How are multinational oil and gas companies talking about renewable energies in their annual reports since the Paris Agreement? This study looks at the sentiments embedded within companies' messages about renewable energies in their communications to shareholders, to understand how corporate messages may shape the perceived values and roles of solar, wind, and biofuels.
By Dipa Desai, NULab and DITI Fellow and Public Policy PhD Student
For my research project with the NULab, I used NVivo 12 software to qualitatively analyze the text in corporate annual reports from major oil and gas companies to see how companies are talking about renewable energy technologies. This research project was borne out of a broader Tier 1 Grant project from Northeastern University looking at the energy communications of multinational fossil fuel companies on social media (Si et al., 2023) and corporate reports. Specifically, this research focused on the renewable energy discourse of Shell, BP, ExxonMobil, and TotalEnergies in corporate annual reports from 2016 to 2021, to assess trends in messaging since the Paris Agreement and global adoption of emissions reductions goals. I chose to study corporate annual reports because I am interested in the discourse being developed and delivered to company shareholders. Content analyses of company annual reports have been used widely across organizational research (Duriau et al., 2007), and is becoming an important research tool to study businesses with international operations (Gaur and Kumar, 2018). The process of qualitatively analyzing text involves iterative research to identify keywords and develop coding schema which can be reliably applied to corporate annual reports from different companies.
When the Digital Integration Teaching Initiative (DITI) teaches the NVivo tool, it is usually in the context of social science and humanities classes. The texts we use for familiarizing students with the NVivo interface are usually sample interview transcripts, or excerpts from historical or literary writings. These sample texts usually have simple content and similar structure to demonstrate how texts can be analyzed and compared across contexts. Applying NVivo to corporate annual reports for research on multinational oil and gas businesses was a little trickier, for several reasons. Currently, multinational oil and gas corporations have no standardized format for reporting their annual results and business outlooks. Often, companies report quantitative financial information, qualitative corporate social responsibility and sustainability information, and a variety of other business performance data. There has been a recent push to standardize corporate reporting so companies’ annual reports include financial data as well as corporate sustainability information (International Financial Reporting Standards Foundation, 2022). Only one of the companies in my study, TotalEnergies, produces a form of this integrated report, and the rest of the companies under study produce annual reports of varying length and content. However, since all of the companies are listed on the New York Stock Exchange, they all are required to report certain data to the US Securities and Exchange Commission. The format and level of detail of this required information can vary across the corporate annual reports. Moreover, several of the companies in my study are UK-based companies, and are required by the UK government to produce a report of business strategy. This strategic report section in companies’ annual reports similarly vary in content and structure. If I planned on analyzing the entirety of the corporate reports, the range of variation across the texts would make it harder for me to compare across the companies.
Thus, an initial step in my research was to define the bounds of text I would be analyzing for each company’s report. I decided to only qualitatively analyze the narrative text within the strategic reporting sections of the corporate reports. For companies without an explicit strategic report section, I focused on analyzing the non-financial sections on company outlook, business strategy, and corporate social action that roughly correspond to the strategic report contents of other companies. Since the format of text varied even within the strategic report section of corporate reports, I further limited the unit of text analysis to sentences and phrases which contain mentions to renewable energies.
The Qualitative Coding Process
In my first round of coding, I used NVivo to search for keywords within these report sections that could reliably be used across all companies’ reports to find mentions of renewable energy technologies, such as “solar” and “wind”. The term “solar” finds more references to solar renewable energy than the keyword “photovoltaic” even though both refer to energy from sunlight. For other forms of renewable energies, I found that finding a reliable keyword involved testing keywords across companies’ reports. I used the word frequency feature in NVivo to find the keywords that represented the most mentions to biofuel energy, wave and tidal energy, hydropower, and geothermal energy. For biofuels, I discovered that across the multinational oil and gas companies, terms such as “biodiesel,” “biogas,” and “bioenergy” were all used at different rates by individual companies and these terms were used interchangeably to refer to liquid, ethanol-based fuels produced predominantly from corn and sugarcane. For the other alternative energies, I discovered major oil and gas companies rarely mention wave energy, hydropower, and geothermal energy.
