Do political connections affect corporate
environmental performance?
A panel data analysis of the influence of connected Boards over the
environmental and financial performance of firms using UK data.
Author: S.B.M Renoult Supervisor: Dr. Homroy, Swarnodeep University of Groningen
Faculty of Economics and Business Business Administration, MSc Finance
Student Number: S3399133 RENOULT Sonia
2
C
ONTENTS Abstract ... 3 1. Introduction ... 4 2. Literature review ... 8 3. Methodology ... 13 3.1 Estimation strategy... 13 3.2 Multicollinearity ... 14 3.3 Model specification ... 14 4. Data ... 174.1 Accounting and GHG emission data ... 17
4.2 Political affiliation ... 17
4.3 Board Composition ... 18
4.4 Industry dummy ... 18
4.5 Summary statistics... 18
5. Results ... 20
5.1 T-test statistics results (independent samples) ... 20
5.2 CO2 emission results (Model 1 and 2)... 21
5.3 Determinants of ROA ... 23 5.4 Green Bonds ... 25 6. Conclusion ... 27 7. References ... 29 8. Appendix ... 32 8.1 Definition Variables ... 32
8.2 Variance Inflation factor ... 32
3
A
BSTRACTGiven the growing debate over corporate governance and environmental investments, this study examines the relationship between political connections and corporate environmental performance. In order to do so this study analyses different political environment measures to determine the connected firms and analyse those connection effects on greenhouse gas (GHG) emissions. Politics may have a significant influence on connected firms’ behaviour due to their advisory role but also mainly from their networks and political connections. Particularly, the thesis stresses the importance of studies linking corporate governance and environmental performance. It also contributes to the existing literature regarding firm’s financial performance and political connectedness. Using greenhouse gas (GHG) emissions data from FTSE 350 firms, as a proxy for the environmental performance of firms, we show that the presence of political directors or connected CEOs, increase the level of GHG emissions and reduce the profitability of the firm (ROA). In this thesis, political connection appears clearly to have a negative influence over the environmental and financial performance of the firms. However, the political connection is only found to be significant at explaining the level of CO2 emissions when the political party of the connected firms was in power.
Keywords: Political connection; greenhouse gas emissions; environmental performance; politically connected;
profitability; UK
4
1. I
NTRODUCTIONOver the last few years, interest in the protection of the environment has grown and the environmental impact of companies have been brought to light. Climate change has come to dominate the public debate on the environment unlike any other issue today. It could explain why it became more than an environmental issue and shifted to a security matter. This trend also results from an increasing attention regarding environmental sustainability and corporate responsibility (Stern, 2006). Firms are investing more and more into greener business manufacture to transition their economic production. We have also noticed this evolution through sustainable responsible investing (SRI) for instance, which has developed and reached a record by rising by 10 percent in 2017 (Omaha World-Herald, 2018). Corporate social responsibility (CSR) is also becoming a significant field for investors, $3.07 trillion of professionally managed US assets are tied to CSR through socially responsible investing (SRI) (Di Giuli and Kostovetsky, 2014). Private politics and corporate social responsibility have a direct effect on the costs of the firm (Baron, 2001). Stakeholders are also highlighting the importance of ESG (environmental social and corporate governance) factors and how important it is for the company to implement a management promoting sustainable business activities to avoid negative publicity but also technology rendered obsolete by stricter legislation.
5
expect companies to reduce their carbon footprint? As climate change and environment issues have been an increasing concern this last decade, it seems rational to expect politics and directors of companies to promote and act toward more ethical and sustainable approach. Various studies (Fisman, 2001; Faccio, 2006; Hutton et al., 2013) demonstrated the important influence of politicians over firm’s behaviour. A political director plays an important resource and advisory role inside the board, which may then lead connected firms to reduce their negative environmental exposure. Previous studies have shown the importance of board characteristics and especially director’s advisory role in the firm’s environmental performance (de Villiers, Naiker and van Staden, 2011). Therefore, the presence of a politician in the board of a company might have a positive impact on the environmental performance of companies because politicians rather seek for positive opinions from the public by promoting ‘’greener’’ actions among companies. Additionally, some papers mentioned the cost of ‘’acting green’’ (Hong and Kostovetsky, 2012; Di Giuli and Kostovetsky, 2014) showing that ‘’goodness is costly’’. Therefore, a connected board of directors might also influence in a positive way, the environmental performance of the company by providing additional support regarding the implementation cost of an environmental policy. From this background, we could then expect a reduction of CO2 emissions when there is a political director presents in the board of the firm. This will represent a positive aspect of corporate governance over companies’ environmental performance. However, politically connected directors might also be able to bypass environmental regulations thanks to their networks (Bunkanwanicha and Wiwattanakantang, 2009). In contrast, connected firms in this situation would display a higher level of GHG emissions since they could be covered thanks to their political connections. Assuming that greenhouse gas emissions are an outcome of board’s decision-making process with regard to political connection would provide more insight into the link between corporate governance and firm’s environmental policy. Nevertheless, political influence on firms depends obviously on the region. As (Faccio, 2006) paper mentioned, Malaysia, Ireland, Russia, Thailand and the United Kingdom account for more than 20% of the market capitalization. In the United Kingdom, connected firms actually represent 39.02 percent. Political connections between company directors and politicians are valuable for UK companies. It is one of the main reason, the studied sample has been on UK firms. (Geys and Mause, 2013) mentioned in their paper that 49% of the Member parliaments in UK House of Commons, have extra-parliamentary activities which are paid.
