annual carbon emission and enhancement in carbon performance over time. Finally, by focusing on the European emission trading scheme (ETS), Clarkson, Li, Pinnuck and Richardson (2015) are one of the first to include a mandatory element. Using an European sample, they find that carbon emissions tend to vary negatively with GHG emissions, but the impact depends on the difference between a firm’s emissions and its allowance under the European ETS and on firm and industry characteristics.

(8)

8 consumers (Clarkson et al., 2015). Moreover, firms that operate in higher carbon-emitting industries are more prone to put effort in managing carbon risks (Kim, An and Kim, 2015). For this reason, in this paper, the value implication of a firms’ absolute GHG emission relative to its industry mean is investigated.

In sum, investors seem to incorporate climate risks to evaluate total firm risk, which influences investment decisions. Following previous literature, the relationship between annual GHG emission and firm value is presumably negative. Firms emitting more GHG face more reputational damage and higher regulatory costs. As a result, they are penalized by the market through a lower firm value, although value implications seem to depend on firm and industry characteristics. In particular, firms that report more GHG emissions relative to their industry peers, are likely perceived as environmentally underperforming and might experience a lower firm value.

H1a: Firms reporting less GHG emission relative to their industry peers experience higher firm value.

Nevertheless, firm value effects have a relative long-term character. To determine whether firm value effects are indeed caused by GHG emission-related information, and therefore if hypothesis 1a can be reinforced, an event study is conducted. By using daily stock price data, an event study determines the influence of a specific event on firm value (MacKinlay, 1997). The event of interest is the press release of a firm’s annual GHG emission. By focusing on investors’ response to firm disclosures about GHG emissions, short-term value impacts can be observed. By doing so, the issue whether GHG emission disclosures contain valuable information for investment decisions is addressed. An event study on GHG emission disclosure is done in one recent study by Griffin et al. (2017). They discover that firms listed on the S&P 500 with lower (higher) GHG intensity, measured as GHG emission per dollar of revenue, show higher (lower) unsigned excess returns around the announcement date. This outcome appears to be coherent with the aforementioned negative association between GHG emission and firm value.

H1b: Firms reporting less (more) GHG emission relative to their industry peers experience higher (lower) stock excess returns on the day they disclosure their annual GHG emissions.

(9)

9 marketing instrument. As of 2013, disclosure had no information value, as every company needed to disclose. Accordingly, environmental regulation increased transparency which alleviated information asymmetry on the market. Therefore I expect that the effect of GHG emission on firm value is less strong after 2013.

2.2 GHG emissions and the cost of capital

A firm’s cost of capital is the cost of a firm’s financing including both borrowings and stockholder equity (Li et al., 2014). Determining how GHG emissions, a non-financial performance measure, is reflected in the cost of capital, is therefore important for both management and investment decisions. In general, finance and accounting literature show that increased transparency through better information disclosure by firms lead to a lower cost of capital because of reduced information asymmetries and uncertainties (see Verrecchia, 2001; Lambert, Leuz, and Verrecchia, 2007). As discussed earlier, information asymmetry occurs when the market knows less relative to the manager (Verrecchia, 2001). In particular, the accuracy of information available to investors decreases their uncertainty regarding future cash flows, and therefore the risk premium they require for holding the firm (Easley and O'hara, 2004). Moreover, Lambert et al. (2012) identify that, if there is perfect competition on the market, a key source of information risk that can impact the cost of capital is information accuracy. Accordingly, Francis and Khurana (2005) find evidence that a higher level of disclosure results in a lower cost of external financing. Hence the prediction is that a higher level of financial disclosure is related to a lower cost of capital. More specifically on non-financial disclosures, Dhaliwal, Li, Tsang and Yang (2011) find that firms that voluntarily issue CSR reports encounter a decrease in their cost of capital if their CSR performance is superior. Consistently, firms providing an integrated report, in which social and environmental responsibilities are included, experience a lower cost of capital (García-Sánchez and Noguera-Gámez, 2017).

(10)

10 associated with the CoD for Australian firms. Correspondingly, in their study on voluntary disclosure, Kleimeier, Stefanie, and Viehs (2015) conclude that higher transparency regarding carbon emissions results in more favorable loan conditions. Subsequently, Jung, Herbohn and Clarkson (2014) document a positive relationship between the CoD and carbon risk for Australian firms that do not respond to the Carbon Disclosure Project (CDP) survey. Although their findings relate to the CoD, they appear consistent with Matsumura et al. (2014) that the market penalizes non-disclosing firms by a lower firm value.

In sum, previous literature suggests a negative relationship between GHG emission and the CoD. Increased transparency leads to lower uncertainty among investors. Moreover, in terms of regulatory costs, less carbon-intensive firms have less uncertainty regarding future cash flows. As a result of less uncertainty, firms can obtain more favourable borrowing conditions.

H2a: Firms reporting less GHG emission relative to their industry peers experience a lower cost of debt.

(11)

11 that disclose their GHG emission to the CDP project. The project obtains emission data through the use of extensive questionnaires (Chapple et al., 2013). The CDP project has a voluntary character, because firms are not obliged to respond to the surveys. Secondly, in their Korean study, Kim et al. (2015) examine firms who are affected by the Korean trading scheme. This plan, comparable to the European emission trading scheme (ETS) studied by Clarkson et al. (2015), requires firms that emit more than a particular amount of CO2 to disclose their carbon emissions. Although there is a mandatory element in these country-level studies, not all publicly listed firms are affected under these reduction plans. This underlines ones more that the UK provides an unique setting and that research on the mandatory disclosure of GHG emission on cost of capital is still in its infancy.

