This post is a summary of the findings from data science research on the financial materiality of labor rights carried out by undergraduate and graduate students of the Data Science Institute of Columbia University under the direction of Rights CoLab Cofounders Joanne Bauer and Paul Rissman. It is part of a multi-year collaboration between Rights CoLab and the Institute that began in January 2020. For this round of the research, Sabrina Jade Shih was the lead researcher on Diversity, Equity, and Inclusion (DEI) topics, and Yuwen Zhang coordinated a team of students — Catalina Ramos, Eubin Park, Jay Trevino, and Jittisa Krapayoon — on Labor Conditions in Supply Chains (LCSC) topics. (See Meet our Data Team.) Alexis Cheney consolidated the two research reports into this post.

Summaries of prior research can be found here.

The notebook containing the code can be accessed in a Github repository here.

Introduction

In 2015, the New York Taxi Workers Alliance filed a complaint with the New York Office of the Attorney General, sparking an investigation into wage theft and in 2023 resulted in Lyft and Uber being ordered to pay a combined $328 million in settlement fees.1 Cases like this one highlight the financial and reputational costs to companies that fail to respect labor rights. In response to investor interest and gaps related to human capital management disclosure,2 the Sustainability Accounting Standards Board (SASB) launched its Human Capital Research Project in 2019 to “assess the scope and prevalence of various human capital management themes across SASB’s sectors and within its 77 industries to develop a solid evidenced-based view on this cross-cutting theme.”3 This project and the SASB standards themselves have since been integrated into the International Sustainability Standards Board (ISSB), which announced in April that it would be building upon the SASB industry standards to “research disclosure about risks and opportunities associated with: biodiversity, ecosystems and ecosystem services; and human capital.”4

Starting in 2019, until SASB was folded into the ISSB, Rights CoLab led an initiative, together with 22 civil society partners, to provide subject matter expertise to SASB’s human capital work, with a goal of  better integrating “human rights at work” into the SASB Standards. Partnering with Columbia University’s Data Science Institute, Rights CoLab worked with graduate students to apply data science techniques to large datasets, aiming to uncover evidence of the financial materiality of human capital and human rights issues and to use these findings to inform the development of new financially material metrics. 

This post demonstrates how using data science can help us identify new human capital metrics linked to investor interest and financial impact. It discusses findings in support of metrics related to the DEI practices — particularly recruitment practices — of formerly incarcerated workers and older workers, as well as metrics related to practices affecting Labor Conditions in Supply Chains (LCSC), particularly regarding the alternative workforce.5

Ultimately, this work serves as a proof of concept that data science techniques can be used to isolate and ascribe financial materiality to new metrics. It further suggests that data science provides a viable route for informing the SASB industry standards in the context of dynamic materiality.

Methods

We began the research by constructing a term dictionary of corporate practices linked to potentially material financial outcomes. We identified terms through an extensive literature review. For DEI metrics, we added to the term dictionary terms from 429 metrics found in 21 influential DEI frameworks that Rights CoLab mapped for a  2022 report, “What Is DEI? Market Signals of Diversity, Equity, and Inclusion.” To aid our manual analysis, we labeled each term according to the themes established in this report. For supply chain metrics, we collaborated with the Center for Business and Human Rights at NYU Stern, which developed metrics based on their fieldwork in Bangladesh on practices that lead to financial risk and augmented our term dictionary with these metrics.6 Using these term dictionaries, we then applied the algorithm to a set of databases, including FactSet’s TruValue SASB Spotlight DataFeed,7 shareholder resolutions, and earnings call transcripts, to identify practice-outcome co-occurrences.

If at least a practice term, an outcome term, and a context term (see box above for definition of terms) appear within 3 to 15 words of each other in the text, depending on the relationship between the topics, the algorithm flags the article as potentially containing a financially material practice related to DEI or LCSC. Next, for each topic, DEI or LCSC, we ranked the co-occurrences by salience and checked for relevance and accuracy. See the Appendix for the paragraphs and terms that form the basis  of our conclusions. For more on our approach to the research, see “Our Data Science Methodology”.