6
7
By examining political affiliation through political donation and check for the presence of the political directors in the company’s board, it serves as the measure for political connection. GHG emissions are the equivalent of CO2 emissions and have been used as a measure for the environmental performance of firms. It has been then regressed on the political connection (PC) and other firm’s characteristics mentioned in the Data section.1 From there on, the same PC variable has been used as a dummy variable for the other test regressions and for testing the political effect on firm’s financial productivity (ROA). The results show that political connection have no significant impacts on the financial and environmental performance. However, when testing a sample of only connected firms, when the political party of the firm is in power (PowerParty), we see a positive significant effect upon the CO2 emissions level and an adverse effect on ROA. The results are robust to several different specifications. The PC variable remains insignificant when testing on high polluting industries sample while PowerParty variable is still significant and positive.2
The remainder of the paper is organized as follows. The next section reviews the literature on the relationship among corporate governance, environmental performance with different measures than GHG emissions, firm’s performance, and connected firms. It will also include the research objectives and hypotheses of this paper. Section 3 pursues with the Methodology, and the different regressions that have been performed and Section 4 notes the sample and data. Finally, section 5 presents the results on the relation between political connection and greenhouse gas emissions issued by companies as well as companies issuing green bonds. It also displays the effect on the profitability of the firms (ROA) and their correlation. To conclude this master thesis, the last section concludes with a summary and the limitations.
8
2. L
ITERATURE REVIEWThe literature review provides an overview of the existing theoretical and empirical concepts. The different research hypotheses are also going to be stated in this section throughout the literature overview.
In the last years, we assisted to different political and corporate scandals regarding the environment such as the Volkswagen emissions scandal. There has been a continuous attention in corporate governance and socially responsible investing (SRI) as globalization brought changes in the structure of business management and regulations. ‘US firms spent more than $120 billion in 1994 to comply with environmental laws’ (Konar and Cohen, 2001b). Environmental protection has become part of a company’s budget cost. New environmental restrictions put in place by governments and institutions are also going to support this transition and encourage firms to act towards the environment. All the different incentives are put in place to incentivize further the firms to improve their environmental and ethical behaviour. One of the key paper that I would also like to extend in this area is the study of (Di Giuli and Kostovetsky, 2014) where they found that firms with high Kinder, Lydenburg, Domini (KLD) score are mainly the ones managed by Democratic CEOs or Directors.3 Most studies use the KLD score as a proxy for environmental performance, but in this study, greenhouse gas emission of companies will be taken as an alternative. In the empirical study of (Hutton et al., 2013), the researchers support the current conclusions by showing that having managerial conservatism, most likely to be put in place by Republican political-side CEOs, can cause firms to adopt conservative policies. They tested these conjectures by using personal political contributions of corporate managers to identify their political orientation and degree of financial conservatism.
Thus, from the growing importance of corporate governance over the last 20 years, a connection could be made with corporate environmental performance. As it has been mentioned in the introduction, politicians face pressure to promote ‘’good’’ behaviour to firms, especially regarding sustainability. Some papers in the political science literature (Roberts, 1990; Chen et al., 2014) showed that political connections matter and that it had even an influence on share prices of companies. Although, large UK companies also voluntarily made environmental disclosures. Environmental disclosures have been viewed as a way to avoid adverse regulatory or legislative pressures in the future (Brammer and Pavelin, 2008). Depending on the type of the company, its financial health or industry characteristics,
3 They examined the political side of companies to find a connection with the environmental score (KLD). I will
9
companies have different patterns in terms of environmental disclosure (Patten, 2002; Cormier and Magnan, 2003) and into their decisions of reducing their CO2 emissions. In (Brammer and Pavelin, 2008) paper, they found that larger and less indebted companies with dispersed ownership characteristics are significantly more likely to make voluntarily environmental disclosures. Political connectedness may bring a positive contribution in an ecological perspective since it might generate a positive opinion about their political behaviour. Therefore, one might conclude that a politically connected director inside the board of the companies will instead incite the firms at reducing its GHG emissions.
But is there a counterargument that might also explain an opposite outcome when testing PC (political connection) significance over GHG emissions? Because one way for political connectedness to bring benefits to companies is through rent-seeking activities (Chen
et al., 2014). Connected firms could derive some advantages from rent-seeking activities
thanks to political networks with the politicians currently in power. For instance, connected firms might be actually able to bypass environmental regulations thanks to their connection with the current government in place. Studies have suggested that companies are still exercising political pressure by affecting regulatory changes in relation to social and environmental issue through active lobbying, membership in advisory committees and other traditional political channels (den Hond et al., 2014; Frynas and Stephens, 2015). In this case, political connections will not end up bringing positive effect on the environmental performance of the company, but even a negative impact as the company will not feel pressured to comply to environmental regulations if supported by politicians and will continue issuing high level of CO2 emissions. So, we can see from this overview, two different aspects of how political connection can affect corporate environmental performance. One is positive by reducing the greenhouse gas emissions of the company which would then show connected firms having a lower CO2 emission in the result. And the other one negative which, therefore, will end up underline in the result a significant and negative coefficient for political connection when testing hypothesis 1:
H1. Political connections have no explanatory value in predicting greenhouse gas (GHG)
emissions.
10
of this study to investigate whether having its political party in power changed the significance of the environmental performance of the firm. A paper in the political science literature (Roberts, 1990) showed that political connections matter and that it had even an influence on share prices of companies. Connected firms are more likely to get benefit from the government as it was appointed in (Faccio, Masulis and Mcconnell, 2006) study where it compared non-connected companies and connected companies. They also found that connected firms are more likely to be bailed out by the government in case of financial distress. As demonstrated by some previous studies such as (Hong and Kostovetsky, 2012), there are differences in environmental exposure whether a company or a fund is managed by a Democrat or Republican manager. In this paper, the political side will not be tested since it evaluates a sample of UK companies. The second tested model is a similar regression than the first model, except that instead of examining simply if the company is politically connected, Model 2 will only consider a sample of connected firms to test the significance of political connection when the political party of the company is in power. Because as we saw previously, a politician on the board of the company has been found to influence the managerial financial decisions of the firm. Another study (Saeed, Belghitar and Clark, 2016) also reveals that firms that have a politician on its board of directors are going to affect the managerial financial decisions of the company. Again, these effects escalate with the strength of the connected politician and whether the party is currently in power (Saeed, Belghitar and Clark, 2016). Theories reviewed in this literature review appears to predict that political affiliation may have an effect on firm’s decision-making behaviour. (Bonaparte, Kumar and Page, 2017) study shows that people’s optimism towards financial markets and the macroeconomy is also affected by political affiliation and the current political climate. Investors will perceive less risk when their preferred party is in power, which leads them to invest differently compared to what they would have done if it was the opposite party in power. Value of companies increases when their executives enter politics (Faccio 2006). And because of the growing importance of environmental issues, the board of directors is more likely to influence the company’s environmental policy which makes board composition an important element to analyse as political environment measure.