To finalize this section, empirical results on an association between GHG emission and CoE are ambiguous. However, it is of great importance for firms to know whether they can benefit from a lower cost of capital, both debt and equity, by reducing their GHG emissions. Furthermore, all three studies use the same line of reasoning. Firms with an effective carbon management actively reduce their exposure to carbon risks. Subsequently, firms with lower carbon emission intensities have lower uncertainty regarding future cash flows. As a result, they are rewarded by the equity market by a lower CoE.

H2b: Firms reporting less GHG emission relative to their industry peers experience a lower cost of equity.

In addition to hypothesis 1, it is likely that also hypothesis 2 is not constant over time. The effect of GHG emissions on the cost of capital is probably less strong after 2013, because the Companies Act increased transparency and alleviated information asymmetry on the market.

2.3 The moderating effect of foreign holdings

(12)

12 information asymmetry problem (Oh et al., 2011). Moreover, firms can apply legitimacy strategies to convince the public that it is acting socially responsible (Haniffa and Cooke, 2005). According to Haniffa and Cooke, foreign stakeholders have different interests and power and might therefore exert other pressure to legitimize the company.

When foreign investors, concerned with reputational damage, seek legitimacy in the UK by being environmental responsive, they may influence CSR decision making within the firm and therefore the relationship between GHG emission and financial performance. To my knowledge, such an investigation has not yet been conducted and it is therefore timely to investigate this phenomenon. The outcomes are helpful for firms who want to generate social value alongside shareholder value, for shareholders as the results reflect their potential impact on the environment and for investors who want to invest in more environmental responsive firms.

(13)

13 In order to reduce GHG emissions, firms need to make investments and allocate resources. It is therefore important to understand how CSR decision making is influenced by foreign ownership and add evidence to the little, but contradicting literature. CSR can be seen as a form of investment, but CSR engagement is likely to be perceived as costly in the short-term. I argue that because foreign owners are more inclined to seek legitimacy with the wider stakeholder community, they are more long-term oriented and are therefore more aware of the long-term benefits of investments to reduce carbon emissions. As a result, they will encourage GHG emission reduction and disclosure. Moreover, shareholders holding a substantial proportion of firm equity cannot easily divest and might therefore influence management to pursue long-term goals (Johnson and Greening, 1999). Hence the influence on CSR decision making is probably larger when a higher proportion of shares is held by foreigners, which leads to the proposition of hypothesis 3.

H3a: The negative relationship between firm GHG emissions and firm value is strengthened by foreign ownership.

H3b: The positive relationship between firm GHG emissions and cost of debt is strengthened by foreign ownership.

(14)

14

3. Research design

From the first two sections, the central question that arises is whether firms can financially benefit from lowering and disclosing their GHG emissions. Accordingly, three main hypothesis were constructed. In this chapter, data collection, variable construction and research methods are discussed.

3.1 Sample and data

To invesitgate the impact of GHG emissions and its mandatory disclosure on financial performance, a sample is constructed containing firms that are affected by the Companies Act. The legislation obligates all publicly listed firms in the UK to report comprehensive data on their GHG emissions. For this reason, all companies listed on the Financial Times Stock Exchange 350 (FTSE350) are examined over the years 2007 – 2016. This timespan allows for a comparisson between voluntary and mandatory disclosure period as the Companies Act was enacted in 2013. Moreover, GHG emission data is barely available for UK firms before the year 2007. The effect of GHG emission on financial performance, reflected in firm value and cost of capital, is characterised by both cross-sectional and time series elements. However, the event study, which examines the effect of GHG emission on stock prices, serves as a verification analysis and is based on daily stock prices. As both analysis have different characteristics, the research methods and data used in this paper are twofold.

(15)

15 firms often only set targets or report poor intensity values to give a perception that they actively manage their GHG emissions. Although time consuming, publication dates can be accurately determined, because traded companies use Regulatory News Services (RNS). RNS is a communication channel through which firms make announcements to investors, such as the publication of annual reports or the notice for an annual meeting. In total, 2,761 announcements are identified in which firms report their GHG emission. Subsequently, these announcements are matched with their GHG emission values and stock prices from datastream to split the sample in two groups: high versus low intensity firms. This process will be explained in more detail in the following sections and resulted in a final sample of 2,090 events2.

Table 1. Distribution of panel observations over the years and industries.

Industry 2008 2009 2010 2011 2012 2013 2014 2015 2016 1 11 10 16 13 13 13 12 19 19 126 2 14 10 13 15 15 15 14 14 15 125 3 3 3 4 4 4 5 5 6 9 43 4 12 11 9 12 12 13 13 10 9 101 5 27 25 28 32 34 32 32 36 36 282 6 30 24 28 34 34 33 33 32 32 280 7 11 9 8 11 9 10 11 10 12 91 8 6 3 4 16 16 19 23 25 36 148 9 11 11 12 14 14 9 11 10 12 104 10 5 5 6 6 5 5 5 5 4 46 11 5 5 7 9 9 9 9 12 12 77 135 116 135 166 165 163 168 179 196 1,423 3.2 Variable construction