For the second round of coding, I used the Sentiment feature on NVivo 12 Pro to code mentions to solar, wind, and biofuels in the narrative text as having positive, neutral, or negative sentiment. This process required me to develop a coding schema that would help determine the features of text which situate mentions to renewable energies in positive, neutral, or negative language. For example, sentences or phrases that referred to renewable energies as “intermittent” and needing to be paired with natural gas or batteries were coded as negative, because of the implication that the renewable energy technology has dips in supply which need to be augmented by fossil fuels or battery storage (Figure 2). Text that referred to renewable energies as being critical to achieving emissions reductions was coded as positive. Neutral mentions to renewable energies were located in text that did not directly attribute a positive or negative sentiment, or were located in text with qualified positive or negative sentiment. Cases which illustrate neutral sentiment were the mentions of solar and wind described with the adjective “low-carbon” situated in the same report against mentions to biofuels described with the adjective “lower-carbon.” I developed the coding schema while coding the 2021 reports for the four companies in my study.
To guarantee the reliability of the coding schema, I asked another Public Policy graduate student working on energy discourse, Yutong Si, to co-code annual reports from three of the four companies. I chose to co-code the 2018 reports of Shell, BP, and ExxonMobil since this year of reporting captures the middle of the study period, and the three companies have varying content and detail in their reports. The annual reports which were co-coded represent an eighth of the sample, and capture the range of messaging deployed by multinational oil and gas companies. In the co-coding process, my colleague and I went through each mention of solar, wind, and biofuels in narrative text of companies’ strategic report sections and independently identified the sentiment as positive, negative, or neutral. After independently deciding, we discussed why we thought the mention had a specific sentiment and collaboratively decided to code it as positive, negative, or neutral. This process of discussion resolved instances where our independent sentiment coding did not match, and refined the coding schema. I used the Coding reliability tool on NVivo to determine that the finalized coding schema had an intercoder reliability above 98%.
After establishing a reliable set of keywords and coding scheme, I entered the final round of coding the remainder of annual reports in the study period for Shell, BP, ExxonMobil, and TotalEnergies. This round of coding allowed me to see longitudinal trends in renewable energy discourse within and across companies over time. In this process, I encountered an unexpected discourse trend, prompting another research question. I found that all of the companies frequently re-use text in annual reports. Specifically, I found companies repeat verbatim or near-verbatim sentences or phrases within the same year’s annual report and in reports from year to year, with minor changes in syntax over time. Since my research is concerned with how oil and gas companies are designing and delivering messages about renewable energies to their shareholders and company decisionmakers, I thought the use of boilerplate text for renewable energy messaging was interesting and could have unexpected effects on the message delivery. Given that annual reports are dense documents meant to showcase the business’ performance for the year and their future actions, the re-use of text about renewable energies may work to distract shareholders, and over-emphasize certain discourse narratives about renewable energies.
Current Work & Next Steps
Currently I am testing manual and computational methods to reliably identify instances of near-verbatim, or fuzzy, textual re-use (Figure 3). I use the Coding Query feature on NVivo to review and clean the codes attributed to mentions of solar, wind, and biofuels in the annual reports. In this process, I am able to see phrases and sentences which are coded as being in a positive, neutral, and negative context, and compare the messages across different contexts. To facilitate analysis, I am transferring the coded phrases and sentences for positive and negative mentions of renewable energies into an Excel spreadsheet. This will allow me to manually distinguish the positive and negative messages delivered by each company, as well as support computational text analysis since I can use the spreadsheet with quantitative coding tools like Python.