H2. There is no statistical significance that firms’ political party when in power, affects GHG
emissions level.
11
connectedness from firm contributions to politicians also appears to be important for firm value (Cooper, Gulen and Ovtchinnikov, 2010). In (Goldman, Rocholl and So, 2009) paper, they demonstrate that firms connected to the US Republican party increased in value after the Republican win in the 2000 Presidential elections. Therefore, one might expect the political connection to enhance the company’s value when it is in power but also when simply politically connected via firms’ contributions. A company’s board can add value via the connections it provides with politicians or either by monitoring and advising management (Adams and Ferreira, 2007). In most cases, connected firms are associated with better financial performance (Hillman, 2005; Faccio, Masulis and Mcconnell, 2006). However, some studies also mention the harmful effect that could bring political connection. Study of (Fisman, 2001) shows that during a period of negative rumours about Suharto’s health, connected firms lost more value during this period that did less-dependent firms. In (Fan, Wong and Zhang, 2007) paper, they investigated 790 partially privatized firms in China and found that politically connected CEOs underperform those without connected CEOs.4 Because firms managed by a politically connected CEO are more likely to appoint directors based on their political ties than on relevant professional backgrounds. Also, it is important to consider that other factors might impact a company value. Environmental value can also affects the market value of a company (Konar and Cohen, 2001b). Then, the last tested models will contribute to the existing literature concerning the effect of director/CEO connections and observe which effect it has on the financial performance of companies.
H3. There is no statistical significance of political connection on firm’s return on assets.
Financial literature provides various studies that examine investment decisions and the different issues associated with such decisions inside firms. Majority of the studies focused on analysing the sensitivity of the investment decision to the availability of cash flow. However, we can see in the paper of (López-Gutiérrez, Sanfilippo-Azofra and Torre-Olmo, 2015) that there are more factors that affect the relationship between investment and cash flow. In this paper, they show how financial distress firms behave differently regarding the sensitivity to cash flow and investment decision. A similar approach will be studied by analysing the relationship of all the different variables through independent t-test when firms have a politically connected CEO and/or director in the board to examine how they behave in both situations. From the previously studied background, we might expect a relationship between political connection and companies operating in polluting industry. Then, together with Hypothesis 3, an interaction effect would be performed between polluting industry (HighGHG) and political connection (PC) on the profitability of the firm (ROA) to test whether there is a
12
connection between the two explanatory variables and which effect it might have on the profitability of the firm.
H4. There is no interaction effect between Political connection (PC) and High polluting industry
(HighGHG) upon the profitability of the company (ROA).
13
3. M
ETHODOLOGYThose different research aims have been studied to be part of an increasing literature regarding corporate governance and sustainable investment and bring a focus on how it might affect the environmental and financial performance of firms.
3.1
E
STIMATION STRATEGYWe obtain a panel data and run regressions for this thesis by collecting different samples. It contains observations of multiple time periods for the same firms.
First, I performed a cross-check with the first sample from the FTSE 350 index and the political data to determine which companies are politically connected via political donation. I used past political donation from 2008 to 2017 from the Elective Commission website. If a company donated to any political party during this period, it was set equal to 1 and 0 when no political donation found. It reduced then my sample from 350 companies to 35. Thus, I extended the political environment measure to the presence of political directors inside the company. In addition, I cross-checked for names of member of parliaments with the names from Board composition of all the 350 companies. Those data were found on Boardex. Those variables have been used in previous studies testing for political environment impact on firm performance. Since ‘corporate boards have the power to make, or at least ratify, all important decisions including decisions about investment policy, management compensation policy, and board governance itself’ (Bhagat and Bolton, 2008). Therefore, those variables are relevant to include in the test regression and were able to complete the political environment measure and made it more accurate. There are many ways to create a political connection. In this thesis, we are focusing on campaign contributions which are publicly available data in the UK, and the presence of politicians on the board of directors. The political environment measure (PC) includes, then, 2 sub measures:
- Political donation – dummy variable set to 1 when the company donated a certain
amount to a political party and 0 if no political donation found.
- Board director – dummy variable that will take 1 if a firm has at least one board
director politically connected, and 0 otherwise.
And an additional extent to account for the political party in power is going to be used to test the second model:
- Political party in power (Powerparty)– dummy variable set to 1 whether the political
14
3.2
M
ULTICOLLINEARITYWhen performing a regression analysis, it is important to look at possible collinearities between the different explanatory variables as it could influence the results. Table 1 gives a correlation matrix for all the considered variables in this paper:
Table 1. Correlation Matrix
As expected, there is a high correlation between Powerparty and politically connected variable. Highly correlated variables, above 0.7, should be taken separately into the regression analysis (Brooks, 2008). For this reason, PC and Powerparty dummies have both been tested separately when regressing the models. For most of the other variables, the results show that they have relatively small correlations. Except for Revenues and Size which display significant correlations between them. This high correlation between both variables is intuitively logical since both variables proxy firm size. Similar reasoning regarding Cash and Revenues. Since
Revenues variable is highly correlated with other independent variables, it has then been
removed when testing the models. Therefore, to assess potential multicollinearity in the model, collinearity test is also performed.5
3.3
M
ODEL SPECIFICATIONPolitical connection and GHG emissions
The first equation used to test the formulated H1 hypothesis is related to the equation applied by (López-Gutiérrez, Sanfilippo-Azofra and Torre-Olmo, 2015), their empirical method consisted of testing the influence of the existence of financial distress on the investment behaviour of companies. Thus, I tested a similar regression while focusing on environmental performance and political connection instead. Together with the different independent t-tests