The dependent variables in this paper are firm value, CoD, CoE and stock price. Firstly, Tobin’s Q is taken as proxy for firm value. Tobin’s Q is often used by researchers, among whom Krüger (2016), who calculates Tobin’s Q as the sum of market value and liabilities divided by total assets. Tobin’s Q has an accounting dimension, but can also be used as stock market performance indicator (Lewandowski, 2017). Secondly, CoD is calculated by dividing interest expense on debt by total long-term debts (Li et al., 2014). This formula is relatively straightforward and is applied in many other studies. Although measuring CoE is more complicated, the implied cost of equity is used by most related studies (see Li et al., 2014; Dhaliwal et al., 2011; Kim et al., 2015). The implied CoE is based on a discount rate that balances present share prices to expected future payoffs (Li et al., 2014). One CoE method, the

2_{158 in industry 1, 176 in industry 2, 76 in industry 3, 124 in industry 4, 363 in industry 5, 397 in industry 6,}

(16)

16 price-earnings-growth (PEG) ratio from Easton (2004), is considered as superior. The PEG ratio is valuable as it separates the effects of the growth and cash flow (García-Sánchez and Noguera-Gámez, 2017). The calculation is displayed in formulas 1, 2 and 3.

Cost of equity = √1/(𝑃𝐸𝐺 𝑥 100) (1)

PEG = P’ E ratio / Annual EPS growth (2)

P’ E ratio = P0 / EPS (3)

P0 reflects the share price at 31 december, because annual EPS growth is the expected EPS for the following year. The one-year growth rate is retrieved from Datastream. The last dependent variable, stock price, is based on excess returns on the daily stock price, where excess returns are calculated by (𝑃𝑟𝑖𝑐𝑒𝑡 - 𝑃𝑟𝑖𝑐𝑒𝑡−1)/𝑃𝑟𝑖𝑐𝑒𝑡−1.

(17)

17 observations are classified as low carbon-intensive, 686 as high carbon intensive and 51 firm-year observations emit their industry mean.

Regarding the event study, firms have different year end dates and announcements are unevenly distributed throughout the year. Therefore GHG emission relative to industry peers is not a good proxy, nor is the lagged industry mean due to big differences over time. For this reason, a firm’s carbon intensity is measured as GHG emission divided by annual sales, as in Chapple et al. (2013) and Griffin (2017). Subsequently, based on GHG emission to sales ratio’s, the sample is split in two groups; low carbon intensive firms which have a lower ratio than the median ratio within their industry over all years and high carbon intensive firms which have a higher ratio than their industry peers over the years 2007 untill 2016. Accordingly, a dummy is made for 1,042 firm-year observations which are classified as low carbon intensive. Furthermore, 1,042 firm-year observations are high carbon intensive and 6 emit their industry median. Figure 1 displays the median GHG emission to sales ratio’s over the different years and industries. There is little variation over time, except for industry 9, expecially in the first two years. Industry 9 encompasses, among other sectors, metal mining, beverages, wholesale medical trade, retail clothing trade and health services.This could be due to improvements in GHG emission measurements over time relative to the beginning years. Often companies make adjustments for previous years because they have enhanced their carbon measurement capabilities.

Figure 1. Median GHG emission to sales ratio’s over the years and industries.

0.000 0.200 0.400 0.600 0.800 1.000 1.200 1.400 1.600 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

GHG emission to sales ratio

(18)

(19)

19

Table 2. Descriptive statistics of firm-level variables in the panel regression analysis.

In the table, summary statistics of firm-level variables are presented. The sample period runs from 2008-2016. TQ = Tobin’s Q, CoD = Cost of debt, CoE = Cost of equity. GHG emissions are the annual total GHG emissions as CO2 equivalents in tonnes. Financial variables are measured in British pounds. Net income is expressed in million units.

Variable Mean Median Min Max SD N

TQ 3.09 2.29 0.73 10.4 2.49 1,423 CoD 0.077 0.056 0.021 0.380 0.079 1,423 CoE 0.139 0.119 0.014 0.370 0.096 1,423 GHG emissions 1217 109 1.69 11580 2907 1,423 Foreign holdings 8.1 5 0 83 13.7 1,423 LN(assets) 15.18 14.84 11.63 21.44 1.69 1,423 Leverage 0.26 0.24 0.00 1.73 0.18 1,423 Capital expenditures 0.07 0.03 0.00 0.41 0.10 1,416 Net income 477,550 156,900 24,036 3,010,812 773,336 1,423

Table 3. Descriptive statistics of variables in the panel regression analysis per sub-group.

Low-intensity firms Industry medians High-intensity firms

Variable Mean Median Mean Median Mean Median

N = 686 N = 51 N = 686 TQ 3.29 2.55 2.90 2.07 3.24 2.36 CoD 0.087 0.056 0.067 0.055 0.071 0.055 CoE 0.133 0.111 0.144 0.124 0.140 0.115 GHG emission 77 28 2426 489 296 68 Foreign holdings 7.28 0 8.82 5 11.04 0 LN(assets) 14.31 14.19 16.05 15.73 15.18 15.07 Leverage 0.249 0.227 0.262 0.244 0.246 0.235 CAPEX 0.060 0.024 0.090 0.045 0.064 0.030

Table 3 shows a breakdown of descriptive statistics for both low carbon intensive and high carbon intensive firms. Firms defined as low (high) carbon intensive have, on average, a slightly higher (lower) firm value compared to the overall sample, which is in line with the expectations. Moreover, the cost of equity is lower (higher) for low (high) carbon intensive firms. However, the opposite is true for the cost of debt, which contradicts the predictions made in section 2. Significance is tested in the chapter results.

(20)

20

3.3 Research method

As discussed, the data and research methods in this paper are twofold. Whereas a unbalanced panel regression analysis is needed for nearly all hypothesis, the event study methodology is used to test hypothesis 1b.

