Faculty of Economics and Business
June 2019
Master Thesis Final Version
Mandatory GHG emissions disclosure and
corporate market value: Family vs nonfamily firms
Author: Supervisor:
Geert Sturm Dr. Swarnodeep Homroy
Student ID:
2555948
ABSTRACT
In this article, I estimate the effect of mandatory greenhouse gas (GHG) emissions disclosure on corporate value. Specifically, I examine the effect of family ownership on the disclosure-firm value relationship. Using The Companies Act 2006 Regulations of 2013 in the United Kingdom as an exogenous shock, I find that overall there is no different valuation effect for firms that started to disclose post-regulation versus firms that already disclosed
pre-regulation. Family firms were less likely to publicly disclose pre-regulation and more to likely to do this post-regulation. Restricting the sample size to the 50% largest firms, I find that firms that started to disclose in or after 2011 saw decreases in corporate value. This effect seems to dampen the more family control is present in a firm.
2
I. Introduction
The general consensus on mandatory reporting requirements of firm-specific information is that it allows capital markets to work more efficiently (Greenstone, Oyer, and
Vissing-Jorgensen (2006)). Financial information such as balance sheets and cash flow statements are a good example of this. In contrast, it is more ambiguous if mandatory disclosure of non-financial information is beneficial for firm value (Flöstrand and Ström (2006)). With the issue of climate change emerging more and more in the public debate, pressure on firms increased to address risks and opportunities relating to this phenomenon. In regards to firm value after disclosure of climate change related information, the central line of thought is that investors value such initiatives, especially when the external business environment becomes more climate conscious (Kim and Lyon (2011)). Family firms form an interesting subgroup of firms to examine in this regard. The general thought is that family firms are less willing to disclose climate change related information because they view this as a way of giving away control (Terlaak, Kim and Roh (2018)).
British firms were able to pick up signs of an upcoming government regulation as early as January 2011. From October 1, 2013 onwards, all firms listed on the Main Market of the London Stock Exchange (LSE) were required by law to disclose their Greenhouse Gas (GHG) emission values. Using the introduction of this regulation, I can partly circumvent the issue of endogeneity that usually arises when examining the relationship between firm value and Corporate Social Responsibility (CSR) related data. Without this regulation, the disclosure of emission values could influence the firm value and vice versa, a higher firm value could make firms more likely to disclose emission values. The regulation circumvents this issue because firms are required to respond by disclosing their emission values either publicly or privately. Furthermore, the introduction of this regulation allows for an interesting setting since it did not affect all firms equally. Part of the firms were already disclosing their GHG emission values and thus compliant with the regulation before it even took effect. This allows to divide firms between two (homogeneous in other aspects) groups: the group that was already
complaint with the regulation and the group that was not.
In this article, I examine whether investors valued the mandatory disclosure of GHG
3
different groups for different periods. Using the DID method, the parallel trends assumption is the most important. If this assumption holds and we can rule out any other time-variant
changes and DID is an appropriate method to use in this context. I also explore the cross-sectional and time-series variation of the DID coefficients. In terms of time-series, I show that there are no differences between the two main groups prior to the regulation period. Cross-sectionally, the results show that the valuation effects differ widely among firm size and industry.
This paper is related to certain different literatures. First of all, it is related to the literature that examines the effect of mandatory reporting regulations on valuation (Greenstone, Oyer, and Vissing-Jorgensen (2006)). Secondly it contributes to papers concerned with the financial effects of environmental regulation (Porter and van der Linde (1995), Palmer, Oates and Portney (1995)). Thirdly, it relates to papers examining the link between family firms and disclosure of climate change related data (Chau and Gray (2002), Nekhili, Nagati, and Chtioui
2017), Ali, Chen and Radhakrishnan (2017)).
The rest of the paper is organized as follows. In section II, I will provide a background on The Companies Act 2006 Regulations 2013 that is central to this paper. Section III covers relevant literature that has been published relating to family firms, GHG disclosure and corporate valuation. According to this literature, three hypotheses will be formed. I will discuss the dataset that has been used as the basis of the empirical part of this paper in Section IV. The empirical part where I will work with Difference-in-Differences regressions and perform robustness checks takes up Section V. Section VI concludes and discusses some implications following the results of this paper.
II. Background on The Companies Act 2006 Regulations 2013
4
collecting worldwide corporate climate risk data. Before this regulation, firms had the choice between responding publicly, privately to CDP, and not responding at all.
On 17 January 2011, the first version of ‘Impact Assessment of Options for Company GHG Reporting’ was released. This Impact Assessment is a document that helps the policymaker to think through the consequences of possible government interventions. Basically, this was the first sign for British firms that a government regulation concerning GHG reporting was coming in the future. Following this, on the 19th of June 2012, Deputy Prime Minister Nick Clegg announced in The Guardian that it is the UK’s intent to pass legislation requiring UK quoted companies to disclose GHG emissions in their annual reports. In July 2012, the Department for Environment, Food and Rural Affairs (DEFRA) published the report ‘Measuring and reporting of GHG emissions by UK companies’ in which detailed information about the upcoming regulation could be found. In June 2013, the draft of the legislation was laid before the parliament and it got approved by the House of Commons on July 16, 2013. The Companies Act 2006 Regulations officially took effect from October 1, 2013 onwards.
The legislation requires firms to report their annual CO2 emissions in metric tonnes. Every
company is required to disclose Scope 1 and 2 emissions. Scope 1 emissions are direct
emissions from owned or controlled resources while Scope 2 emissions are indirect emissions from the generation of purchased energy. Next to that, there are Scope 3 emissions: all
indirect emissions that occur in the value chain of the reporting company, including both upstream and downstream emissions. Firms must also report their relative emissions; their emissions in relation to a factor that measures the activities of the respective company. Besides this, firms are required to report the methodologies used to calculate these numbers. Subsequently, the information needs to be reported in the Director’s Report, this is a
document prepared annually by the board of directors under the requirements of UK company law.