5 See Table 8 in the Appendix for Variance inflation factor (VIF) value.
15
outcome, it will answer one the main question underlined in this thesis: Are connected firms increasing or reducing their environmental exposure (GHG emissions)? The model is estimated using panel data methodology which allows controlling for unobservable heterogeneity. If the null hypothesis H1 is rejected, the coefficient β1 is statistically different from zero and measures the influence of the political environment on CO2 gas emission issue by companies: Following regression test Model 1:
𝐶𝑂2𝑒𝑚𝑖𝑠𝑠𝑖𝑜𝑛𝑖𝑡 = 𝛽0 + 𝛽1𝑃𝐶𝑖𝑡 + 𝛽2𝑅𝑂𝐴𝑖𝑡 + 𝛽3𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽4𝐶𝐴𝑆𝐻𝑖𝑡 + 𝛽5𝐷𝐸𝐵𝑇𝑖𝑡 + ∑Indsutryit + ԑit
Additionally, I controlled the industry of each company. Whether the company was performing in a high polluting industry, the variable will be set equal to 1 and 0 otherwise. This dummy was created by attributing 1 to companies which issued more than 100 000 tons of CO2 emission per year. As it was mentioned above, the European government threshold is 100 million Kg per year. If below this, it is set equal to 0. 32% of my company sample were considered to perform in a polluting industry. Although, the firm-fixed effect has not been included in the regressions because PC is a time-invariant variable.
Second tested model is focused only on connected firms to analyse further the association between the environmental performance of companies with their political connections. However, in this regression, I considered whether the political party of the firm was in power during the studied period: Powerparty variable, in order to test the null hypothesis H2:
𝐶𝑂2𝑒𝑚𝑖𝑠𝑠𝑖𝑜𝑛𝑖𝑡 = 𝛽0 + 𝛽1𝑃𝑜𝑤𝑒𝑟𝑝𝑎𝑟𝑡𝑦𝑖𝑡 + 𝛽2𝑅𝑂𝐴𝑖𝑡 + 𝛽3𝑆𝐼𝑍𝐸𝑖𝑡 + 𝛽4𝐶𝐴𝑆𝐻𝑖𝑡 + 𝛽5𝐷𝐸𝐵𝑇𝑖𝑡 + ∑Industryit + ԑit
Firm’s performance and its political affiliation
Modelling the third null hypothesis H3, I use a similar approach from (Di Giuli and Kostovetsky, 2014), but instead of using a degree of financial conservatism, I perform a test on the return on assets (ROA) as a proxy for the productivity of the firm. Since the previous literature demonstrated the impact of political connection over firms’ performance, one would expect to find PC significant. The next two models are regressing the different explanatory variables on ROA as a dependent variable (Model 3):
16
this test, I created a similar variable, HighGHG, which is the industry level sets to 1, referring to HighGHG in this equation. The purpose was to analyse the interaction between those independent variables which may affect the return on assets of the companies. I also controlled for the different firm’s characteristics (Model 4):
17
4. D
ATAFour separate types of data were acquired for this thesis: 1) GHG emissions and other firm controls variables of the companies traded on the London Stock exchange from the FTSE 350 index; 2) data on group affiliation of all FTSE 350 firms; 3) List of the Board composition of all the firms and list of the Members parliament for each parliament mandate
4.1
A
CCOUNTING ANDGHG
EMISSION DATAThe sample consists of an unbalanced panel of the 350 largest companies by capitalization primarily listed on the London Stock Exchange (FTSE 350) from 2008 and 2017. The different companies’ data were taken from Datastream on an annual basis. In (López-Gutiérrez, Sanfilippo-Azofra and Torre-Olmo, 2015) paper, they mentioned various explanatory variables, which were proved by previous studies, to influence firm’s investment behaviour. And as we want to analyse the effect of political connection to firm’s decision-making, specifically regarding their corporate environmental approach, most of those variables have been considered. The variables are internal funds generated for each firm (CF); firm’s size (SIZE); the firm’s leverage (LEV); and the firm’s industry (SECTOR). Those factors have been included in the panel data because they are considered endogenous and have a strong influence on the investment decisions. The remaining variables are the return on assets (ROA), to display the firms’ profitability and Net sales (Revenues) as an indicator of growth and size of the companies. See Appendix table 7
Greenhouse gas emissions (GHG) is the main dependent variable in this study, which represents the gas emission that emits companies in the atmosphere and has been selected to represent the environmental performance of the firms. It has been found through Datastream under “ENERDP123” code that reports the annual total GHG emissions as CO2 equivalents. This variable also helped in determining which industry were the most polluting. GHG emission variable has been chosen to examine the role of politics in their degree of emission.
4.2
P
OLITICAL AFFILIATION18
companies in order to observe which companies are politically connected. From this first selection, it reduced my unbalanced panel of 351 companies to 35 companies.
4.3
B
OARDC
OMPOSITIONAfter selecting across political donation, the Board composition list of the same 350 companies has been taken as a second measure of political connectedness to increase the number of connected firms’ observations. I would check whether the firm has a politically connected director who is a member of the Board of director of the company. Additionally, I checked for names in the Board composition list to match every name which is also part of the parliament composition at the corresponding period. It added six additional connected companies to the sample. BoardEx is one of the sources of data where we can find information on firm CEOs, directors, and founders. It also provides detail information about education, employment history, role at the firm. Whenever I cannot find the corresponding name with a political connection or party, I set the dummy variable to 0.
A second variable has been created from this data information: Powerparty. Companies which were recognized politically connected and where it was possible to affiliate a political party to the firm, has been matched depending on the time to another data table providing the political party in power. Whenever the political party of the connected firm was also the one in power in the specified period, the dummy variable has been set to 1 and 0 otherwise. The
Powerparty variable has been applied to a sample of 362 observations.
4.4
I
NDUSTRY DUMMYLast dummy variable was Industry level which refers to the different sectors of companies. The level of CO2 emission that a company could emit is strongly influenced by which industry the company is performing in. Therefore, a dummy has been created in order to take into account the industry more likely to pollute. An industry group emitting higher than 100 million Kg/year were set to 1 and set to 0 when lower. In order to give a label on which industries were the most polluting, this value came from (Homroy and Slechten, 2017) study which reported the threshold set up by the European Commission regarding the impact on human health and on the environment.