Given the rapid evolution of data science methods and that some of our data sources date back to 2022, the relationships we discuss below should not be considered definitive. Instead, we present select examples of research findings and invite other researchers to build upon these methods to validate our findings and uncover new financially material relationships. It is noteworthy that the examples we present here –  a sampling of the most interesting relationships our method identified – are from the United States. This resulted despite the fact that our largest data set, FactSet’s TruValue SASB Spotlight DataFeed, is global, covering “all major positive and negative ESG events according to the Sustainability Accounting Standards Board’s 26 categories for 267,000+ companies.”8 Therefore another area of exploration for future data science researchers is identifying data sources and methods that can surface evidence from other geography contexts. 

Financial Materiality and DEI: Criminal Background Checks 

Our co-occurrence method and DEI keyword dictionary flags news articles and shareholder resolutions tying criminal background checks during the hiring process to material impact. The identified cases occurred in the United States, where it is illegal to deny employment based on minor, outdated, or job-irrelevant criminal convictions. Due to the misuse of criminal background checks, the companies paid fines or reached other settlements. Overall, the texts suggest that the use of criminal background checks during the hiring process led to allegations of racial and ethnic discrimination with negative financial consequences for the companies involved. The majority of the identified texts relate to the Multiline and Specialty Retailers & Distributors industry.

To demonstrate how the co-occurrence method flags an article, let’s look at the following passage from an April 2018 Reuters article:9

The terms used to flag this news article, which highlights the financial impact of discriminatory background checks at Target, are bolded and highlighted according to the color key described in Table 1.

Table 1: Key for the Term Pattern in the Reuters Article

The co-occurrence method also flags articles reporting on the misuse of criminal background checks and their financial consequences at Pier One and Macy’s. Excerpts from these articles can be found in Appendix parts I and II.

The algorithm also flagged three shareholder resolutions with a practice or outcome term linked to hiring practices of individuals with criminal backgrounds. One resolution was filed with Amazon in 2018 and the other two were filed with Microsoft in 2021 and 2022. Identifying these practices in shareholder resolutions is significant because these texts strongly reflect investor interest, indicating financial materiality. Each proposal argues that the company’s practices related to hiring and retaining employees with criminal records expose them to material risks: legal liability for racial discrimination under the U.S. Civil Rights Act of 1964 or under the Equal Employment Opportunity Commission guidelines, and reputational damage for failing to meet their own public commitments to DEI. Appendix part III provides an excerpt of the 2022 Microsoft shareholder resolution linking the discriminatory use of criminal background checks to financially material consequences.

Financial Materiality and DEI: Age Discrimination

The co-occurrence algorithm and DEI keyword dictionary also identified news articles and shareholder resolutions linking certain recruitment methods alleged to involve age discrimination to financial impacts. Specifically, university on-campus recruitment programs and online advertising directed at certain demographic groups while excluding others. Similar to criminal history, age discrimination is illegal in the US through the Age Discrimination in Employment Act (ADEA) of 1967. 

One notable case flagged was a lawsuit against Eli Lilly, a pharmaceutical company, for allegedly biased hiring practices related to its on-campus recruitment programs for sales representatives. The lawsuit claimed that these practices discriminated against older workers and led to a class action seeking back pay, front pay, punitive damages, and injunctive relief. For details on the terms that flag this article, see Appendix part IV. This example illustrates how company disclosures about recruitment strategies and applicant demographics can reveal potentially discriminatory practices with associated financial risks. 

The algorithm also identified the use of online recruitment advertising that intentionally targets one demographic group while hiding the ads from others. The U.S. Equal Employment Opportunity Commission (EEOC) ruled that seven companies in various industries used Facebook ads in a way that discriminated against older workers, leading Facebook to change its ad targeting algorithm. See Appendix part V for the terms that appear in this article on Eli Lilly. Since recruitment practices such as these can lead to regulatory and legal risk (and, in Facebook’s case, business model risk), their disclosure may be financially material. A potential metric could center around the company’s outreach methods (in-person vs. online recruiting and advertising practices), and how well designed they are to enable companies to reach a diverse pool of qualified candidates.