3.3.1 Panel regression analysis

Regarding the cross-sectional analysis, regressions are run in Stata. From table 1, the dataset is an unbalanced panel as some firms have fewer observations or observations at different times to others. Hence a pooled ordinary least square (OLS) approach is applied to estimate the effect of GHG emission on financial performance. According the Hausman test (P-value < 0.05 for all models) a fixed effect model is preferred over a random effect model. This demonstrates that the random effect model does not sufficiently model for unobservable firm characteristics that might correlate with the explanatory variable (Lewandowski, 2017). Per hypothesis, two regressions are run. The first regression addresses the full sample and the second regression shows the effect of the regulation with respect to the previous period. In this way, the relative difference between the mandatory and voluntary disclosure environment can be examined.

TQit = βo + β1LowGHGIntensity + β2YEAR + β3SIZE + β3LEV + β4CAPEX + uit (4)

TQit = βo + β1LowGHGIntensity + β2PostReg + β3SIZE + β3LEV + β4CAPEX + uit (5)

Formulas 4 and 5 are used to test hypothesis 1a. TQit is Tobin’s Q in year t . Low GHG intensity is a lagged dummy variable marking all firm-year observations with less GHG emission than the median within the industry, in a particular year. Year is a dummy variable to capture time-related effects. ‘PostReg’ is a dummy variable which equals one for years 2013-2016. Uit denotes the entity fixed effect. Because I expect that firms reporting less GHG emissions relative to their industry peers experience a higher firm value, the coefficient (β1) is probably positive. Moreover, the effect of GHG emission on firm value is supposedly less strong after 2013. Therefore the coefficient (β1) in the fifth equation, which shows the effect of the Companies Act, is likely less high than in formula 4.

CoDit = βo + β1LowGHGIntensity + β2YEAR + β3SIZE + β3LEV + β4CAPEX + uit (6)

CoDit = βo + β1LowGHGIntensity + β2PostReg + β3SIZE + β3LEV + β4CAPEX + uit (7)

CoEit = βo + β1LowGHGIntensity + β2YEAR + β3SIZE + β3LEV + β4CAPEX + uit (8)

(21)

21 Formulas 6,7, 8 and 9 test hypotheses 2a and 2b respectively. CoD is the cost of debt in year t and CoE is the cost of equity in year t. According to previous literature, firms with lower carbon emission intensities have lower uncertainty regarding future cash flows. As a result, they are rewarded by the market by a lower cost of capital. Hence the coefficient (β1) is likely negative, with an even more negative coefficient in equations 6 and 8.

TQit = βo + β1LowGHGIntensity + β2Foreignholdings + β3LowGHGIntensity ×

Foreignholdings + β4YEAR + β5SIZE + β6LEV + β7CAPEX + uit (10)

TQit = βo + β1LowGHGIntensity + β2Foreignholdings + β3LowGHGIntensity ×

Foreignholdings + β4PostReg + β5SIZE + β6LEV + β7CAPEX + uit (11)

CoDit = βo + β1LowGHGIntensity + β2Foreignholdings + β3LowGHGIntensity ×

CoEit = βo + β1LowGHGIntensity + β2Foreignholdings + β3LowGHGIntensity ×

Equations 10 to 15 are an extension to the previous discussed regressions, but add the independent variable foreign holdings and the interaction effects between foreign holdings and the low GHG intensity dummy. The signs and magnitudes of the coefficients as described above are likely strenghtened by a higher proportion of foreign ownership.

3.3.2 Event study methodology

(22)

22 The constant mean return model is used for determining the abnormal returns. Although relatively simple, the constant mean return model often generates outcomes identical to those of more complicated models (MacKinlay, 1997).

Firstly, excess returns are calculated daily and then averaged over the 88 days preceding a disclosure. Secondly, abnormal returns are the market returns on days of the disclosure. Abnormal returns are the actual returns over the event window (day -1 to day 1) minus the normal return over the estimation window. By calculating abnormal returns, the effect of GHG emission disclosure is isolated from other activities on the market. Thirdly, to interpret the results and draw conclusions, abnormal returns are summed up and across firms and time. I do this for both groups, low-intensity and high-intensity firms. As discussed in section 3.2, carbon intensity is measured as GHG emission divided by annual sales and the groups are seperated accordingly, by industry. The expectation is that low-intensity firms experience higher abnormal returns than high-intensity firms. Fourthly, to measure the final impact, singular abnormal returns are summed up over the three day event window, leading to cumulative abnormal returns.

CARi = ∑𝑇2_𝑡=𝑇1+1𝐴𝑅̂it (16) To determine statistical significance and therefore which test to use, I determine whether abnormal returns have a normal distribution or not by conducting the Jarque-Bera (JB) test. Results can be seen in table 4 and with large JB values, the null hypothesis of a normal distribution can be rejected, indicating a non-normally distribution of abnormal returns. Hence a non-parametric test is necessary. An elaboration on the calculations and their corresponding formulas can be found in the appendix.

In table 4 it can be seen that the cumulative abnormal returns for low carbon intensive firms are 0.05% on the day of announcement and -0.19% on one day prior and one day after the dat of announcement (day -1 untill 1). Altough table 4 presents summary statistics, they give an indication which is not in line with the prediction, namely that low carbon intensive fims experience a higher stock price on the days surrounding their annual GHG emission

(23)

23

Table 4. Descriptive statistics of the event study.