4.5
S
UMMARY STATISTICS19
Table 2: Summary statistics (2008-2017)
Variable N Mean SD Min Max
Panel A: Firm-control variables
ROA 3,319 7.134806 7.0148 -3.16 23.95 Size 3,319 4710176 8792415 0 3.58e+07 Cash 3,319 298927 547774.2 0 2165000 Revenues 3,319 3851335 7048866 0 2.71e+07 Debt leverage 3,121 .7635207 .9150093 0 3.318146 CO2 emission 3,319 1785.452 8467.842 0 86000 Green bond 11 278.7 200.9 4 600
Panel B: Political environment
Politicallyconnected
(Dummy) 3,320 .1093373 .312109 0 1
Industry level (Dummy) 3,320 .3210843 .4669635 0 1
Power party (Dummy) 362 .6132597 .4876774 0 1
20
5.
R
ESULTSThis section will discuss the results derived from the dataset after performing the panel data analysis. For model 1 and 2, CO2emission is the dependent variable. First, the results regarding the political connection of firms as a main independent variable will be presented. Model 2 will focus on the PowerParty significance dummy and connected firms. Model 3 presents the results regarding the estimation regression on ROA as dependent variable and model 4 includes the interaction terms between PC and HighGHG on ROA. All models are presented with heteroscedasticity robust standard errors. In addition, a small part will be attributed to Green bonds and their political connection. However, because of the lack of data, it will be a qualitative description. The regression outcomes will be presented in table 4 and table 5. Table 3 will display the outcomes from the independent t-tests by political connection.
5.1
T-
TEST STATISTICS RESULTS(
INDEPENDENT SAMPLES)
In this section, independent t-tests statistics have been performed for each variable on the political connection (PC) to examine every relationship between them.
Table 3. T-test results
Not PC PC M SD M SD t-test ROA 7.374 7.105 5.181 5.877 5.6478*** Size 4712706 8905113 4689574 7825519 0.0473 Revenues 3680555 6855627 5242038 8341416 -3.9920*** Cash 287999.6 543403.2 387911.5 575243.1 -3.2844*** Debt .7241909 .8994625 1.084157 .976993 -6.9074*** CO2emission 595334.2 1634739 621106.6 1586342 -0.2844
Table 4: Results of the regression model relating all firm’s control variables with political and industry dummies. Regression coefficients and standard errors are included in the table. *, ** and *** mark statistically significant coefficients at the 10%, 5% and 1% level respectively.Note: M=Mean. SD=Standard Deviation.
21
non-connected firms (not PC). Regarding the debt leverage, the results are consistent with (Saeed, Belghitar and Clark, 2016) which found companies having a politician on its board of directors are highly leveraged and use more long-term debt. And here we can see that connected firms have higher debt ratio. Revenues appear to be bigger for connected firms which are in line with the study of (Fisman and Wang, 2010), politically connected firms are larger in terms of sales. However, the profitability of the firms diminishes when they have a political connection as the mean for ROA is higher for firms without political connections. Which contradicts the analysis of this same paper (Fisman and Wang, 2010) saying that connected firms are more profitable. This divergence of result may be explained by the different study area since this paper investigated data from China. Looking at the global corruption perceptions index released by Transparency International lists, China is ranked 77th while UK is ranked 8th which refers to the UK as a far less corrupted country than China. Therefore, a political connection may not benefit the profitability of the firm as much as it did for companies in this area. Such benefits are usually greater when the firms operate in a corrupted country or with a non-democratic government (Faccio, 2006). Another possible explanation might be that determining most political connections via campaign contributions does not represent a strong enough association with politicians to obtain significant financial benefits in this case.
5.2
CO2
EMISSION RESULTS(M
ODEL1
AND2)
In this section, the panel data regression results will be presented. The regressions have been performed with Stata. As there is a presence of heteroscedasticity (see Appendix), robust standard errors have been reported. In both model 1 and model 2, there is binary independent variables, Industry, PC, and Powerparty.
22
Table 4: Panel data estimates on CO2 emission (dependent variable)
Variable Model 1 Model 2
(restricted sample) Constant -44.60 (38.529) -520.91** (207.798) PC(Dummy) -67.87 (80.848) PowerParty (Dummy) 508.37** (204.324) ROA -9.16*** (3.02) -37.10* (19.888) Size -0.00003*** (0.000) 0.00003 (0.000) Cash 0.0005*** (0.000) 0.001*** (0.000) Debt -38.1 (32.79) 38.61 (108.678) Industry (Dummy) 1064.7*** (75.089) 1741.9*** (264.564) Number of obs 3,121 341 R-squared 0.2704 0.3026
Table 4: Results of the regression model relating all firm’s control variables with political and industry dummies. Regression coefficients and standard errors are included in the table. *, ** and *** mark statistically significant coefficients at the 10%, 5% and 1% level respectively. Robust standard errors are in parentheses. In addition, the table provides the R-squared and the number of observations.
23
(Elsayed, 2007) which found that firm size and available resources are significant predictors of corporate environmental performance. The result of ROA negative coefficient is also consistent with the finding of (Fisher-Vanden and Thorburn, 2011) which reported that companies reducing greenhouse gas emissions appear to conflict with firm value maximization. The negative coefficient of Size variable makes sense since a more prominent company is more likely to produce higher greenhouse gas emissions than a small company. Model 2 concerns a restricted sample as the estimated regression includes only politically connected firms. From this result, we can see that Powerparty variable is statistically significant which means that we can reject the null hypothesis H2 that there is no explanatory power between Powerparty and CO2 emission level. This result is in line with the previously mentioned literature, which stated that companies are benefiting from political connection when this one is in power. Government is more likely to support the companies having the same political affiliation which is then going to influence the firm’s behaviour. However, it does not support one of the first assumption made in this thesis, regarding the fact that connected boards should push companies to further reduce their CO2 emissions. We get the negative aspect discussed previously regarding the harmful impact of political connection on firms’ environmental performance. A possible explanation is that politics may help bypass environmental laws and regulations, therefore these companies are not tempted to reduce their GHG emissions because they are not threatened by government’s sanctions. Industry dummy like in the first model is statistically significant and shows an important positive coefficient since a polluting industry is more likely to issue a higher level of GHG emission.