Financial Materiality and LCSC: Supply Chain Management 

The passage of the European Union’s Corporate Sustainability Due Diligence Directive on March 15, 2024 made the responsibility of companies to reduce human rights and environmental risks in  their supply chains a matter of legal liability, and therefore financially material. In its April announcement, referenced above, the ISSB also recommended adding supply chain management to its human rights–related research project.10 The financial materiality of corporate supply chain management will likely garner intense investor interest in the coming years.

Academic research has likewise examined the impact of supply chain visibility on corporate financial performance.11 Some studies have demonstrated that enhanced supply chain visibility can lead to improved operations, while others have explored the relationship between financial performance and “ESG incidents” in supply chains.12 In line with this research, our co-occurrence method identified investor and corporate management interest in supply chain visibility, based on various sources including news articles, shareholder resolutions, and earnings call transcripts. 

News article example: a 2019 Electrek news article describes Volvo Cars’ practice of using blockchain technology to trace cobalt in its supply chain. The practice was implemented after Amnesty International’s 2019 allegations that Volvo Cars production of EV batteries involved child labor. (See the article excerpt in Appendix part VI). The article suggests the financial materiality of EV battery supply chains since failure to do so could result in reputational damage. 

Shareholder resolution example: A shareholder resolution filed in 2016 to The Kroger Company, a supermarket chain, requesting the implementation of a Human Rights Risk Assessment framework in order to prevent human rights violations within its supply chain, thereby reducing the risk of civil fines resulting from such violations. (See the shareholder resolution excerpt in Appendix part VII).

Earnings call transcript example: On a 2021 American Software Inc. earnings call, Allan Dow described their digital supply chain traceability solution, which enables companies to prevent forced labor in the supply chain. (See the earnings call excerpt in Appendix part VIII).

Financial Materiality and LCSC: Alternative Workforce

Among the most salient subjects for a company’s human capital management is its alternative workforce. During its human capital research project, the SASB Standards Board took an interest in the topic, which it defined as:

a provisional group of workers who work for an organization on a non-permanent basis, which include freelancers, independent professionals, temporary contract workers, independent contractors, or consultants (but excludes workers who form part of an outsourcing arrangement with a third-party as part of the alternative workforce).13

The co-occurrence method flags that “alternative workforce” may also be financially material for a company’s supply chain through several channels. One channel relates the alternative workforce to the heightened risk of exploitation given independent contractors’ ineligibility for various labor protections, including collective bargaining agreements and grievance mechanisms. Limited labor protections sometimes link employers to wage theft allegations. Our algorithm flags several cases including the following:

XPO, a trucking company which paid drivers nearly $30 million to settle class action lawsuits, is flagged in a news article and a shareholder resolution. (See Appendix parts IX and X for the excerpts.)

Woolworths’ Tasmanian supermarkets in Australia, which violated workplace laws when procuring and contracting cleaners in gig work, leading to a Fair Work Ombudsman inquiry into the supermarket. (See Appendix part XI for the article excerpt.)

The co-occurrence method also reveals an association between employing non-permanent workers and increased threats to worker safety. For example, the method flags an earnings call from Scandi Standard, a Swedish food processor, which attributed an increase in the proportion of temporary workers during the COVID-19 pandemic to an increase in adverse safety incidents. See Appendix part XII for the earnings call excerpt on Scandi Standard.

Conclusion

This report demonstrates the promise of data science to identify new, financially material human capital metrics. The co-occurrence data science method produces findings that provide directions for shaping financially material DEI metrics related to certain recruitment practices affecting formerly incarcerated and older workers. In particular, the method suggests the salience of metrics on the use of criminal background checks in recruitment processes and the disclosure of recruitment strategies. The method also identifies potential for financially material LCSC metrics related to supply chain management and employing the alternative workforce. Relevant metrics could involve the use of technology to trace supply chains in the former case and percentage of alternative workforce of suppliers or  contract conditions  in the latter. 