In the table, summary statistics of the event study observations are presented. The sample period runs from 2007-2016. E(R) is the mean return over the estimation window of 88 days. AR are the abnormal returns and are the actual returns over the event window day minus the normal return over the estimation window. CAR are the cumulative abnormal returns and are retrieved by summing up all singular abnormal returns over the event window. Jarque-Bera values determine whether abnormal returns have a normal distribution or not. A detailed overview of the formulas can be found in table 9 of the Appendix.

Total announcements Low-intensity firms High-intensity firms

N = 2,090 N = 1,042 N = 1,042 E(R) 0.08% 0.08% 0.07% AR (0) 0.10% 0.13% 0.08% AR (it) 0.02% 0.05% 0.00% CAR (-1, 1) -0.12% -0.19% -0.06% Skewness 0.85 0.80 0.92 Kurtosis 10.25 13.50 6.62 Jarque-Bera 4,709 4,899 608 P-value 0.000 0.000 0.000

Mean Median Mean Median Mean Median

GHG emission 2333.21 77.71 598.28 28.11 4046.32 261.42

Sales 7793.91 1300.85 6803.74 1267.73 8696.12 1369.00

GHG intensity 0.36 0.05 0.03 0.02 0.69 0.14

To test for significane, the Corrado test is appriopriate when only a single day is tested rather than an event window consisting of multiple days. Corrado ranks the returns on the event day (day 0) relative to all returns in both the the estimation window and event window.

𝜃_𝑐 = 1 𝑁∑ (𝐾𝑖0− 𝐿1+𝐿2 2 ) 𝑁 𝑖=1 /𝑠(𝐾) (17) 𝑠(𝐾) = √ 1 (𝐿1+𝐿2)∑ ( 1 𝑁∑ (𝐾𝑖𝜏− 1 𝐿1+𝐿2) 𝑁 𝑖=1 ) 2 𝑁 𝜏=𝑇0+1 (18)

Where L1 is the estimation window (88 days), L2 the event day (day 0) and K the mean rank of

the announcement within the estimation window and event day. However, this nonparametric test is normally used in combination with parametric counterparts (MacKinlay, 1997). Moreover, for an event study with more than 50 events, one may assume a normal distribution of returns (Brown and Warner, 1985). Therefore a parametric statistical test is applied and formula 19 used to arrive at the t-statistic.

𝜃1 =

𝐶𝐴𝑅𝜏

(24)

24

4. Results

In this section, results of the panel regression analysis and event study are discussed. Furthermore, issues regarding endogeneity are addressed in section 4.3.

4.1 Panel regression analysis

Results for the unbalanced panel regression analysis with firm-fixed effects are presented in table 5. In model 1a, hypothesis 1a is tested, in model 2a hypothesis 2a, in model 2b hypothesis 2b, etcetera. Outcomes are partially in line with the hypotheses. Firstly, the results indicate that firms with low GHG intensities have a higher firm value. More specifically, firms that emit less carbon than the median within their industry are valued 0.270 times higher by the market than their peers who emit the same or more than that industry median. Although the effect tends towards significance at a 5% level, the result is only statistically significant at the 10% level (p = 0.070). Moreover, firms with low GHG intensities after the introduction of the Companies

Table 5

Unbalanced panel regression analysis with firm-fixed effects

Variable (1a) Tobin’s Q (2a) Cost of debt

Full sample Post regulation Full sample Post regulation Low GHG intensity 0.270* 0.269* -0.009 -0.009 (0.149) (0.153) (0.006) (0.006) Foreign holdings GHG emission × Foreign holdings Post regulation 0.724*** -0.014*** (0.090) (0.003) Size -1.006*** -0.768*** -0.023*** -0.031*** (0.185) (0.172) (0.007) (0.007) Leverage 0.177 -0.050 -0.206*** -0.177*** (0.664) (0.662) (0.026) (0.026) Capital expenditures -0.764 -0.751 -0.105** -0.096 (1.106) (1.133) (0.044) (0.044) Constant 18.052*** 14.345*** 0.510*** 0.614*** (2.752) (2.588) (0.109) (0.101)

Year dummies Yes No Yes No

Industry dummies No No No No

R² overall 0.037 0.034 0.152 0.125

(25)

25

Table 5 (continued)

Unbalanced panel regression analysis with firm-fixed effects

Variable (2b) Cost of equity (3a) Tobin’s Q (3b) Cost of debt (3c) Cost of equity

Full sample Post regulation Full sample Post regulation Full sample Post regulation Full sample Post regulation

Low GHG intensity -0.002 -0.002 0.161 0.196 -0.006 -0.007 -0.007 -0.007 (0.008) (0.008) (0.172) (0.177) (0.007) (0.007) (0.010) (0.010) Foreign holdings -0.015* -0.006 0.001* 0.000 -0.000 -0.000 (0.009) (0.009) (0.000) (0.000) (0.000) (0.000) GHG emission × Foreign holdings 0.010 (0.008) 0.007 (0.008) -0.000 (0.000) -0.000 (0.000) 0.000 (0.000) 0.000 (0.000) Post regulation -0.012** 0.732*** -0.015*** -0.011** (0.005) (0.091) (0.004) (0.005) Size -0.002 -0.011 -1.038*** -0.773*** -0.021*** -0.031*** -0.003 -0.012 (0.010) (0.010) (0.185) (0.172) (0.007) (0.007) (0.011) (0.010) Leverage 0.097** 0.114*** 0.319 0.027 -0.210*** -0.180*** 0.103*** 0.119*** (0.038) (0.037) (0.671) (0.668) 0.027 (0.026) (0.038) (0.037) Capital expenditures -0.031 -0.028 -0.798 -0.788 -0.105** -0.095** -0.033 -0.031 (0.063) (0.063) (1.107) (1.136) (0.044) (0.044) (0.063) (0.063) Constant 0.171 0.287** 18.627*** 14.466*** 0.488*** 0.604*** 0.189 0.297** (0.157) (0.144) (2.772) (2.597) (0.110) (0.101) (0.158) (0.144)