5.3
D
ETERMINANTS OFROA
24
Table 5: Panel data estimates on ROA (dependent variable)
Variable Model 3 Model 4
(Interaction terms included) Constant 9.16*** (0.201) 9.28*** (0.204) PC(Dummy) -1.608*** (0.356) -2.68*** (0.481) HighGHG -0.861*** (0.244) -1.21*** (0.263) HighGHG*PC 1 1 2.94*** (0.672) Size 9.82e-08*** (0.000) 1.03e-07*** (0.000) Cash -1.99e-06*** (0.000) -2.04e-06*** (0.000) Debt -1.36*** (0.134) -1.36*** (0.133) Number of obs 3,121 3,121 R-squared 0.0650 0.0690
Table 5: Results of the regression model relating all firm’s control variables with political and industry dummies. Regression coefficients and standard errors are included in the table. *, ** and *** mark statistically significant coefficients at the 10%, 5% and 1% level respectively. Robust standard errors are in parentheses. In addition, the table provides the R-squared and the number of observations.
At first glance, we notice that in model 3, all variables are significant. When looking at our main explanatory variable, PC, we can reject the null hypothesis H3, which means that political connection does have significant explanatory power upon the return on assets (ROA). In addition, it shows a negative coefficient indicating that connected firms will be more likely to display lower ROA (similar finding than the t-test result). Which is in line with some studies arguing that politically connected managers may have deleterious effect instead of benefiting firm performance (Fan, Wong and Zhang, 2007). This outcome could be explained by the different incentives and misallocation of investment (Ang, Ding and Thong, 2013). From this result, another regression was then performed6 on PowerParty variable in order to observe whether the productivity of the company is only impacted when its political party is in power. The result demonstrates that Powerparty is statistically insignificant. Therefore, we cannot conclude that having a connected board director when its political party is in power increases
6 See Table 11 in the Appendix. It shows a similar regression than Model 3 except that it considers only connected
25
the firm’s ROA, even if it displays a positive coefficient in this case. Additionally, HighGHG has been found significant and shows a negative coefficient so it appears that polluting companies have a lower financial performance. This result is in line with (Fisher-Vanden and Thorburn, 2011) paper which says that companies could gain potential benefits by improving environmental performance. For example, this increase in profit could come from a differentiation of product that can increase customer demand and hence increase the sales and/or margins. It also prevents companies to get into a future environmental liabilities and lawsuits that might damage their revenues.
Then, the interaction term between PC and HighGHG has been included in the regression. By including the interaction term in the model, it allows capturing the relation that changes based on the value of another variable. The statistical significance of the term shows that there is an interaction effect between those two variables. As we can see, the coefficient of both explanatory variables is negative. When the industry is polluting and not politically connected it decreases the ROA of the firm. Same if the firm is politically connected and not performing in a polluting industry, it negatively impacts the profitability of the firm. As we already observed in Model 3, firms with lower GHG emissions have then higher return on assets (ROA) while connected firms display a lower productivity. However, we see that interacting connected firms (PC=1) with polluting industry (HighGHG=1), the coefficient becomes positive which means that the interaction between the two manages to counteract their adverse effect on ROA when combined. Then, polluting firms manage to perform well when they are politically connected at the same time. It is consistent with the previous finding in model 2, where Power party variable was found to be significant and positive when testing the CO2 emissions level. So, companies that are polluting have not been affected by their financial performance when they do have political ties with the government in power. It is in line with previous findings (Faccio, 2006; Bunkanwanicha and Wiwattanakantang, 2009; Fisman and Wang, 2010) that demonstrate the positive effect that brings political connectedness and networks to the company.
5.4
G
REENB
ONDS26
socially responsible investing (SRI) mutual funds outperformed conventional funds (Margolis, Hillary and Walsh, 2009). However, due to lack of data for UK companies, this aspect will not be further tested in this thesis.
Afterward, I wanted to extend my area of study to Green bonds and companies issuing it. However, the collected data has been limited and lead to analyse only a small sample of 10 UK companies.
Table 6. Green bonds (United Kingdom)
Company Politically connected
Unilever (UK) No
Belectric Solar No
Renewi No
HSBC Bank No
Anglian Water Not determined
SSE No
Barclays No
Bazaglette Finance Not determined
27
6.
C
ONCLUSIONEnvironmental issues became an important concern in our society and it is legitimate that firms have been pressured to reduce their negative environmental impact. Now, it is more likely that the board of directors implement ‘greener’ business practices among the firm. Therefore, the role of politically connected director and/or CEO has been in the interest of this paper, which examines the relationship between corporate governance and environmental performance of UK companies. The analysis is made by a sample of around 350 large UK companies drawn from a diverse range of 82 industrial sectors. Even if the two estimation regressions have the same dependent variable and similar explanatory variables, the study distinguishes both due to the different content of the sample and its measure. The first regression includes the full sample with 3 321 observations and the second regression contains only the connected firms which represents 363 observations. I created different sub-measures that account for the degree of the political environment of the firm. It accounts for the political contributions and for the presence of a politician or a member of the parliament in the Board of directors. I found that companies having a political director at the Board or being politically affiliated are not a significant measure to determine the level of GHG emission. Although, connected firms display higher CO2 emissions compare to non-connected companies from the 350 companies sample. Therefore, one of the first assumptions regarding the fact that political director should help companies to reduce CO2 emissions appears to be disproved by the results. However, the political connection (PC) was found significant only when the political party of the connected firms was in power. It supports the previous findings on this topic, political affiliation does have an influence on firm’s behaviour when this one is in the government. Connected board of directors or CEOs are more likely to apply government recommendations when this one is the same political party affiliated to the firm.
28
29
7.