Rather than promote the development of one metric over another, this post seeks to demonstrate the power of data science to suggest additional ISSB human rights metrics. We invite data scientists to explore the data sources, experiment with data science methodologies, and discover additional financially material practices that could inform the ISSB’s human capital metrics. 

To learn more about  this project, contact Rights CoLab co-founders Paul Rissman at paul@rightscolab.org or Joanne Bauer at joanne@rightscolab.org.

Appendix

See Appendix here.

Endnotes

  1. Jonathan Stempel, “Uber, Lyft to pay $328 million to settle New York wage theft claims,” Reuters, November 2, 2023, https://www.reuters.com/business/autos-transportation/uber-lyft-pay-328-million-settle-new-york-wage-theft-claims-2023-11-02/. ↩︎
  2. Thomas Riesenberg (SASB Director of Legal and Regulatory Policy), letter to SEC Secretary Vanessa Countryman, October 17, 2019, https://www.sec.gov/comments/s7-11-19/s71119-6313644-193668.pdf. ↩︎
  3. “Human Capital: Project Overview,” SASB, last modified June 22, 2023, https://sasb.ifrs.org/standards/process/projects/human-capital/. ↩︎
  4. “ISSB to commence research projects about risks and opportunities related to nature and human capital,” IFRS, April 23, 2024, 
    https://www.ifrs.org/news-and-events/news/2024/04/issb-commence-research-projects-risks-opportunities-nature-human-capital/. ↩︎
  5. ‘Alternative workforce’ is defined in SASB’s Human Capital: Preliminary Framework as, “a provisional group of workers who work for an organization on a nonpermanent basis, which include freelancers, independent professionals, temporary contract workers, independent contractors, or consultants (but excludes workers who form part of an outsourcing arrangement with a third-party as part of the alternative workforce).” (7) ↩︎
  6. This work has been published as Natasja Sheriff-Wells and Chana Rosenthal, “A Broken Partnership: How Clothing Brands Exploit Suppliers and Harm Workers — And What Can Be Done About It,” April 2023, NYU Stern Center for Business and Human Rights. https://bhr.stern.nyu.edu/publication/a-broken-partnership-how-clothing-brands-exploit-suppliers-and-harm-workers-and-what-can-be-done-about-it/. ↩︎
  7. A dataset of ESG events tied to over 230,000 public and private companies globally since 2016 based on 100,000 sources including international news, government agencies, and NGOs in over 30 languages. ↩︎
  8. “FactSet Truvalue SASB Spotlight DataFeed,“ FactSet, accessed August 9, 2024, https://www.factset.com/marketplace/catalog/product/sasb-spotlight-datafeed. ↩︎
  9. “Target Settles Claims It Screened Blacks, Hispanics out of Jobs | Reuters,” accessed August 5, 2024, https://www.reuters.com/article/target-lawsuit/target-settles-claims-it-screened-blacks-hispanics-out-of-jobs-idUKL2N1RI0XG/. ↩︎
  10. “Staff Paper: Projects to add to the work plan ” IFRS, April 2024, https://www.ifrs.org/content/dam/ifrs/meetings/2024/april/issb/ap-2-agenda-consultation-projects-to-add-to-the-work-plan.pdf ↩︎
  11. Supply Chain Visibility refers to “the ability to trace the points of origin of materials used in a product” (Lee & Rammohan, 2017). It entails activities such as firms developing public supplier lists and monitoring supplier violations of codes of conduct. ↩︎
  12. Caroline Swift, V. Daniel R. Guide Jr., and Suresh Muthulingam, “Does Supply Chain Visibility Affect Operating Performance? Evidence From Conflict Minerals Disclosures,” Journal of Operations Management 65, no. 5 (2019): 406–429, https://doi.org/10.1002/joom.1021.; Xuanpu Lin, Guoman She, Aaron Yoon, and Haoran Zhu, “Shareholder Value Implications of Supply Chain ESG: Evidence from Negative Incidents” (June 30, 2023). http://dx.doi.org/10.2139/ssrn.4500888. ↩︎
Photo by Uriel SC on Unsplash

Related Posts