Making the proper business case for diversity

Here’s a note from Alex Edmans:

McKinsey’s (2015, 2018, 2020) reports claiming that diversity improves performance have already been discredited. They’ve doubled down in their latest report, entitled “Diversity Matters Even More”, claiming that “the business case is the strongest it has been”. While they are not alone in producing flimsy research on diversity, the concerns here are particularly serious for the following reasons:

1. The Errors are Particularly Basic

They correlate diversity in 2022 with financial performance in 2017-21. This makes it particularly likely that financial performance causes diversity, rather than diversity causes financial performance. This is an even more basic error than made by other papers (e.g. failing to control for industry).

Yet their belief that diversity improves performance is so entrenched, and their confirmation bias so strong, that they dismiss this possibility without any analysis, even though their research design makes it very likely: “It is theoretically possible that financial outperformance enables companies to achieve greater levels of diversity … However, in practice this seems unlikely.” This is the opposite of diversity of thought.

2. The Errors Have Been Extensively Documented

Green and Hand (2021) previously showed that the McKinsey studies are irreplicable even with their chosen performance measure (EBIT) and preferred methodology. Moreover, there is no link between diversity and other performance measures, or with more established methodologies that do not throw away data. McKinsey are aware of the critique since Green and Hand called one of the authors.

Yet they do not address any of the critiques, using the same flawed methodology as prior papers for “consistency”. But continuing to do something wrong and refusing to heed suggestions for improvement is the opposite of diversity of thought. Indeed, McKinsey have gone the other way and obfuscated their methodology (they don’t even say how they collect their sample) to prevent future replication.

3. The Errors Contradict McKinsey’s Own Research

McKinsey produces a gamut of papers claiming to have identified the secret sauce that improves performance. Yet they control for none of these factors in their tests. They simply link diversity to whether performance is above the industry average with no controls.

Even if you are an advocate of DEI (as I am, properly practiced), you should be concerned about such papers. They blind you to the fact that properly-measured DEI, that doesn’t reduce a person to their gender and ethnicity, may be genuinely correlated with performance, or that DEI initiatives can be justified on social grounds.