R
EFERENCESAbbott, K. W. and Snidal, D. (1998) ‘Why states act through formal international organizations’, Journal of Conflict Resolution, 42(1), pp. 3–32. doi: 10.1177/0022002798042001001.
Adams, R. B. and Ferreira, D. (2007) ‘A theory of friendly boards’, Journal of Finance, 62(1), pp. 217– 250. doi: 10.1111/j.1540-6261.2007.01206.x.
Ang, J. S., Ding, D. K. and Thong, T. Y. (2013) ‘Political connection and firm value’, Asian Development Review, 30(2), pp. 131–166. doi: 10.1162/ADEV_a_00018.
Baron, D. P. (2001) ‘Private Politics, Corporate Social Responsibility, and Integrated Strategy’, Journal of Economics & Management Strategy, 10(1), pp. 7–45. doi: 10.1162/105864001300122548.
Bernstein, S. and Cashore, B. (2007) ‘Can non-state global governance be legitimate? An analytical framework’, Regulation & Governance, 1(4), pp. 347–371. doi: 10.1111/j.1748-5991.2007.00021.x. Bhagat, S. and Bolton, B. (2008) ‘Corporate governance and firm performance’, Journal of Corporate Finance, 14(3), pp. 257–273. doi: 10.1016/j.jcorpfin.2008.03.006.
Bonaparte, Y., Kumar, A. and Page, J. K. (2017) Political climate, optimism, and investment decisions, Journal of Financial Markets. doi: 10.1016/j.finmar.2017.05.002.
Brammer, S. and Pavelin, S. (2008) ‘Factors influencing the quality of corporate environmental disclosure’, Business Strategy and the Environment, 17(2), pp. 120–136. doi: 10.1002/bse.506. Brooks, C. (2008) Introductory Econometrics for Finance 2nd edition, Cambridge. doi: 10.1017/CBO9781107415324.004.
Bunkanwanicha, P. and Wiwattanakantang, Y. (2009) ‘Big business owners in politics’, Review of Financial Studies, 22(6), pp. 2133–2168. doi: 10.1093/rfs/hhn083.
Chen, C. M. et al. (2014) ‘Does a firm’s political connection to government have economic value?’, Journal of the Asia Pacific Economy, 19(1), pp. 1–24. doi: 10.1080/13547860.2013.860761.
Climate Bonds Initiative (2017) ‘Climate Bonds Standard’, Climate Bonds Standard, Version 2.1, p. 16. Cooper, M. J., Gulen, H. and Ovtchinnikov, A. V. (2010) ‘Corporate political contributions and stock returns’, Journal of Finance, 65(2), pp. 687–724. doi: 10.1111/j.1540-6261.2009.01548.x.
Cormier, D. and Magnan, M. (2003) ‘Environmental reporting management: A continenal European perspective’, Journal of Accounting and Public Policy, pp. 43–62. doi: 10.1016/S0278-4254(02)00085-6.
Elsayed, K. (2007) ‘Does CEO duality really affect corporate performance?’, Corporate Governance: An International Review, 15(6), pp. 1203–1214. doi: 10.1111/j.1467-8683.2007.00641.x.
Faccio, M. (2006) ‘Politically connected firms’, American Economic Review, 96(1), pp. 369–386. doi: 10.1257/000282806776157704.
Faccio, M., Masulis, R. W. and Mcconnell, J. J. (2006) ‘Political connections and corporate bailouts’, Journal of Finance, 61(6), pp. 2597–2635. doi: 10.1111/j.1540-6261.2006.01000.x.
30
Fisher-Vanden, K. and Thorburn, K. S. (2011) ‘Voluntary corporate environmental initiatives and shareholder wealth’, Journal of Environmental Economics and Management, 62(3), pp. 430–445. doi: 10.1016/j.jeem.2011.04.003.
Fisman, R. (2001) ‘Estimating the value of political connections’, American Economic Review, 91(4), pp. 1095–1102. doi: 10.1257/aer.91.4.1095.
Fisman, R. and Wang, Y. (2010) ‘The mortality cost of political connections’, Review of Economic Studies, 82(4), pp. 1346–1382. doi: 10.1093/restud/rdv020.
Frynas, J. G. and Stephens, S. (2015) ‘Political Corporate Social Responsibility: Reviewing Theories and Setting New Agendas’, International Journal of Management Reviews, 17(4), pp. 483–509. doi: 10.1111/ijmr.12049.
Geys, B. and Mause, K. (2013) ‘Moonlighting Politicians: A Survey and Research Agenda’, Journal of Legislative Studies, 19(1), pp. 76–97. doi: 10.1080/13572334.2013.737158.
Di Giuli, A. and Kostovetsky, L. (2014) ‘Are red or blue companies more likely to go green? Politics and corporate social responsibility’, Journal of Financial Economics. North-Holland, 111(1), pp. 158–180. doi: 10.1016/J.JFINECO.2013.10.002.
Goldman, E., Rocholl, J. and So, J. (2009) ‘Do politically connected boards affect firm value’, Review of Financial Studies, 22(6), pp. 2331–2360. doi: 10.1093/rfs/hhn088.
Hillman, A. J. (2005) ‘Politicians on the board of directors: Do connections affect the bottom line?’, Journal of Management, 31(3), pp. 464–481. doi: 10.1177/0149206304272187.
Homroy, S. and Slechten, A. (2017) ‘Do Board Expertise and Networked Boards Affect Environmental Performance?’, Journal of Business Ethics. Springer Netherlands, (2015), pp. 1–24. doi: 10.1007/s10551-017-3769-y.
den Hond, F. et al. (2014) ‘Playing on Two Chessboards: Reputation Effects between Corporate Social Responsibility (CSR) and Corporate Political Activity (CPA)’, Journal of Management Studies, 51(5), pp. 790–813. doi: 10.1111/joms.12063.
Hong, H. and Kostovetsky, L. (2012) ‘Red and blue investing: Values and finance’, Journal of Financial Economics, 103(1), pp. 1–19. doi: 10.1016/j.jfineco.2011.01.006.
Hutton, I. et al. (2013) ‘Corporate Policies of Republican Managers’, Working Paper.
