To some observers, this might have seemed like an unusual choice. But to Yifei Lu, an assistant professor of accountancy at Gies College of Business at the University of Illinois Urbana-Champaign, Nike’s decision to go with a CEO with experience in software and in e-commerce was a clue to the company’s future.
“Although industry knowledge is a very important aspect of managerial talent, it’s definitely not 100 percent,” Lu says. “A CEO’s skills can be much broader than expertise in a particular industry. It can also tell you something about the strategy change. Like this Nike example: they wanted to focus on e-commerce.”
Why Nike Hired a CEO From Outside the Shoe Industry
Lu (right) has been thinking about executive labor markets for a long time, since his first year as a PhD student when he became interested in corporate disclosure statements. While looking for a research subject, he started reading proxy statements – company filings that, among other things, contain information about executive compensation. One section of the proxy statements in particular caught his eye: peer selection, where the firm typically lists 15-20 peer firms in order to set a benchmark for executive salaries and benefits.
Some of the peer selections were expected: Delta Airlines, for example, listed other airlines, such as American and United. They also predictably listed hotel chains such as Marriott, and other companies in the travel and leisure industry. And then, unexpectedly, they listed Coca-Cola.
“At first I couldn’t connect the dots,” Lu says now. “But once I saw that Coca-Cola was in the section called transportation distribution, I started to realize, well, Delta is transporting people and Coca-Cola is transporting goods. So that makes sense, actually. They have some similarities in operations.”
Other researchers who had studied proxy statements believed that companies listed larger corporations as peers in order to justify paying their CEOs larger salaries. But Lu saw something else: by identifying peers in different industries and then hiring executives from those industries, a corporation was also indicating the direction in which it wanted to grow. Nike’s hiring of Donahoe seemed like proof that he was on the right track.
Lu discussed this insight with Ray Rui Gao, then a fellow grad student, now an assistant professor at the Marshall School of Business at the University of Southern California. Together they examined how proxy statements could improve understanding of the executive labor market and compensation. Their paper, “Aggregated Compensation Peer Group Disclosure and Managerial Labor Market Competition: A Network Analysis,” was recently published in Journal of Accounting Research.
Mapping the Executive Labor Market
Lu and Gao began their study by collecting data from the ISS Incentive Lab between 2006 and 2018, specifically which companies different firms listed as their peers. Then they created a network of corporations, dividing the relationships into three Managerial Labor Classifications (MLCs).
The first two groups were straightforward: direct peers, where both firms identified each other as peers, and indirect peers, which included peers of peers (Lu likens this to friends of friends in a social network). On average, each firm had about 20 direct peers and roughly 100 indirect peers.
Then there was the third group, the Louvain peers, named after the Louvain community detection algorithm used to identify them. Computer software uses the algorithm to group corporations based on how frequently they benchmark against one another, directly or indirectly, and sorts firms into roughly a dozen clusters each year – 12 in 2007 and 13 in 2018 – with their membership shifting over time.
“You actually see how the dots move together to form that cluster in real time,” Lu says. “It’s very fun.”
The Louvain method allowed the researchers to see more unexpected connections between companies in different industries, connections that emerged from the network as a whole rather than from any single firm’s disclosure. In 2007, for instance, food producers, restaurant and hotel chains, and transportation companies commonly fell into one Louvain group; by 2018, those industries had drifted into separate clusters.
Next, the researchers compared their MLC model with several other measures of talent transferability. They found that the MLC model was best at predicting executive job-hopping, beyond obvious cues like industry, size, or location. The executive labor market, it turned out, was different from the product market. So Nike’s hiring of Donahoe wasn’t as unexpected as it might have seemed, especially if Nike was planning to expand its e-commerce operations.
Finally, Lu and Gao looked at executive compensation within the firms in their study. They found, for instance, that in a competitive labor market, firms paid their CEOs more – mostly through equity pay with longer vesting periods, a classic retention tool – and were even more willing to pay for "luck," rewarding windfalls beyond the CEO's control rather than risk losing talent to rivals.
What Corporate Boards Can Learn
For Lu, this study provides an important lesson for corporate boards who are looking to keep executives from leaving. “If a board is trying really hard to retain talent, the first thing they need to know is who they’re competing with. It goes beyond the 15 or 20 firms that they select. They need to benchmark those potential competitors as well. And once they raise their pay to the market level, they can retain their CEO.”
The most important lesson from this study, Lu said, is that people need to realize that the labor market is distinct from the product market. In the future, he’d like to make the MLC model public and update it annually so that companies will know which other firms are in the same labor market. “If that data is available,” he says, “I believe people would look into it.”