III. Literature review
A. Family firms and GHG disclosure
5
who closely identify with their business (Whetten and Mackey (2005)). They see the business as an extension of their family and themselves and consequently are less likely to engage in behaviour that would damage the reputation of the firm. The family reputation for multiple generations is at stake for every decision made (Miller and Le Breton-Miller (2003)). Berrone, Cruz, Gomez-Meija, and Larraza-Kintana (2010) find that family-controlled public firms protect their socioemotional wealth by having a better environmental performance than their nonfamily counterparts. To gauge environmental performance, they weigh each emission by their Human Toxicity Potential (HTP) factor. So according to the aforementioned
literature, family firms seem likely to have good environmental performance. But are family firms eager to disclose this? In terms of disclosing this information to the public, Chau and Gray (2002) show that family firms have little motivation to disclose this information,
especially in absence of mandatory requirements. In excess of mandatory requirements, there is relatively weak demand for public disclosure of information from family-controlled firms. Nekhili et al. (2017) also find that family firms report less information on their environmental duties than do nonfamily firms. In addition, Ali et al. (2007) show that, looking at S&P 500 firms, family firms make fewer disclosures about their corporate governance practices. The main argument brought forward in this paper is that family firms have incentive to reduce the transparency of corporate governance practices to facilitate getting family members on boards without interference from nonfamily shareholders. Terlaak et al. (2018) brings the argument that family firms are less likely to disclose GHG emissions voluntarily. Family firms give priority to maintaining the family’s controlling influence and insuring the firm’s reputation. Based on this, I form hypothesis I:
Hypothesis I: Prior to the regulation, a lower percentage of family firms disclosed their GHG
emission scores compared to nonfamily firms.
6
B. GHG disclosure and valuation
The traditional view regards regulation aimed at reducing the negative impact of firms on the environment as costly for firms. Such regulations force firms to allocate labor and capital to comply and this is detrimental to profits. There exists a tradeoff between social benefits and private costs (Palmer et al. (1995)). Porter and Van Der Linde (1995) challenge this
traditional view and argue that a properly designed regulation does not have to be costly for firms. A correctly designed environmental regulation can trigger innovation that may offset the costs attached to the regulation. The Porter hypothesis recognizes that this innovation brings about improvements in energy and resource efficiency. In line with the Porter
hypothesis, Konar and Cohen (2001) find a negative correlation between bad environmental performance and the value of intangible assets of the firm. Their research relates
environmental performance of firms in the S&P 500 to their financial performance. The results imply that good environmental performance is valued in financial markets. In line with the traditional view on this matter, there is literature that contradicts these findings. Fisher-Vanden and Thorburn (2011) find that corporate commitments to reduce GHG emissions seem to reduce firm value. They track the financial performance of firms that committed to participate in a program targeting reductions of GHG emissions and conclude that it conflicts with firm value maximization.
The potential of the benefits and costs of GHG disclosure differ among firm characteristics. Firstly, disclosure benefits are larger when firm environmental performance is poor or when the firm operates in a polluting industry (Cho and Patten (2007)). Secondly, since larger firms tend to be more exposed, they are under higher pressure for disclosure and thus able to reap more benefits of this (Cormier and Magnan (1997)). Thirdly, firms with greater resource slack are better able to absorb costs associated with disclosing potentially damaging information as well as basic identifying and measuring costs (Brammer and Millington (2006)). All off the aforementioned factors can be used as control variables when estimating the relationship between firm value and GHG disclosure.
7
Kim and Lyon (2011) find that participation in the mandatory Carbon Disclosure Project increases shareholder value when the likelihood of climate change regulations rose. Their research was conducted on the Financial Times Global 500 companies, the 500 largest companies in the world. This leads me to forming hypothesis II:
Hypothesis II: The mandatory GHG emission disclosure had a positive effect on firm
valuation.
An interesting study area that remains is to examine what firm characteristics contribute to a higher valuation after disclosure of the GHG emissions. According to Kim and Lyon (2010), voluntarily disclosed emission scores can be very misleading. Firms participating in voluntary disclosure programs engage in highly selective reporting: in the aggregate, they increase emissions over time but report reductions. Participants tend to be large firms facing strong regulatory pressure. Furthermore, pressure from environmental groups reduces the likelihood of participation, suggesting that firms view the program as a form of greenwash.
Terlaak et al. (2018) investigate the effect of family control on environmental performance disclosure in South Korean firms. They theorize that disclosing environmental performance information weakens the owning family’s control over its firm but that it also generates reputational benefits, this is in line with the tradeoff theory from Palmer et al. (1995). The results of this study indicate that these disclosure propensities are greatest when family control of business group firms is most extensive. Since this research was done on business groups in South-Korea, the results should be interpreted with caution. Nevertheless, there seems to be a U-shaped relationship between family control and the gain from disclosing environmental information. The perceived costs and benefits of giving away this control vary with the degree of family ownership. A higher degree of family ownership is associated with a gain from disclosing environmental performance information. Generally, the higher the degree of family control, the higher the gain from disclosing. This leads me to forming hypothesis III:
Hypothesis III: The gain from disclosure of GHG emissions was biggest for firms with a high
8
IV. Data
A. Sample and summary statistics
To test the impact of the regulation on firm value measured as Tobin’s q, I select a sample of firms that is affected by the regulation. I begin with selecting all the equities of the FTSE 350, which consists of the 350 largest companies by market capitalization with their primary listing on the London Stock Exchange. The FTSE 350 is a combination of the well-known FTSE 100 and FTSE 250. It includes firms originating from all over the world with a large share of firms being of British origin.
