Diversity Tokenism

Abstract

Using data from over 4,000 Black Lives Matter (BLM) protests across 600 U.S. counties from 2014 to 2021, we examine how BLM activism shapes corporate diversity at different organizational levels. We develop an approach integrating OpenAI's GPT-4 with Chain-of-Thought prompting to classify race and ethnicity. In our validation tests, this method achieves higher accuracy than several tested open-source algorithms. Our main findings are as follows. First, although firms headquartered in protest-affected counties add more Black directors, particularly in larger protests, this gain appears to largely offset the representation of other non-Black minority directors. Second, these board-level shifts do not consistently extend to executives or the general workforce. In contrast, a gap may emerge between a firm's workforce composition and local labor-market demographics, particularly in the representation of Black employees. This pattern is consistent with diversity tokenism, which suggests firms may prioritize high-visibility board appointments and potentially downplay broader, transformative change. Our findings indicate that although board-level diversity gains are highly visible and attract notable public attention, they may not be accompanied by an organization's transformative commitment to company-wide diversity.