Source: Li, Mai, Shen & Yan (2021) — “Measuring Corporate Culture Using Machine Learning”
At first this sounds like a technical finance question. Then you look closer and realise it is also a question about attention, incentives, and whether people are using information or just being impressed by it.
Corporate culture is one of those concepts that every business leader talks about and almost no researcher can measure. Culture surveys are expensive and biased toward what people say rather than what they do. Ethnographic approaches don’t scale. This paper offers a different path: using machine learning on earnings call transcripts at scale to create quantitative culture scores. The researchers use word embedding models — which capture semantic relationships between words based on how they co-occur in text — to construct a culture dictionary for five corporate values: innovation, integrity, quality, respect, and teamwork. Applied to 209,480 earnings call transcripts from over 62,000 firm-year observations, the resulting scores provide a quantitative, systematic measure of corporate culture that can be studied empirically. First, the innovation culture score captures something distinct from standard proxies like R&D spending and patent counts. These correlate with the ML-based score but don’t explain all of it — firms with similar R&D budgets have meaningfully different innovation culture scores. Culture is not just investment; it’s an orientation that affects how resources are deployed. Second, the culture scores predict a range of business outcomes: operational efficiency, risk-taking behaviour, earnings management, executive compensation design, firm value, and deal activity.
So I would not read this as a neat technology story. It is a messy information story. The tools improve, but people still have to decide what deserves attention. Annoying, but true.