I retrieve CO2 (scope 1 + scope 2) emission values, book value of total assets and market
value data from Thomson Reuters’ Eikon database. Market value is measured as the number of shares outstanding multiplied by the share price. Book value of total assets and total market value are measured in British Pounds. Tobin’s q is measured as the ratio of market value over the book value of total assets. GHG emissions are measured in kilotonnes (kt) of CO2. A value of 0 for CO2 emissions implies that either the firm does not report its CO2
emissions but there is the rare case that the CO2 emissions are actually zero. I distinguish
firms that actually have zero CO2 emissions from firms that did not disclose their emission
values (a ‘private’ response) by considering the rest of their emission values. Furthermore, for identification, I download the ISIN code and industry type for each separate firm from Eikon. The dataset covers every year from 2009 to 2018, implying that I start with 3500 firm-year observations. Several missing observations for market value, and hence Tobin’s q, are deleted from the dataset. Outliers in Tobin’s q are removed from the dataset because of their ability to skew results. On top of that, I take the natural logarithm of total firm assets as a proxy for firm size. To observe the difference between longer term effects of the regulation on Tobin’s q and shorter term effects, I create a dummy variable for the years 2009-2014.
This leaves me with 3034 firm-year observations observations for 341 different firms. Furthermore, I add a couple of dummy variables; Treat indicates firms that did not disclose their CO2 values in 2011 and prior to that year but started doing so after 2011, After stands for
9
family control.
[Table I about here]
As can be observed from Table I, the average FTSE 350 company has a market value of £6.48 billion and assets worth £35.7 billion. It has 2,4 million kt of scope 1 and scope 2 CO2
emissions. The sample size of CO2 is only 1,980 because this is the amount of publicly
disclosed firm-year observations. Comparing the mean with the median CO2 emissions, it
appears the distribution is highly skewed.
B. GHG emissions per industry
Table II presents the summary statistics of GHG emissions per industry.
[Table II about here]
Unsurprisingly, the industries Basic Materials, Utilities and Oil and Gas have the highest overall mean emission scores. Surprising is the high standard deviation of GHG emissions in the Industrials sector, suggesting that there are a few active firms with unproportionate emission scores compared to the rest of the firms in that industry.
C. Summary statistics for different subgroups
Table III shows the summary statistics of firms with different levels of family involvement and also for the treatment and the control group.
[Table III about here]
10
the treatment group have substantially higher assets, on average. This is consistent with the notion that larger firms have more pressure on them to disclose data relating to climate change.
D. GHG disclosure for family firms vs nonfamily firms
Table IV shows the amount of firms per emission response to CDP, per year.
[Table IV about here]
This provides an answer to hypothesis I. In the pre-regulation period, a lower percentage of family firms disclosed their emission scores compared to nonfamily firms. Firms with a high level of family control even have a slightly lower percentage of firms that disclose their emission values compared to firms with a moderate level of family control.
Post-regulation, the share of family firms publicly reporting their emissions was generally higher than the share of nonfamily firms doing this. The firms with the highest level of family control seemed to be the most eager to publicly report their emission values post-regulation.
V. Methodology
A. Baseline Analysis
To start with, I create a table outlining the means of firm value for the treatment and control group, family firms and nonfamily firms, before and after the regulation period. Performing t-tests on the differences in these means is a relatively straightforward method to compare the effect of a regulation on a treatment group versus a control group. The results of this are represented in Table V.
[Table V about here]
11
consistent with the fact that the coefficient for Treat, in table VI is positive and statistically significant. Moreover, the table shows that there are no statistically significant differences in the means before the regulation compared to the means after the regulation.
Panel B of Table V shows the same results for the 50% largest firms in terms of market value. Striking is that the means of firm value for the treatment group are significantly higher
compared to the full sample in panel A. The values of the control group are slightly higher. The most important difference between panel A and panel B of Table V is that, for family firms in the post-regulation period, the means of firm value of the treatment group are significantly different from those of the control group. This was not the case in the pre-regulation period, implying that the firm values of the control group and the treatment group reacted differently. Next to that, I observe a significant drop in firm value of nonfamily firms belonging to the treatment group after the regulation. In Panel A, this difference was not significant.
Figure I shows the means of firm value over the sample period for the treatment and the control group. Remarkable is that the mean of firm value of the treatment group is well above the mean of the control group for the whole period. This finding is consistent with the
statistics from Panel A of Table V. For both the control and the treatment group, firm value drops from 2010 onwards until it recovers in 2013.
[Figure I about here]
Figure II depicts the means of firm value for the 50% largest firms in terms of market value. Eye-catching is the spike in firm value for the treatment group in 2012. In 2011 when the regulation took effect, the treatment group saw a small increase in firm value followed by a significant spike in 2012. Firm value of the treatment group dropped below the pre-regulation levels in 2013 and remained quite stable in the years following this. The values of the control group seem to be stable for the whole sample period, with a small dip in the years after the regulation took effect in 2011. Both Figure I and Figure II make clear that the effects of the regulation mainly seem to unfold in the short term.
[Figure II about here]
12
widespread in recent years. The simplest setup is where two groups are compared for two different time periods. One of the groups is exposed to a treatment in the first period, but not in the second. The other group is not exposed to a treatment in any of the two periods. The average gain in the control group is subtracted from the average gain in the treatment group. This removes biases in second period comparisons between the treatment and the control group that could be the result from permanent differences between these two groups. It also removes biases from comparisons over time in the treatment group that could be the result of trends. Measuring firm value by Tobin’s q, I start with estimating equation 1:
Tobin’s qit = α1 + β1Treati + β2Aftert + β3Treati*Aftert + ηt+ εit (1)
Where Tobin’s qit represents firm i in year t. As mentioned in the previous section, Treati is a
dummy variable marking all firm-year observations belonging to the treatment group. After is a dummy variable representing all years in the post regulation period, from 2011 until 2018. The law was publicly announced in 2012 but I chose 2011 as a cut-off point because the information about the introduction of mandatory GHG reporting circulated as early as September 2011. ηt is a set of year dummies and εit is an error term.
β3 is the most critical coefficient following from this regression and is called the DID
coefficient, in this case it measures the difference in the before-after difference of Tobin’s q between the treatment and the control group. Β1 measures the difference in Tobin’s q
between the treatment and the control group in the years before 2011 and β2 measures the
difference in Tobin’s q between the periods 2009-2011 and 2011-2018 for the control group.
