^{Disclaimer: Not Investment Advice}

^{‍The content provided on this blog is for informational and educational purposes only. It is not intended to be, nor should it be considered as, financial or investment advice. The opinions expressed here are based on personal experiences and research, but they do not take into account the specific financial circumstances, goals, or needs of any individual.}

^{‍Before making any investment decisions, you should consult with a qualified financial advisor or conduct your own research. Investing involves risk, and you should be aware that you could lose some or all of your invested capital. The author of this blog is not responsible for any financial losses you may incur as a result of relying on the information provided herein.}

‍

Recently we decided to take a brief look at Tesla's (NASDAQ: TSLA) sustainability disclosures using our AI platform Responsible Capital.  Tesla are renowned under the leadership of Elon Musk for being unafraid to be different and not follow the crowd and so we wanted to see how well our AI could perform extracting key metrics from PDF format sustainability reports for quantitative comparisons over a period of a few years.

‍

As always the challenge in analysing these disclosures is that even analysing the standalone performance of a single company can be challenging as from year to year the style of reporting, methodologies and metrics of choice can change. We are still some way away from reporting standards such as ISSB, CSRD / ESRS, TNFD and GRI adopting specific, consistent, structured, machine readable data standards like we see in other open data contexts. Historically Natural Language Processing (NLP) was widely employed to try and solve this problem but increasingly Large Language Models (LLMs) are taking over.

‍

Another challenge is identifying where metrics are idiosyncratic (completely unique to Tesla such as self driving car crashes or Elon related mental meltdowns!), typical industry metrics or core global disclosure data points well covered by frameworks. Fortunately Retrieval Augmented Generation (RAG) architectures such as Responsible Capital can to a large extent solve this challenge.

‍

We wanted to assess the degree to which our AI could rapidly identify any notable trends or changes in reporting approach adopted by Tesla - perhaps in response to evolving reporting standards, increased data capture and measurement at Tesla or possibly in response to Elon's more recent anti-esg backlash railing against the 'woke mind virus'....who knows.

‍Our new Indicator module

‍

Responsible Capital’s AI capabilities extract indicators from disclosures and conform these to a global indicator data model. It's a relatively new feature we developed to overcome the plethora of global reporting standards, lack of standardisation in digital formats previously mentioned. The freedom of disclosing companies to adopt and change their disclosure formats as they choose can be helpful if you have something to hide, not so ideal if you are looking to consistently compare performance across a peer group, industry or even historically for a single company like Tesla.  Hence we developed an indicator module to deal with this issue until if / when reporting standards achieve sufficient further conformity (we’re not holding our breath here).

‍

‍What did we find?

‍

Below is a spreadsheet containing a core set of disclosed indicators for comparison our AI was able to extract from 4 years of Tesla impact reports (2020-23) so you can see the results yourself.  The TLDR summary is:

‍

- Tesla's impact report length increased by 200% between 2019 and 2021 as they gathered more data to a peak in 2021 when they started to disclose new indicators such as scope 1,2,3 emissions. 2021 was the longest report Tesla have issued to date at a whopping 144 pages.

- Since 2021 there has been a consistent decline in length of Tesla's impact reports - averaging an almost 50% decline in length per year down from 144 pages in 2021 to 38 pages in 2023.  Perhaps this is the ruthless efficiency of Elon in action cutting down on wasteful whitespace, cliched imagery and charts?  Not entirely....

- There has also been a trend of omitting certain past disclosures and changing the format of the report.  In addition to numerous charts and statistics the 2021 report contained a 20 page long appendix with tables of neatly laid out metrics.  The appendix disappeared in the 2022 report in favour of interweaving statistics into general narrative and imagery.

- Tesla also stopped publishing waste disposal, diversion and recycling statistics after 2021. Previously they published quite detailed statistics about total, hazardous and non hazardous waste generated and diverted overall on a per vehicle basis including recycling levels of key battery metals cobalt, nickel and copper.  All are omitted from 2022 and 2023's reports - at least 16 data points are omitted on waste alone.  It's quite stark that mentions of the word "waste" (and related synonyms) decline from 33 in 2021 report to just 1 mention in 2023.

‍

Download the full csv file of extracted indicators

‍

Mentions of waste and waste related words 2021 - 2023 

‍

‍

- On water Tesla also cease disclosing from 2022 onwards its total water withdrawal in cubic meters.  They do however continue to disclose the water withdrawal per vehicle produced which shows a nice decline from 3.02 m3/vehicle in 2021 to 2.48 m3/vehicle in 2023.  Given Tesla produced 1.3m vehicles in 2022 we can calculate that around total water withdrawal increased from 2.8m cubic meters to 3.3m cubic meters in total was withdrawn globally in 2023 - an 18% increase since 2021.

- Similarly on the social indicator side Tesla have been embroiled in a number of labor controversies in recent years so this is an area of particular focus for investors. Tesla historically disclosed a wide range of data points here including Days Away from Work, Restricted Time (DART), American Society for Testing and Materials (ASTM) Level One Rate and the Global Recordable Injuries per 1000 cars - all ceased to be disclosed in the 2023 report.  

‍Conclusion

‍

In conclusion as with many Tesla / Elon related trends it can be challenging to understand the rationale.  Perhaps there was a simple change in reporting strategy and style, perhaps the intern publishing the reports online omitted several pages by accident or perhaps there was a deliberate strategy to shift the focus from certain disclosure metrics to others.  Fundamentally however TNFD and other frameworks are leading to growing interest from investors in performance on waste, water and recycling rates so Tesla should focus on disclosing as consistently as possible to make the process of downstream analysis as efficiently as possible.  For a global technology and data leader this should be easy to achieve.

‍

Another conclusion is that AI tools can achieve significant time savings - particularly in analysing idiosyncratic metrics not easily available via terminals and ESG data products. If you are interested in testing these capabilities out for yourself you can sign up for a free trial of Responsible Capital.

‍

Sources:

‍

https://www.tesla.com/ns_videos/2023-tesla-impact-report-highlights.pdf

https://www.tesla.com/ns_videos/2022-tesla-impact-report-highlights.pdf

https://www.tesla.com/ns_videos/2021-tesla-impact-report.pdf

https://www.tesla.com/ns_videos/2020-tesla-impact-report.pdf

https://www.tesla.com/ns_videos/tesla-impact-report-2019.pdf

‍