Konar, S. and Cohen, M. A. (2001a) ‘Does the Market Value Environmental Performance?’, Review of Economics and Statistics, 83(2), pp. 281–289. doi: 10.1162/00346530151143815.
Konar, S. and Cohen, M. A. (2001b) Does the Market Value Environmental Performance?, Review of Economics and Statistics. doi: 10.1162/00346530151143815.
López-Gutiérrez, C., Sanfilippo-Azofra, S. and Torre-Olmo, B. (2015) ‘Investment decisions of companies in financial distress’, BRQ Business Research Quarterly, 18(3), pp. 174–187. doi: 10.1016/j.brq.2014.09.001.
31
Patten, D. M. (2002) ‘The relation between environmental performance and environmental disclosure: A research note’, Accounting, Organizations and Society, 27(8), pp. 763–773. doi: 10.1016/S0361-3682(02)00028-4.
Roberts, B. E. (1990) ‘Political Institutions, Policy Expectations, and the 1980 Election: A Financial Market Perspective’, American Journal of Political Science, 34(2), pp. 289–310. doi: 10.2307/2111448. Saeed, A., Belghitar, Y. and Clark, E. (2016) ‘Do Political Connections Affect Firm Performance? Evidence from a Developing Country’, Emerging Markets Finance and Trade, 52(8), pp. 1876–1891. doi: 10.1080/1540496X.2015.1041845.
Shleifer, A. and Vishny, R. W. (1994) ‘Politicians and Firms’, The Quarterly Journal of Economics, 109(4), pp. 995–1025. doi: 10.2307/2118354.
Schroders 2015. Green Bonds – A Primer. Talking Point
Stern, N. (2006) ‘The Economics of Climate Change’, Stern Review, p. 662. doi: 10.1257/aer.98.2.1. de Villiers, C., Naiker, V. and van Staden, C. J. (2011) ‘The effect of board characteristics on firm environmental performance’, Journal of Management, 37(6), pp. 1636–1663. doi: 10.1177/0149206311411506.
32
8.
A
PPENDIX8.1
D
EFINITIONV
ARIABLESTable 7: Dependent and Independent variables
Variables Definition Measurement
GHG emission Greenhouse gas emission Annual GHG emission as C02
equivalents (in tonnes)
ROA Return on assets Ratio of net income over total book value of asset (%)
Revenues Net Sales Gross sales + other operating
revenue less discounts, returns and allowances
CF Cash Flow (cash available for the normal operations of the
company)
(Book value on equity + Total Debt) – cash
Debt Debt/Equity ratio Ratio of total liabilities over
shareholder’s equity
Size Market capitalization Total market value of a company’s
outstanding shares
Industry level Industry business (Sector) Branch category of the company
PC Politically connected Dummy variable sets to 1 whether the firm was politically connected
and 0 otherwise
Powerparty Political party in power Dummy variable set to 1 whether
the political party of a connected firm was in power during the
studied period.
HighGHG High GHG emission Dummy variables sets to 1
whether the industry level was equal to 1 and 0 otherwise.
8.2
V
ARIANCEI
NFLATION FACTORIn Table 8, the results highlight that there is moderate multicollinearity (VIF between 1 and 4) among variables but not significant as all the VIF (Variance Inflation Factor) values are below 4. As mentioned above, revenues variable has some correlation between other predictors (independent variables) in the model like Size and Cash variables. Which is the reason why
33
Table 8. Variance Inflation Factor test
Variable VIF 1/VIF
PC 2.44 0.410274 Powerparty 2.41 0.415677 Size 3.34 0.299186 Cash 1.84 0.544767 Industry 1.37 0.728970 Debt 1.15 0.869603 ROA 1.07 0.934954 Revenues 4.22 0.237055 Mean VIF | 2.38
8.3
R
OBUSTNESS OF CHECKSThis section shows the different techniques used to check the model robustness. The test performed is the heteroscedasticity test in order to detect whether it is present in the regression model.
After fitting the model, I tested whether it had the presence of heteroscedasticity:
Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho: Constant variance
Variables: fitted values of C02emission chi2(1) = 13090.62
Prob > chi2 = 0.0000
34
It can be explained from a simultaneity bias where the explanatory variables are jointly determined with the dependent variable. In this case, ROA, Revenues, size are deeply linked with the dependent variable CO2emission which lead to unstable results regarding the significance of politically connected variable (PC) when adding further financial variables such as, Revenues, which is also highly correlated with those variables. Therefore, the heteroscedasticity problem should not be regarded as a negative result since the used variables are all correlated together and impact only the standard errors. And to avoid bias standard errors, robust standard errors are used which group heteroscedasticity and therefore control for it.
Determinants CO2 emissions
Table 9: Panel data estimates on CO2 emission
Variable Coeff Std Error
C 2623.142*** 520.029 PC (Dummy) 723.348 867.747 ROA -162.716*** 59.912 Size .000*** 0.00002 Cash .001*** 0.0003 Debt -261.484 237.279 Model statistics Number of obs 1,056 R-squared 0.2108
35
Table 10: Panel data estimates on CO2 emission
Variable Coeff Std Error
C 2545.136*** 517.604 Powerparty (Dummy) 2479.92* 1296.369 ROA -160.735*** 59.541 Size .000*** 0.00002 Cash .001*** 0.0003 Debt -278.508 234.325 Model statistics Number of obs 1,056 R-squared 0.2151
Table 10: Results of the regression model relating all firm’s control variables with Powerparty dummy on subsample restricted by selecting only companies from the most polluting industries. Regression coefficients and robust standard errors are included in the table. *, ** and *** mark statistically significant coefficients at the 10%, 5% and 1% level respectively. In addition, the table provides the R-squared and the number of observations.
Table 11: Panel data estimates on ROA (connected firms)
Variable Coeff Std Error
C 4.254*** 0.775 Powerparty (Dummy) 0.209 0.663 Size .000*** 7.55e-08 Cash -.000** 7.93e-07 Debt 0.034 0.320 Industry (Dummy) 0.273 0.663 Model statistics Number of obs 341 R-squared 0.0.384