[Table VI about here]
The results of the standard DID estimation for the years 2009-2018 are reported in column 1, Panel A of table VI. The most important coefficient from this regression, β3, is statistically
insignificant. This suggests that the change in value as a result of the regulation of firms belonging to the treatment group is not different from those belonging to the control group. Hypothesis II is rejected. β1 is positive and statistically significant, implying that Tobin’s q of
the treatment group in the years 2009 and 2010 is higher than that of the control group. The coefficient for After, β2, is insignificant, suggesting that there is no evidence to assume that
13
Doing the same regressions for firms with a moderate and a high level of family control in Panel C and Panel D yields similar results in terms of β3. Noteworthy is that β2 becomes
statistically significant at the 5% level; this suggests that the firm value of family firms is lower in the post-regulation period compared to the pre-regulation period. β 1 is positive and
statistically significant suggesting that the firm value of family firms belonging to the treatment group is higher compared to the control group.
By adding the natural logarithm of firm assets to equation (1), I make sure that the differences in firm valuation are not due to a difference in firm size between the control and the treatment group. I estimate equation 2:
Tobin’s qit = α1 + β1Treati + β2Aftert + β3Treati*Aftert + β4ln(Assets)it + ηt+ εit (2)
By adding size to the first equation, I also make sure that any changes in asset size during 2009-2018 are accounted for. Column 2 of Table VI reports the results of this estimation. The coefficient β4 is negative and statistically significant, this makes sense since firm value is
defined as market value divided by assets. In terms of the other coefficients, β1 decreases but
remains statistically significant. This suggests that firms belonging to the treatment group generally had a higher market value.
To test whether the results of these estimations would be different for a change in sample period, I performed the same regressions for the years 2009-2014. The results can be observed in columns 3 and 4 of Table VI. β3 remains insignificant, implying there is no difference in
firm value for the post-regulation period between firms belonging to the treatment group as opposed to those belonging to the control group. Noteworthy is that β2 becomes statistically
significant in all estimations, this entails that firm value is generally higher in 2009 and 2010 compared to 2011-2014.
14
B. Robustness checks
In this section, I explore whether the results of the DID estimation presented in table VI vary among firm size, industry and time. In the last subsection in perform some placebo tests.
B.1. Firm Size
As mentioned in related literature, the DID coefficient is expected to be stronger for larger firms because issues related to climate change are potentially more relevant for bigger firms. Larger firms generally have higher GHG emissions, causing them to be more financially vulnerable to regulations regarding this. I thus restrict the sample to the largest 50% firms in terms of market value. The results can be observed in column 5 and 6 of table VI. β3 now
becomes statistically significant and is negative. This implies that firms belonging to the treatment group saw a drop in market value after the regulation period. Moreover, looking at column 5 and 6 of Panel C and D in Table VI, β3 is closer to zero the more family control is
present in the firm. This suggests that family firms belonging to the treatment group saw less severe decreases in firm value compared to the control group and can be considered as
evidence in favour of Hypothesis III. Adding the natural logarithm of assets to the estimation, and thereby accounting for pre-existing differences in firm size influencing firm value, the results remain. The results can be interpreted as showing a selection bias; the firms that already disclosed their emission values expected a favourable reaction from investors, hence they had an incentive to disclose. The firms who did not disclose expected that doing this would affect their firm value negatively. Hence when they were forced to disclose, they saw a drop in their firm value.
B.2. Carbon Intensive Industries
15
[Table VII about here]
The results are somewhat surprising since the consumer goods sector has the highest positive coefficient that is statistically significant. The DDD coefficient of the utilities sector is
positive and statistically significant, as expected. However, the DDD coefficient for the Basic Materials sector is negative and also significant. This could be due to the reason that investors underestimated the GHG emissions of this sector. Another probable reason is that various FTSE 350 firms are conglomerates; they consist of multiple business entities operating in entirely different sectors. For this reason, it might be hard to assign them to one particular industry.
B.3. Time-series Variation in DID coefficient
Now I explore the time pattern that is the dynamics of the average treatment effect. This is of major importance to validate the DID approach because it displays whether the parallel trends assumption is violated or not. Instead of interacting the dummy Treat with the dummy After, I now interact it with every year dummy.
[Table VIII about here]
The results of this estimation are displayed in Table VIII. Most importantly, in both
specifications the interaction term between the Treat and the year dummies for 2009 and 2010 are insignificant. This confirms that the parallel trends assumption is not violated and the DID approach is justified. Furthermore, most interaction effects of the year dummies with the treatment dummy in the post-regulation period are also insignificant. Noteworthy is that this interaction effect is significant for the year 2011 in the second column, suggesting a strong impact of the new regulation right after firms could pick up signals of this. As mentioned in Section II, companies could have anticipated the new government regulations as early as January 2011.
B.4. Placebo tests
16
marks all firm-year observations of firms that did not publicly report their emission values prior to 2013 and started doing so in or after this year. Columns 1 and 2 cover the whole sample of 3,034 firm-year observations while columns 3 and 4 restrict the sample to the 50% largest firms in terms of market value. Panel A covers the whole sample of firms, Panel B nonfamily firms, Panel C firms with moderate level of family control and Panel D firms with a high level of family control.
[Table IX about here]
The DID coefficients are not significant for any of the specifications. The average treatment effect is positive and significant in all specifications in columns 1-4. This implies that firms that did not disclose their emission values prior to 2013 but started doing so in the period afterwards generally had a higher firm value compared to the control group. Limiting the sample to the 50% largest firms in column 3 and 4, I observe that for most specifications the coefficient of After is negative and significant. This means that the firm value of the larger firms was generally lower in the period after 2013 compared to the period before.