Year dummies Yes No Yes No Yes No Yes No

Industry dummies No No No No No No No No

R² overall 0.005 0.001 0.040 0.035 0.155 0.127 0.006 0.000

Observations 1,416 1,416 1,416 1,416 1,416 1,416 1,416 1,416

(26)

26 Act in 2013, have a 0.724 times higher firm value than firms with low GHG intensities before 2013. This contradicts the expectation that the negative effect of GHG emission on firm value, for low-intensity firms, is stronger before 2013. Control variables LEV and CAPEX are insignificant. However, the coefficient for firm size is negative (β = -1.006) and significant at the 5% level, which underlines the theory that large firms seem to be more affected by environmental regulation as they tend to emit more.

Secondly, although the coefficient for lowcarbon intensive firms on CoD is negative (β = -0.009), thus demonstrating a positive association between GHG emission and CoD, there is no significant association (p = 0.130). On the other hand, control variables SIZE, LEV and CAPEX are significantly and negatively correlated with CoD. This is consistent with the assumption that larger firms are perceived as less risky, but conflicts the notion that firms with more debt suffer from unfavorable borrowing conditions. Overall, where previous literature was in congruence that higher levels of disclosures would generally lead to more favorable borrowing conditions, because increased transparency leads to lower uncertainty among investors, hypothesis 2a cannot be supported.

(27)

27 Fourthly, hypothesis 3 tests the moderating effect of foreign ownership on the relationship between GHG intensity and financial performance, reflected in firm value and cost of capital. As can be seen in model 3a, the coefficient of low-carbon intensive firms on TQ is positive (β

= 0.161), as in hypothesis 1a, but insignificant. The coefficient of foreign holdings on TQ is

negative (β= -0.015) and significant at the 10% level. This suggests that firms with a higher share of foreign holdings have a lower firm value. However, the interation effect of low GHG intensity and foreign holdings on firm value is insignificant (p = 0.236), which means that firms with low GHG intensities and foreign owners do not have a higher firm value.

Fifthly, and similar to insignificance in its main relationship, no support is found for hypothesis 3b and hypothesis 3c, indicating no moderating effect of foreign holdings on the relationship between GHG intensity and cost of capital. There are several alternative explanations for why the results differ from the expectations. Firstly, as ownership share increases, shareholders gain more power to promote CSR activities, but also bear more of the costs (Barnea and Rubin, 2010). Secondly, most literature on the moderating effect of foreign ownership focuses on emerging countries. As developed country, the UK might already be more environmental conscious. Therefore the effect could be stronger when UK investors target developing countries instead of vice versa. Thirdly, I have not taken into account investors’ country of origin. Foreign owners from different countries might differ in their profile and preferences in terms of favoring environmental investments.

(28)

28 debate on carbon emissions among listed firms was only initiated recently and understanding of potential investment implications is still incomplete, reaction on the market can be delayed (Li et al., 2014). Thirdly, disclosures can be affected by a country’s legal and financial systems (Francis and Khurana, 2005). As the paper focuses on one country, it does not address cross-country issues. For example, environmental regulation and therefore regulatory costs might be less predictable in less developed countries. Fourthly, I focus on GHG performance relative to industry peers, where GHG emission to sales ratio’s or GHG improvements over time are often used. Recently, Lewandoski (2017) and Trumpp and Guenther (2017) propose a U-shaped relationship between environmental performance and financial performance, thereby pointing out the need for more investigation on a complex relationship. Finally, inferences on pre versus post regulation capital effects should be made with caution. Capital market effects are not necessarily attributable to regulation adoption (Leuz and Wysocki, 2016). For example, the Companies Act created more peers and therefore the effect might be stronger in industries where fewer firms have previously reported. Moreover, as most studies focus on individual firms, they do not account for externalities or market-wide effects that emerge over time (Leuz and Wysocki, 2016). One discernible effect is the financial crisis in 2008, which is the first year in my sample period. It is conceivable that, over the succeeding years, as the economy grew, firms’ financial performance increased. Future research could take into account these external effects.

4.2 Event study

(29)

29

Table 6. AAR and CAAR based the constant mean return model

High-intensity firms Low-intensity firms

N = 1,042 N = 1,042

CAR (0) CAR (-1, 1) CAR (0) CAR (-1, 1)

Corrado

Average rank 45 45

Average rank event 44 46

Test statistic 1.049 0.412 P-value 0.192 0.172 T-test AR 0.00% -0.06% 0.05% -0.19% AAR ϴ 0.011 -0.445 0.694 -1.626 P-value 0.459 0.672 0.244 0.948

On the other hand, carbon intensive firms encounter cumulative abnormal returns of 0.00% and -0.06% one the day of the announcement and on one day surrounding the announcement respectively. However, the results are not significant, neither for one day event windows nor for three day event windows. As a result, hypothesis 1b cannot be supported and hypothesis 1a cannot be reinforced. An attainable reason is that even if markets are perfectly efficient (Fama, 1970) and CSR engagement is often perceived as beneficial, stock markets may not value CSR investments properly because it is difficult to determine how a specific social initiative will pay off as the future is uncertain (Oh et al, 2011). Due to uncertain future payoffs, short-term investors may consider environmental investments as risky and not value GHG emission disclosures appropriately, whereas long-term investors tend to encourage environmental responsibility, and low GHG intensities in particular, as is found in section 4.1.

