UBC study uses machine learning to uncover what makes great corporate culture
New research from the UBC Sauder School of Business finds that corporate culture can have a huge impact on a company’s bottom line.
If you’ve ever worked in a corporate environment, from a large company to a small boutique firm, you’ll know: corporate culture can be the difference between a wonderful workplace and an environment that employees dread.
But what makes some workplaces enjoyable while others are unhealthy and inefficient? A new study from UBC Sauder has used big data to help nail down what makes great corporate culture — and what that means for the bottom line.
Previous researchers relied on questionnaires or surveys with hypothetical scenarios to try to quantify companies’ internal culture; others used a company’s corporate social responsibility (CSR) as an indicator — but none had come up with a reliable measure.
“Corporate culture is something intangible. You can feel it and experience it when you are in the environment, but it’s very difficult to quantify,” says Kai Li (she, her, hers), the W.M. Young Chair in Finance and Professor in the Finance Division at UBC Sauder. Professor Li co-authored the study with Feng Mai of the Stevens Institute of Technology, Rui Shen of The Chinese University of Hong Kong, and Xinyan Yan of the University of Dayton. “So there was a gap that needed to be filled.”
For the study, titled Measuring Corporate Culture Using Machine Learning, the researchers used machine learning and textual analysis to examine 200,000 transcripts from nearly two decades of corporate earnings calls, which typically involve CEOs and other top executives speaking to analysts about their companies’ operations and performance.
The team first searched for the five values most often mentioned by S&P 500 firms: innovation, integrity, quality, respect and teamwork.
Relying on a time-tested concept in linguistics — that words with similar meanings tend to occur near each other in sentences — the researchers then used a complex algorithm to create a “culture dictionary” by finding hundreds of words and phrases that appeared in close proximity to those five terms. For example, among the words most likely to appear near “innovation” were creativity, excellence, passion, technology and capability. Words that often appeared near “teamwork” included cooperation, collaboration and partnership.
The algorithm also counted common phrases such as “pushing the envelope”, “proof is in the pudding” and “shoulder to shoulder.”
The researchers found that innovation was the most frequently mentioned of the five values, while integrity was the least common; innovation and quality were the most closely associated, while innovation and integrity were the least connected. Among specific industries, healthcare stood out for its commitment to integrity and teamwork, while high tech was most likely to emphasize innovation and quality.
The researchers also found that, even though organizational culture evolves over time, there was also a fair bit of stability, with the same companies ranking high in culture across many years. “For example, Procter & Gamble scored very high in innovation across all of the periods,” says Professor Li, “so their culture of innovation is relatively stable over time.”
Another major finding was that firms that rank highest in corporate culture tend to be more resilient in bad times. (Professor Li co-authored a separate study showing how companies with strong corporate culture fared better during the Covid-19 pandemic, and were more likely to support their community, embrace the shift to digital, develop new products and cut costs).
Corporate culture is also shaped by major events such as going public, going private, or through mergers and acquisitions. “We found evidence of a concept called acculturation,” says Professor Li. “In the corporate world, when a smaller company like Red Hat is subsumed into IBM, the value system of the combined entity will exhibit a trace of both firms — both the acquirer and the target.”
Interestingly, firms scoring high in innovation and respect were more likely to be acquirers, whereas firms scoring high in integrity and quality were less likely acquire other companies.
To validate their findings, the study authors looked at companies’ patents and R&D spending (to confirm innovation), product quality and safety ratings (to confirm quality), and rankings in diversity and best employer status (to confirm respect), as well as other factors.
The study is the first of its kind to apply natural language processing models to the analysis of corporate disclosures, and to use machine learning-driven textual analysis to quantify corporate culture.
Professor Li says companies that have strong corporate culture tend to be more innovative, efficient and resilient, and employees and executives tend to stay longer-term; they also tend to be more highly valued by the stock market. Businesses with poorer culture often lose talent, are less competitive and are more likely to fail.
Developing a strong corporate culture can’t happen overnight and it can cost money, says Professor Li, but the research proves that in the long run, it pays off.
“Executives shouldn't lose sight of cultivating and investing in their organization's culture in order to excel in the business world,” says Professor Li. “They shouldn't just look at short-term targets; they should invest in their people and invest in their communities. They have to play the long game.”
Interview languages: English, Mandarin