In column 5-8 of Table IX, I reestimate the equation (2) with a randomly selected treatment group. Given that 1,007 firm-year observations belong to the treatment group in the initial dataset, I randomly assign the same amount of firm-year observations to a placebo treatment group. For every firm in the treatment group, I assign all firm-year observations to the treatment group. The purpose of this placebo group is to account for effects from the treatment that do not depend on the treatment itself. Again, the DID coefficients are not significant in any of the specifications.
VI. Conclusion
17
variation in disclosure policies.
Although the evidence is far from conclusive, I find some evidence against and in favour of the hypotheses. Firstly, I find that prior to the regulation, family firms indeed disclosed less of their GHG emissions compared to nonfamily firms. Secondly, I find that firms not compliant with the new regulation experience no different valuation effects compared to firms that were already compliant with the regulation. Going by the reasoning that climate change is more relevant for bigger firms, I restrict the sample to the 50% largest firms. Consistent with this, I find that family firms disclosing their emission values after 2011 saw less of a decrease in their firm value compared to nonfamily firms. This result is robust to two placebo tests. Lastly, performing a DID regression with interaction effects between year dummies and the treatment dummy, I find that the year 2011 together with the treatment dummy generates a significant and positive interaction effect. This implies that firms that did not disclose prior to 2011 and started doing so after this year, saw positive valuation effects in 2011.
18
References
Ali, A., Chen, T. Y., & Radhakrishnan, S. (2007). Corporate disclosures by family firms. Journal of accounting and economics, 44(1-2), 238-286.
Angrist, J. D., & Pischke, J. S. (2008). Mostly harmless econometrics: An empiricist's companion. Princeton university press.
Berrone, P., Cruz, C., Gomez-Mejia, L. R., & Larraza-Kintana, M. (2010). Socioemotional wealth and corporate responses to institutional pressures: Do family-controlled firms pollute less?. Administrative Science Quarterly, 55(1), 82-113.
Brammer, S., & Millington, A. (2006). Firm size, organizational visibility and corporate philanthropy: An empirical analysis. Business Ethics: A European Review, 15(1), 6-18. Chau, G. K., & Gray, S. J. (2002). Ownership structure and corporate voluntary disclosure in Hong Kong and Singapore. The International journal of accounting, 37(2), 247-265.
Cho, C. H., & Patten, D. M. (2007). The role of environmental disclosures as tools of legitimacy: A research note. Accounting, organizations and society, 32(7-8), 639-647. Cormier, D., & Magnan, M. (1997). Investors' assessment of implicit environmental liabilities: An empirical investigation. Journal of accounting and public policy, 16(2), 215-241.
Fisher-Vanden, K., & Thorburn, K. S. (2011). Voluntary corporate environmental initiatives and shareholder wealth. Journal of Environmental Economics and management, 62(3), 430-445.
Flöstrand, P., & Ström, N. (2006). The valuation relevance of non-financial information. Management Research News, 29(9), 580-597.
Frankel, R., McNichols, M., & Wilson, G. P. (1995). Discretionary disclosure and external financing. Accounting Review, 135-150.
Greenstone, M., Oyer, P., & Vissing-Jorgensen, A. (2006). Mandated disclosure, stock returns, and the 1964 Securities Acts amendments. The Quarterly Journal of
19
Hughey, C. J., & Sulkowski, A. J. (2012). More disclosure= better CSR reputation? An examination of CSR reputation leaders and laggards in the global oil & gas industry. Journal of Academy of Business and Economics, 12(2), 24-34.
Konar, S., & Cohen, M. A. (2001). Does the market value environmental performance?. Review of economics and statistics, 83(2), 281-289.
Kim, E. H., & Lyon, T. P. (2011). Strategic environmental disclosure: Evidence from the DOE's voluntary greenhouse gas registry. Journal of Environmental Economics and Management, 61(3), 311-326.
Kim, E. H., & Lyon, T. (2011). When does institutional investor activism increase
shareholder value?: the carbon disclosure project. The BE Journal of Economic Analysis & Policy, 11(1).
Krüger, P. (2015). Climate change and firm valuation: Evidence from a quasi-natural experiment. Swiss Finance Institute Research Paper, (15-40).
Lyon, T. P., & Maxwell, J. W. (2008). Corporate social responsibility and the environment: A theoretical perspective. Review of environmental economics and policy, 2(2), 240-260.
Miller, D., & Le Breton-Miller, I. (2003). Challenge versus advantage in family business. Strategic Organization, 1(1), 127-134.
Nekhili, M., Nagati, H., Chtioui, T., & Rebolledo, C. (2017). Corporate social responsibility disclosure and market value: Family versus nonfamily firms. Journal of Business
Research, 77, 41-52.
Palmer, K., Oates, W. E., & Portney, P. R. (1995). Tightening environmental standards: the benefit-cost or the no-cost paradigm? Journal of economic perspectives, 9(4), 119-132. Porter, M. E., & Van der Linde, C. (1995). Toward a new conception of the environment-competitiveness relationship. Journal of economic perspectives, 9(4), 97-118.
Tauringana, V., & Chithambo, L. (2015). The effect of DEFRA guidance on greenhouse gas disclosure. The British Accounting Review, 47(4), 425-444.
20
21
Appendices
A Figures
Figure I
Parallel trends figure: all firm-year observations
This figure reports the mean of firm value measured as Tobin’s q for the treatment and the control group. The treatment group consist of all firms that did not disclose their GHG emissions prior to 2011 but started doing so in or after this year. The dataset includes all firm-year observations from 2009 to 2018.
22
Figure II
Parallel trends figure: 50% largest firms
23
B Tables
Table I
Summary statistics overall
This table reports summary statistics of firm-level variables. The sample runs from 2009 to 2018. Tobin’s q is defined as Market value divided by Total Assets. Market value and Total Assets are measured in millions. All
financial variables are measured in domestic currency, i.e. British Pounds. CO2 emissions consists of the sum of
Scope 1 and Scope 2 emissions and is measured in kilotonnes. Treat is a dummy variable marking all firm-year observations belonging to the treatment group. Firms belonging to this group did not publicly report their GHG
emission values prior to 2011. SD displays the standard deviation, P25 and P75 represent the 25th and the 75th
percentiles respectively. N represents the amount of observations.