4.3 Endogeneity

(30)

30 valued more in a timespan of three days, cannot suddenly emit less. For this reason, the event study deals with endogeneity in some way. While long-term value effects are observed (at the 10% significance level), insignificant event study results raise doubt on the direction of the relationship. Therefore, in the paper’s context, I cannot eliminate the possibility that the level of firm value was indeed caused by previous GHG intensities.

(31)

31

5. Conclusion

Today, little is known about the consequences of mandatory carbon disclosure on financial performance. With environmental problems that are increasingly global and markets that incorporate non-financial information disclosures, the issue that emerges is how carbon emissions specifically impact management and investment decisions. I address this issue by answering the question whether firms can financially benefit from lowering their greenhouse gas (GHG) emissions. More specifically, the purpose of this paper is to investigate the potential impact of GHG emissions on financial performance, measured by firm value and cost of capital, and the moderating effect of foreign ownership.

I do so by studying firms listed on the FTSE350 over the years 2008 until 2016, as the U.K. was the first country to introduce mandatory disclosure on GHG emissions for publicly listed firms. The regulation became operational in 2013 and lends itself for a unique comparison between pre and post mandatory regulation effects. The effect of GHG emissions on financial performance, reflected in firm value and cost of capital, is analyzed by a panel regression analysis using 1,416 firm-year obervations. Subsequently, in an event study, the short-tem value effects are investigated. Using 2,090 hand-collected disclosure dates, I investigate how daily stock prices react to firms’ disclosing their annual GHG emission.

(32)

(33)

33

References

Andersson, M., Bolton, P., & Samama, F. (2016). Hedging climate risk. Financial Analysts

Journal, 72(3), 13-32.

Bansal, R., Kiku, D., and Ochoa, M. (2015). Climate change and growth risks. Working paper, Duke University, Durham, NC.

Barnea, A., & Rubin, A. (2010). Corporate social responsibility as a conflict between shareholders. Journal of Business Ethics, 97(1), 71-86.

Brown, S. J., Warner, J. B. (1985). Measuring security price performance. Journal of finance

and economics, vol. 8(3), 205-258.

Busch, T., & Hoffmann, V. (2011). How hot is your bottom line? linking carbon and financial performance. Business & Society, 50(2), 233-265.

Chapple, L., P. Clarkson, and D. Gold. (2013). The cost of carbon: Capital market effects of the proposed emission trading scheme (ETS). Abacus, 49 (1): 1–33.

Clarkson, P., Li, Y., & Richardson, G. (2004). The market valuation of environmental capital expenditures by pulp and paper companies. The Accounting Review, 79(2), 329-353. Clarkson, P. M., Li, Y., Richardson, G. D. and Vasvari. F. P. (2011). Does it really pay to be green? Determinants and consequences of proactive environmental strategies. Journal

of Accounting and Public Policy, 30, 122-144.

Clarkson, P., Li, Y., Pinnuck, M., & Richardson, G. (2015). The valuation relevance of greenhouse gas emissions under the European union carbon emissions trading scheme. European Accounting Review, 24(3), 551-580.

Dam, L., & Scholtens, B. (2012). Does ownership type matter for corporate social responsibility? Corporate Governance, 20(3), 233-252.

Dell, M., Jones, B., & Olken, B. A. (2009). Temperature and income: Reconciling new cross- sectional and panel estimates. American Economic Review, 99(2): 198–204.

Dell, M., Jones, B., & Olken, B. A. (2014). What do we learn from the weather? The new climate-economy literature. Journal of Economic Literature, 52(3): 740–798.

Dhaliwal, D., O. Z. Li, A. Tsang, and Y. G. Yang. (2011). Voluntary nonfinancial disclosure and the cost of equity capital: The initiation of corporate social responsibility reporting. The Accounting Review, 86 (1): 59–100.

Easley, D., & O'hara, M. (2004). Information and the cost of capital. The Journal of Finance, 59(4), 1553-1583.

Easton, P. (2004). PE ratios, PEG ratios, and estimating the implied expected rate of return on equity capital. The Accounting Review, 79(1), 73-95.

Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. The

Journal of Finance, 25, 383.

(34)

34 García-Sánchez, I., & Noguera-Gámez, L. (2017). Integrated information and the cost of capital. International Business Review, 26(5), 959-975.

Griffin, P., Lont, D., & Sun, E. (2017). The relevance to investors of greenhouse gas emission disclosures. Contemporary Accounting Research, 34(2), 1265-1297.

Haniffa, R., & Cooke, T. (2005). The impact of culture and governance on corporate social reporting. Journal of Accounting and Public Policy, 24(5), 391-430.

Hoskisson, R., Hitt, M., Johnson, R., & Grossman, W. (2002). Conflicting voices: The effects of institutional ownership heterogeneity and internal governance on corporate innovation strategies. The Academy of Management Journal, 45(4), 697-716.

Johnson, R., & Greening, D. (1999). The effects of corporate governance and institutional ownership types on corporate social performance. Academy of Management Journal,

42(5), 564-576.

Jung J, Herbohn K, Clarkson P. 2014. The impact of a firm’s carbon risk profile on the cost of debt capital: evidence from Australian firms. Working Paper, University of Queensland, Australia.

Kim, Y., An, H., & Kim, J. (2015). The effect of carbon risk on the cost of equity capital.

Journal of Cleaner Production, 93, 279-287.

King, A., & Lenox, M. (2002). Exploring the locus of profitable pollution reduction. Management Science, 48(2), 289-299.