Even in this step of manual data cleaning and organization, I am able to distinguish trends in companies’ messaging about renewables. My first main takeaway from this research is that major oil and gas companies are prioritizing the development of solar, wind, and biofuels, and seem to be unconcerned with developing wave, tidal, geothermal, and hydroelectric renewable energies based on the messaging to shareholders in corporate annual reports. Second, I see a trend of companies in my study using negative messaging to describe solar and wind as “intermittent” energy sources, citing times when the wind does not blow and the sun does not shine. This messaging strategy implies solar and wind energies are unreliable, and insufficient to meet energy demand. Interestingly, two of the four companies under study use this messaging as justification for expanding natural gas development since liquid natural gas can be easily stored and withdrawn in times of high energy demand (Figure 4). Another company in my analysis uses the message that solar and wind are intermittent to justify investments into and business partnerships with battery developers to improve storage of solar- and wind-generated electricity (Figures 2 and 3). In contrast, I found that companies promote the positive messages that solar and wind are crucial to their internal company decarbonization efforts, as well as to global emissions reductions goals. In fact, three of the four companies under study cited using solar and/or wind energies in their fossil fuel production to offset emissions and power oil and gas operations (as seen in Figure 1). It seems like solar and wind can help meet the high energy costs of fossil fuel production, and are not so intermittent and unreliable as purported by company messaging.
The mixed messaging of major oil and gas companies to their shareholders works to confuse the value of renewable energies, with potential cascading effects on its perceived investment potential and development. Multinational oil and gas companies control a large portion of the world’s energy supply, and are increasingly moving to set up leading positions in alternative energy development and distribution. In addition, oil and gas companies have historically had power and resources to shape and mis-inform climate change messaging in order to continue fossil fuel production and profits (House Oversight and Reform Committee, 2021; InfluenceMap, 2022; Supran and Oreskes, 2017). As fossil fuel companies re-brand into integrated energy companies, the messaging from companies will play an important role in shaping the energy transition to renewable energies and away from fossil fuels.
Citations:
Duriau, V. J., Reger, R. K., & Pfarrer, M. D. (2007). A content analysis of the content analysis literature in organization studies: Research themes, data sources, and methodological refinements. Organizational research methods, 10(1), 5-34.
Gaur, A., & Kumar, M. (2018). A systematic approach to conducting review studies: An assessment of content analysis in 25 years of IB research. Journal of World Business, 53(2), 280-289.
House Oversight and Reform Committee. (2021). FUELING THE CLIMATE CRISIS: EXPOSING BIG OIL’S DISINFORMATION CAMPAIGN TO PREVENT CLIMATE ACTION (No. 117–50; p. 105). US Congress. https://www.congress.gov/117/chrg/CHRG-117hhrg46026/CHRG-117hhrg46026.pdf
InfluenceMap. (2022). Big Oil’s Real Agenda on Climate Change 2022 (p. 56). InfluenceMap. https://influencemap.org/report/Big-Oil-s-Agenda-on-Climate-Change-2022-19585
International Financial Reporting Standards Foundation. (2022). Adoption of integrated reporting. https://www.integratedreporting.org/when-advocate-for-global-adoption/find-out-what-is-happening-in-your-region/
Royal Dutch Shell plc. (2019). Annual Reports and Accounts for the Year Ended December 31, 2019 (p. 316). Royal Dutch Shell plc. https://reports.shell.com/annual-report/2019/
Si, Y., Desai, D., Bozhilova, D., Puffer, S., & Stephens, J. C. (2023). Fossil fuel companies’ climate communication strategies: Industry messaging on renewables and natural gas. Energy Research & Social Science, 98, 103028.
Supran, G., & Oreskes, N. (2017). Assessing ExxonMobil’s climate change communications (1977–2014). Environmental Research Letters, 12(8), 084019.
TotalEnergies. (2019). Universal Registration Document 2019 including the Annual Financial Report (p. 486). TotalEnergies. https://totalenergies.com/sites/g/files/nytnzq121/files/atoms/files/2019_total_universal_registration_document.pdf
TotalEnergies. (2021). Universal Registration Document 2019 including the Annual Financial Report (p. 664). TotalEnergies. https://totalenergies.com/sites/g/files/nytnzq121/files/documents/2022-05/TotalEnergies_URD_2021_EN_accessible.pdf