Variable Mean Median SD P25 P75 N
24
Table II
Summary statistics per industry
This table reports the summary statistics of GHG emissions per industry. The sample size ranges from 2009 to 2018 and all values are measured in kilotonnes. SD displays the standard deviation, P25 and P75 represent the
25th and the 85th percentiles respectively. Firm-year observations with undisclosed GHG emission values are not
taken into account for performing the calculations leading up to this table.
Industry Mean emissions Median emissions SD P25 P75 N Basic Materials 14,100,000 450,199 24,000,000 191,912 18,800,000 197 Consumer Goods 687,342 135,644 1300,526 63,715 509,000 196 Consumer Services 1,800,482 201,544 4,341,112 42,452 918,067 269 Financials 76,253 8,623 176,164 2,058 45,955 457 Health Care 387,339 90,219 610,967 7,908 570,522 74 Industrials 1,306,318 74,069 5,209,450 21,020 361,300 529
Oil and Gas 2,426,101 949,525 3,121,736 202,755 3,514,675 50
Technology 59,610 72,900 44,241 20,654 84,567 32
Telecommunications 301,652 24,914 715,364 6,931 143,742 116
Utilities 8,009,873 4,211,898 8,682,217 540,401 12,700,000 60
25
Table III
Summary statistics of family firms vs nonfamily firms and treatment firms vs nontreatment firms
This table reports summary statistics of firm-level variables for different sub-groups in the dataset. The sample runs from 2009 to 2018. Tobin’s q is defined as Market value divided by Total Assets. All financial variables are
measured in domestic currency, i.e. British Pounds. Total Assets is measured in millions. CO2 emissions consists
of the sum of Scope 1 and Scope 2 emissions and is measured in kilotonnes. SD displays the standard deviation, N represents the total amount of observations.
Group Variable Mean Median SD N
Nonfamily firms (<3% family ownership) Total assets 43,000 2,000 214,000 2457 Tobin’s q 1.07 0.76 1.38 2457 CO2 emissions 2,508,198 998,48 9,741,543 1606 Moderate family control (>3% family ownership) Total assets 4,790 1,500 13,500 577 Tobin’s q 1.31 0.79 1.62 577 CO2 emissions 2,036,268 59,798 7,158,653 374 <10% family ownership Total assets 40,300 1,920 206,000 2,660 Tobin’s q 1.08 0.77 1.39 2,660 CO2 emissions 2,466,165 99,879 9,602,227 1,721
High family control (>10% family ownership)
Total assets 3,240 1,630 4,110 374
Tobin’s q 1.33 0.76 1.67 374
CO2 emissions 2,106,026 28,047 7064273 259
Treatment group Total assets 2,570 1,100 4,560 1,007
Tobin’s q 1.45 0.90 1.95 1,007
CO2 emissions 1,006,109 29,644 5,256,849 532
Nontreatment group Total assets 52,200 2,750 234,000 2,027
Tobin’s q 0.95 0.72 1.044 2,027
26
Table IV
CDP response permission per year (FTSE 350)
When submitting data relating to climate change to CDP, the firms are asked to respond either publicly or privately. Private responses are only available to CDP and other institutions who can request information from CDP. Public responses are also available to view for the general public. NA marks firms that did not respond to the request of CDP. Panel A tabulates the number of sample firms per response against years for all sample firms. Panel B does this for nonfamily firms, Panel C for firms with a moderate level of family control and Panel D for firms with a high level of family control. Public/total represents the percentage of firms within that sub-group that chose to publicly disclose their GHG emission scores.
Panel A: Number of firms per response; all FTSE 350 firms
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Public 132 152 151 162 167 210 228 249 261 268
Private or NA 148 129 130 129 99 98 90 78 73 80
Public/total 0.47 0.54 0.54 0.56 0.63 0.68 0.72 0.76 0.78 0.77
Panel B: Number of firms per response; <3% family controlled firms (nonfamily firms)
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Public 112 129 126 134 142 169 183 200 205 206
Private or NA 117 103 106 103 77 82 74 63 60 66
Public/total 0.49 0.55 0.54 0.57 0.65 0.67 0.71 0.76 0.77 0.76
Panel C: Number of firms per response; >3% family controlled firms
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Public 20 23 25 28 25 41 45 49 56 62
Private or NA 31 26 24 26 22 16 16 15 13 14
Public/total 0.39 0.47 0.51 0.52 0.53 0.71 0.74 0.77 0.81 0.82
Panel D: Number of firms per response; >10% family controlled firms
2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Public 13 16 16 19 17 28 31 33 37 39
Private or NA 22 18 16 14 13 8 6 6 4 8
27
Table V:
Effect of regulation on Tobin’s q in family and nonfamily firms
This table presents the results of the differential impact of the mandatory GHG emissions disclosure on Tobin’s q. I compare the average of Tobin’s q for family firms and nonfamily firms, in the periods before and after the mandatory GHG emissions disclosure period. ***, ** and * denote significance at 1%, 5% and 10% levels, respectively. Panel B restricts the sample to the largest 50 percent of firm-year observations.
