Can Words Speak Louder Than Actions? A Text Analysis Based Evaluation of the UK Modern Slavery Act

Abstract

There are approximately 40.3 million victims of modern slavery worldwide, and up to $354 billion of products produced by modern slavery victims that are imported by G20 countries annually. As awareness of modern slavery has spread, many countries have begun passing legislation to address risks of modern slavery domestically and mandate business transparency on their international supply chains. The motivating question of this thesis is, when public policy mandates disclosure on corporate social responsibility or human rights, does the increased transparency have any material consequence on the issue at hand? To answer this question, this thesis studies one such policy, the UK Modern Slavery Act, and evaluates whether the text of corporate disclosures that are produced by this transparency law has any meaningful relationship to company action on modern slavery risks. This work uses multinomial inverse regression for text analysis to assign companies a labor performance score based on their modern slavery statement. These corporate labor scores are then regressed against financial indicators to assess whether there exists a financial incentive for improved labor conditions. Based on this analysis, I find that text can be a reliable predictor of corporate action, as proven by the validity of my text analysis model. Furthermore, text analysis can be used to distinguish between companies that use vague and generic language to meet the bare minimum requirement for disclosure and companies that take the threat of modern slavery more seriously. For the financial analysis, however, there is no statistically significant relationship between labor performance scores and my chosen metrics of financial performance, suggesting a greater need for the public and government bodies to put pressure upon companies to improve their supply chains, because the businesses themselves may not readily understand the financial incentives. There are two key implications from this thesis’ findings. First, from a methodological perspective, text analysis has never previously been used to score companies numerically on corporate behavior. Current corporate benchmarking is the product of time- and resource-intensive qualitative research. The text analysis based method of scoring companies in this paper can serve as a model to scaling up company ratings and drive improvements in corporate performance on important social issues. Second, from a policy standpoint, this thesis finds that corporate transparency laws like the UK Modern Slavery Act can serve as a starting point to using companies to create material change. The increased transparency from this act does create statements that can be meaningful and impactful. However, there are a variety of more stringent laws that policymakers can enact to penalize companies for human rights abuses and require greater disclosures to achieve the ultimate goal of eradicating modern slavery.