Kleimeier S, Viehs M. 2016. Carbon disclosure, emission levels, and the cost of debt. Working Paper, Maastricht University, The Netherlands.

Kruger (2016). Climate Change and Firm Valuation: Evidence from a Quasi-Natural Experiment. Working paper, University of Geneva, Switzerland.

Lambert, R., Leuz, C., & Verrecchia, R. (2012). Information asymmetry, information precision, and the cost of capital. Review of Finance, 16(1), 1-29.

Leftwich, R., Watts, R., & Zimmerman, J. (1981). Voluntary corporate disclosure: The case of interim reporting. Journal of Accounting Research, 19, 50-50.

Leuz, C., Lins, K. V., & Warnock, F. E. (2010). Do foreigners invest less in poorly governed firms? Review of Financial Studies, 23(3), 3245–3285.

Leuz, C., & Wysocki, P. (2016). The economics of disclosure and financial reporting regulation: Evidence and suggestions for future research. Journal of Accounting

Research, 54(2), 525-622.

Lewandowski, S. (2017). Corporate carbon and financial performance: The role of emission reductions. Business Strategy and the Environment, 26(8), 1196-1211.

Li, Y., Eddie, I., & Liu, J. (2014). Carbon emissions and the cost of capital: Australian evidence. Review of Accounting and Finance, 13(4), 400-420.

MacKinlay, A.C. (1997). Event studies in Economics and Finance. Journal of Economic

Literature. 35(1): 13-39.

(35)

35 NPR. (2015). Nearly 200 nations adopt climate agreement at COP21 talks in Paris. Retrieved

from http://www. npr.org/sections/thetwo-way/2015/12/12/459464621/final-draft-of- world-climate-agreement-goes-to-a-votein-paris-saturday.

Oh, W., Chang, Y., & Martynov, A. (2011). The effect of ownership structure on corporate social responsibility: Empirical evidence from Korea. Journal of Business Ethics, 104(2), 283-297.

Palmer, K., Oates, W., & Portney, P. (1995). Tightening environmental standards: The benefit-Cost or the no-Cost paradigm? Journal of Economic Perspectives, 9(4), 119 132.

Pinkse, J., & Kolk, A. (2010). Challenges and trade-offs in corporate innovation for climate change. Business Strategy and the Environment, 19(4), 261-272.

Porter, M., & Linde, C. (1995). Toward a new conception of the environment competitiveness relationship. Journal of Economic Perspectives, 9(4) Fall 1995: 97-118.

Sharfman, M., & Fernando, C. (2008). Environmental risk management and the cost of capital.

Strategic Management Journal, 29(6), 569-592

Soliman, M., Din, M., Sakr, A., 2012. Ownership structure and corporate social responsibility (CSR): an empirical study of the listed companies in Egypt. International Journal of

Social Sciences. 5(1), 63–74.

Trumpp, C., & Guenther, T. (2017). Too little or too much? exploring u-shaped relationships between corporate environmental performance and corporate financial performance.

Business Strategy and the Environment, 26(1), 49-68.

Verisk Maplecroft (2015). Heat stress threatens to cut labour productivity in SE Asia by up to 25% within 30 years. Report, Climate Change and Environmental Risk Analytics (CCERA), Verisk Maplecroft, Bath.

(36)

(37)

37

Industry classifications

Table 7. Industry classifications.

This table shows industry classifications based on the (SIC) code list of the U.S. Securities and Exchange Commission (SEC). SIC codes reflect the company’s type of business. A more detailed overview can found on https://www.sec.gov/info/edgar/siccodes.htm.

Industry Industry title

1 Medicinal chemicals & botanical products, pharmaceutical preparations, insurance services.

2 Clothing, watches and jewelry, electric, gas and sanitary services, wholesale furniture and grocery trade, retail home supply trade.

3 Computers and electronic equipment, computer services. 4 Oil and gas, food and kindred products.

5 Agriculture, forestry and fishing, tobacco products, motor vehicles, automotive serivces, aircrafts and parts.

6 Construction, metal and steel works, engineering services, plastic materials, cosmetics.

7 Banks and finance services. 8 Real estate and investment trusts.

9 Metal mining, mining & quarrying of nonmetallic minerals, beverages, wholesale medical trade, retail clothing trade, health services.

10 Engines and turbines, (industrial) machinery and equipment, wholesale electrical parts and equipment.

(38)

38

Correlation Matrix

Table 8

Correlation Coefficients

TQ CoD CoE GHG Foreign holdings SIZE LEV CAPEX NI

(39)

39

Formulas event study

Table 9. Event study formulas.

Formula Explanation

𝑅̅it = 1

88∑ µ 𝐿1 −90 i

Excess returns are calculated daily and then averaged over the 88 days preceding a disclosure. µi is the mean of returns for stock i

over the estimation window and 𝑅̅it is its normal

return.

ARit = Rit − 𝑅̅it

Abnormal returns are the actual returns over the event window day (-1 to day 1) minus the normal return over the estimation window (𝑅̅it).

Where ARit and Rit, are the abnormal and actual

returns for day t. 𝐴𝑅̂it = 1

𝑁 ∑ 𝐴𝑅 𝑁 𝑖=1 it

Abnormal returns are summed up and across firms and time

CARi = ∑𝑇2_𝑡=𝑇1+1𝐴𝑅̂it

Singular abnormal returns are summed up over the three day event window, leading to cumulative abnormal returns

JB = 𝑛

6

(

𝑆 2₊1

4 (K – 3)²