Panel A: All 350 FTSE companies
Family firm Firm group Difference
Treatment group Control group
Before regulation Yes 1.64 1.40 0.24
No 1.65 0.96 0.69***
After regulation Yes 1.38 1.18 0.2
No 1.42 0.89 0.53***
After-before Yes -0.26 -0.22 -0.04
No -0.23 -0.07 -0.16
Panel B: Largest 50% of firms
Family firm Firm group Difference
Treatment group Control group
Before regulation Yes 2.42 1.54 0.88
No 4.06 1.01 3.05***
After regulation Yes 1.64 1.12 0.52**
No 2.04 0.98 1.06***
After-before Yes -0.78 -0.43 -0.35
28
Table VI:
Effect of The Companies Act 2006 Regulations 2013 on Tobin’s q in family and nonfamily firms
This table shows the results of the DID estimations of The Companies Act 2006 on firm value, defined as Tobin’s q. Estimations are done for two sample periods, 2009-2014 and 2009-2018. Tobin’s q is the dependent variable and Treat is a dummy variable marking all observations after 2010 for firms that did not publicly disclose their emission values before 2011. The control group constitutes of all firms that did disclose their emission values before 2011. After is a dummy variable marking all firm-year observations after 2010, i.e. the post regulation period. Panel A covers all firms, Panel B nonfamily firms, Panel C firms with a moderate level of family control and Panel D firms with a high level of family control. Standard errors are displayed in
parentheses. (* p < 0.1, ** p < 0.05, *** p < 0.01)
Panel A: All FTSE 350 companies
(1) (2) (3) (4) (5) (6) ln(assets) -0.272*** (0.014) -0.219*** (0.019) -0.424*** (0.021) Treat 0.617*** (0.129) 0.231* (0.124) 0.617*** (0.130) 0.307** (0.129) 2.340*** (0.281) 1.537 *** (0.255) After -0.096 (0.793) -0.025 (0.075) -0.260*** (0.088) -0.224*** (0.085) -0.084 (0.120) -0.065 (0.107) Treat * After -0.141 (0.142) -0.073 (0.135) -0.096 (0.159) -0.089 (0.019) -1.454*** (0.300) -1.191*** (0.268) Observations 3,034 3,034 1,707 1,707 1,524 1,524 R2 0.029 0.130 0.040 0.108 0.095 0.130
2009-2018 YES YES NO NO YES YES
2009-2014 NO NO YES YES NO NO
50% largest firms NO NO NO NO YES YES
Panel B: <3% family ownership firms (nonfamily firms)
(1) (2) (3) (4) (5) (6) ln(assets) -0.227*** (0.015) -0.192*** (0.020) -0.391*** (0.023) Treat 0.679*** (0.142) 0.359*** (0.137) 0.679*** (0.147) 0.408*** (0.146) 3.050*** (0.338) 2.210*** (0.308) After -0.078 (0.083) -0.019 (0.079) -0.233** (0.094) -0.201** (0.091) -0.029 (0.127) -0.025 (0.115) Treat * After -0.149 (0.156) -0.091 (0.150) -0.107 (0.179) -0.096 (0.174) -1.991*** (0.358) -1.737*** (0.322) Observations 2,457 2,457 1,400 1,400 1,524 1,524 R2 0.035 0.119 0.043 0.101 0.117 0.286
2009-2018 YES YES NO NO YES YES
29
50% largest firms NO NO NO NO YES YES
Panel C: >3% family ownership firms
(1) (2) (3) (4) (5) (6) ln(assets) -0.273*** (0.014) -0.220*** (0.019) -0.429*** (0.022) Treat 0.514*** (0.059) 0.199*** (0.059) 0.617*** (0.079) 0.274*** (0.080) 1.234*** (0.107) 0.596*** (0.101) After -0.134** (0.067) -0.036 (0.063) -0.260*** (0.014) -0.238*** (0.072) -0.294*** (0.110) -0.232** (0.099) Treat * After -0.062 (0.107) -0.129 (0.102) -0.096 (0.549) -0.167 (0.159) -0.558*** (0.212) -0.563*** (0.189) Observations 3,034 3,034 1,707 1,707 1,524 1,524 R2 0.028 0.132 0.040 0.108 0.085 0.272
2009-2018 YES YES NO NO YES YES
2009-2014 NO NO YES YES NO NO
50% largest firms NO NO NO NO YES YES
Panel D: >10% family ownership firms
(1) (2) (3) (4) (5) (6) ln(assets) -0.272*** (0.014) -0.219*** (0.019) -0.430*** (0.021) Treat 0.525*** (0.057) 0.189*** (0.057) 0.557*** (0.077) 0.247*** (0.079) 1.204*** (0.105) 0.573*** (0.099) After -0.130** (0.066) -0.041 (0.062) -0.287*** (0.073) -0.251*** (0.071) -0.300** (0.110) -0.237** (0.099) Treat * After -0.182 (0.129) -0.132 (0.122) -0.053 (0.199) 0.001 (0.019) -0.541** (0.236) -0.597*** (0.211) Observations 3,034 3,034 1,707 1.707 1,524 1,524 R2 0.0289 0.132 0.0395 0.1077 0.084 0.272
2009-2018 YES YES NO NO YES YES
2009-2014 NO NO YES YES NO NO
30
Table VII:
Difference-in-Difference-in-Differences estimation
This table shows the results of the DDD estimations for different industries of The Companies Act 2006 on firm value, defined as Tobin’s q. The estimations are done for the sample period 2009-2018. I obtain the DDD coefficients by interacting the DID coefficient by an industry dummy. This estimation controls for size, includes year dummies and is set-up the same way as equation (2). Standard errors are displayed in parentheses. (* p < 0.1, ** p < 0.05, *** p < 0.01)
Industry DID coefficient
Basic Materials -0.679*** (0.164) Consumer Goods 0.540*** (0.133) Consumer Services 0.036*** (0.002) Financials -0.272*** (0.106) Health Care 0.121 (0.220) Industrials -0.143 (0.106)
Oil and Gas -0.834
31
Table VIII
Dynamics of the effect of The Companies Act 2006 Regulations 2013 on Tobin’s q
