Auditors’ Carbon Risk Consideration under the EU Emission Trading System

ABSTRACT

This paper addresses the effects of clients’ carbon risk on audit pricing. Using data from 438 EU companies for the period 2013–2019, we find a positive relationship between carbon risk (measured by the level of carbon emissions) and audit fees. Furthermore, we find that participation in the European Union’s Emission Trading System, a limited market and regulation scheme to mitigate special industries’ Greenhouse Gas emissions, strengthens the positive relationship between carbon risk and audit fees. Insights from additional tests indicate that auditors price carbon risk particularly for large clients that are under greater public scrutiny and that the increase in fees rather stems from a risk premium charged by the auditor than higher audit effort. With interest in climate change developing rapidly across society, practice and research combined with the increasing importance of reducing carbon risk, our findings are timely and should thus appeal to a wide variety of recipients.

Keywords:

• audit fees
• audit risk
• carbon risk
• GHG emission
• EU emission trading system

Acknowledgements

We thank Max Göttsche, Francesco Mazzi, Frank Schiemann and seminar participants at the 17th Workshop on European Financial Reporting (EUFIN), University of Bamberg and University of Ingolstadt for valuable comments and suggestions. All errors remain our own.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1 In the context of sustainability reporting, Hummel and Szekely (2022) have recently shown that not only financial stakeholders (e.g., analysts, investors, lenders) but also non-financial stakeholders (e.g., media, employees, customers) are important in building pressure on firms’ sustainability behavior.

2 For further information and reference see: https://www.reuters.com/article/us-climate%20change-accounts-exclusive/exclusive-big-four-auditors-face-investor-calls-fortougher-climate-scrutiny-idUSKBN1Y21XK

3 See for instance the results of the Yale Climate Opinion Survey, which are publicly available and discussed in Marlon et al. (2022).

4 However, it is important to note that regulation alone is not sufficient in increasing the awareness for climate change risks but that informal institutions like cultural-cognitive factors are also important (Panfilo & Krasodomska, 2022).

5 Some studies more broadly define carbon risk a set of environmental risks which ‘describe any corporate risk related to climate change or the use of fossil fuels’ (Hoffmann & Busch, 2008, p. 514). According to this definition, carbon risk relates to firms’ reliance on fossil fuels and corresponding socio-political factors such as government-related measures and changes in consumer preferences, but also the direct physical effects of climate change (Jung et al., 2018). In this paper, we follow the definition by the FSB and understand carbon risk mainly as the indirect transitional risks from climate change for carbon-intensive firms.

6 Simunic (1980) shows in his seminal paper that audit fees are the total expected costs of performing an audit, including the auditing hours expended and allowances made for future losses, for instance due to litigation risks.

7 Meanwhile, it is one of the main mechanisms to attain the goals of the European Commission becoming climate neutral by 2050 (European Commission, 2020b).

8 It includes installations, which are heavy energy using e.g. power and heat generation, oil refineries, steel works and production of iron, aluminum, metals, cement, lime, glass, ceramics, pulp, paper, cardboard, acids and bulk organic chemicals and civil aviation (European Commission, 2016).

9 In 2013, the cap of stationary installations is 2,084,301,856 allowances and in the aviation sector it is 210,349,264 allowances (European Commission, 2020a).

10 For further information on the impacts and consequences of free allocation of allowances, we refer to Ellerman et al. (2011).

11 For further information see European Commission (2020c).

12 Information is provided by the European Union Transaction Log (EUTL).

13 Further adjustments focusing on the fourth phase include adjustments due to the commitment of the Paris Agreement signed in 2015, i.e. the reduction of emissions of about 40% compared to the 2005 level (European Commission, n.d.b).

14 The yearly reduction of the cap affects only EAUs (allowances for stationary installations) in phase 3 and from 2021 onwards stationary and aviation allowances (European Commission, 2020c).

15 EU ETS participants are identified with data from the EUTL.

16 All continuous variables in our regression models are winsorised at the 1% level.

17 At the end of 2019, the United Kingdom was a member of the EU. For further information, see https://ec.europa.eu/info/strategy/relations-non-eu-countries/relations-united-kingdom/eu-uk-withdrawal-agreement_en.

18 0.090 = 2.917 * 0.031, in which 2.917 presents the standard deviation of scope1_ln and 0.031 is the estimated coefficient of scope1_ln.

19 Additional tests (untabulated) show that the total GHG emissions, which are calculated as the sum of Scope 1 and Scope 2 GHG emissions, are also positively related to audit fees (p-value <.05).

20 Further tests (untabulated) show that results are unchanged if we calculate the total GHG emission output as the sum of Scope 1 and Scope 2 GHG emissions.

21 We conduct the additional tests and sensitivity analyses only for our Equations 1 (direct effect of carbon risk on audit fees) and 3 (interaction effect), for which we reported significant results in our main tests.

22 In this model, we use l_audfee as a control variable.

23 Of the remaining industries, only the mining industry has a higher level of carbon emissions compared to the manufacturing industry.

24 Additional tests (untabulated) indicate that the findings are more prominent for common-law countries (Great Britain and Ireland). The categorization is based on the classification of La Porta et al. (1998).

25 We again attribute this to the significantly lower number of observations for this test, which naturally reduces t-values, particularly bearing in mind that the interaction term indirect_ln*ets_part was already only marginally significant at the 10% level for the full sample.

26 Examples are the implementation of a market stability reserve and shortages of overall and free allowance certificates see Section 2.3.