This table shows the DID estimates of the effect of The Companies Act 2006 Regulations 2013 on firm value defined as Tobin’s q for each year of the sample period 2009-2018. Treat is a dummy variable marking all firm-year observations after 2010 of firms who did not disclose their GHG emission values prior to 2011. The control group comprises of firms who did publicly disclose their GHG emission values before 2011. Year = t are year dummies. Treat * Year = t are interaction effects between the respective treatment and year dummies. Standard errors are displayed in parentheses. (* p < 0.1, ** p < 0.05, *** p < 0.01)
32
Table IX Placebo Tests
This table shows two placebo tests. In column 1-4, I reestimate the main specification assuming that a law change has happened in 2013. Panel A shows the results for all firms in the sample, Panel B for nonfamily firms, Panel C for firms with a moderate level of family control and Panel D for firms with a high level of family control. In column 5-8 I randomly assign 1,007 firms to the treatment group and reestimate the main specification. The dependent variable in all specifications is Tobin’s q. In columns 1-4, Treat is a dummy variable marking all observations after 2013 for firms that did not publicly disclose their emission values before 2013. Accordingly, the control group comprises of all firms that did publicly disclose their emission values prior to 2013. In columns 5-8, Treat is dummy variable marking 1,007 randomly selected firm-year observations. In columns 1-4, After is a dummy variable marking all firm-year observations after 2012. In columns 5-8, After marks all firm-year observations after 2010. Standard errors are displayed in parentheses. (* p < 0.1, ** p < 0.05, *** p < 0.01)
Panel A: All 350 FTSE companies
(1) (2) (3) (4) (5) (6) (7) (8) ln(assets) -0.284*** (0.014) -0.427*** (0.501) -0.294*** (0.013) -0.472*** (0.021) Treat 0.638*** (0.097) 0.233** (0.094) 3.372*** (0.272) 2.409*** (0.247) -0.149 (0.128) -0.156 (0.119) -0.088 (0.219) -0.317* (0.190) After -0.077 (0.116) 0.060 (0.110) -0.206 (0.183) -0.1888 (0.162) -0.149 (0.122) 0.011 (0.114) -0.351* (0.208) -0.410** (0.180) Treat * After -0.285 (0.121) -0.179 (0.115) -2.791 (0.298) -2.371 (0.265) 0.067 (0.635) 0.049 (0.131) -0.035 (0.240) 0.068 (0.207) Observations 3,034 3,034 1,524 1,524 3,034 3,034 1,524 1,524 R2 0.039 0.152 0.111 0.300 0.019 0.151 0.008 0.261
50% largest firms NO NO YES YES NO NO YES YES
Placebo law 2013 YES YES YES YES NO NO NO NO
Random Treatment NO NO NO NO YES YES YES YES
Panel B: <3% family ownership firms (nonfamily firms)
(1) (2) (3) (4) (5) (6) (7) (8) ln(assets) -0.282*** (0.014) -0.435*** (0.021) -0.295*** (0.014) -0.475*** (0.021) Treat 0.597*** (0.074) 0.247*** (0.072) 1.908*** (0.181) 1.128*** (0.165) -0.037 (0.082) -0.138** (0.077) -0.206 (0.143) -0.462* (0.124) After -0.120 (0.114) 0.043 (0.107) -0.377** (0.184) -0.337** (0.163) -0.105 (0.116) 0.019 (0.108) -0.401** (0.195) -0.471*** (0.168) Treat * After -0.200 (0.098) -0.180 (0.014) -1.335 (0.217) -1.092 (0.193) -0.091 (0.098) 0.035 (0.092) 0.134 (0.166) 0.299 (0.143) Observations 3,034 3,034 1,524 1,524 3,034 3,034 1,524 1,524 R2 0.044 0.154 0.082 0.277 0.019 0.152 0.008 0.263
50% largest firms NO NO YES YES NO NO YES YES
Placebo law 2013 YES YES YES YES NO NO NO NO
33
Panel C: >3% family ownership firms
(1) (2) (3) (4) (5) (6) (7) (8) ln(assets) -0.282*** (0.014) -0.442*** (0.022) -0.294*** (0.136) -0.474*** (0.021) Treat 0.493*** (0.068) 0.165** (0.066) 1.057*** (0.122) 0.432*** (0.113) -0.126** (0.058) -0.114** (0.055) -0.082 (0.094) -0.199** (0.082) After -0.161 (0.113) 0.009 (0.106) -0.489*** (0.185) -0.423*** (0.164) -0.140 (0.114) 0.028 (0.106) -0.348** (0.190) -0.358* (0.164) Treat * After 0.108 (0.135) 0.037 (0.127) -0.139 (0.261) -0.173 (0.231) 0.165 (0.113) -0.015 (0.106) -0.212 (0.197) -0.376 (0.170) Observations 3,034 3,034 1,524 1,524 3,034 3,034 1,524 1,524 R2 0.040 -0.153 0.059 0.268 0.019 0.152 0.009 0.264
50% largest firms NO NO YES YES NO NO YES YES
Placebo law 2013 YES YES YES YES NO NO NO NO
Random Treatment
NO NO NO NO YES YES YES YES
Panel D: >10% family ownership firms
(1) (2) (3) (4) (5) (6) (7) (8) ln(assets) -0.282*** (0.014) -0.442*** (0.022) -0.294*** (0.014) -0.475*** (0.021) Treat 0.524*** (0.065) 0.176*** (0.064) 1.073*** (0.120) 0.446*** (0.111) -0.104* (0.057) -0.112 (0.053) -0.097 (0.093) -0.210*** (0.080) After -0.150 (0.112) 0.013 (0.106) -0.487*** (0.185) -0.426*** (0.164) -0.132 (0.114) 0.029 (0.106) -0.354* (0.189) -0.361** (0.164) Treat * After -0.086 (0.166) -0.035 (0.156) -0.278 (0.284) -0.313 (0.251) 0.082 (0.133) -0.039 (0.124) -0.167 (0.230) -0.452 (0.021) Observations 3,034 3,034 1,524 1,524 3,034 3,034 1,524 1,524 R2 0.043 0.153 0.059 0.263 0.019 0.152 0.008 0.263
50% largest firms NO NO YES YES NO NO YES YES
Placebo law 2013 YES YES YES YES NO NO NO NO
Random Treatment
