Anne Marie Knott
Quite thought provoking.
We have a slight change for our April Driving Org Growth class at Kellogg because of a conflict with Easter Sunday. We will begin the class on Monday, April 2, 2018 and end on Thursday at 4:00PM.
As this will be my last posting this year, I would like to wish all of you the best for the New Year!!
This has been a remarkable year for the markets. The S&P and the Dow indexes are up 18% and 19%, respectively. But this run-up isn’t based on solid business foundations. Quarterly profits have only increased 5% since 2012, but investors’ valuations of those profits (as measured by earnings per share) has increased 59% over the same period. What’s behind the disconnect? Some argue that profits are stagnant because of short-termism—that decades of focusing on current profits over long-run innovativeness has resulted, now, in companies that are hollowed out.
Indeed, a study by Rachelle Sampson and Yuan Shi found that company short-termism is negatively correlated with innovativeness, measured as RQ (“research quotient,” a measure of the return on R&D investments). Investors punish companies with a short-term orientation by applying higher discount rates to them, which increases the cost of capital for those companies. In contrast, companies with a long-term orientation are rewarded with a lower cost of capital, which allows them to afford more innovation—a virtuous cycle.
Most attempts to combat short-termism are flawed because they focus on changing CEO behavior through some combination of pleading and incentives. While well-intentioned, these efforts fail to recognize that CEO behavior is largely circumscribed by firm structure. In particular, there are three widespread, interrelated structural trends that have fostered short-termism and reduced corporate innovativeness: increased hiring of outside CEOs (particularly from the late 1980s through the 2008 recession); the decentralization of R&D (over a similar time frame); and a focus on the “development” side of R&D rather than the “research” side. Below, I’ll expand further on these trends and explain how reversing them would reduce short-termism and revive growth.
Instead of hiring outside CEOs, hire insiders—or at least CEOs with domain expertise.
One trend that has contributed to short-termism and lower innovativeness is the increased prevalence of outside CEOs. From 1970 to 2004, the percentage of CEOs hired from outside the firm increased from 12% to 39%.
While outside CEOs are valued because they bring new perspective, my colleague, Trey Cummings, and I believe they impose a hidden cost to innovation at firms whose growth derives from R&D (roughly 49% of firms). We came to this conclusion through interviews with CTOs across a range of industries, which we conducted as part of an NSF study to identify factors explaining differences in firms’ RQ.
A recurring theme in those interviews was bemoaning major changes in R&D strategy that occurred as a consequence of new, often outside, leadership. In these stories, firms shifted from an orientation of “R&D as a driver of growth” to “R&D as an expense.” What was reported to happen as a consequence of this shift was a steady decline in firms’ R&D intensity (R&D/Sales) and a corresponding decline in firms’ R&D capability. In other words, the new leader’s disinvestment cut meat as well as fat.
While the identity of the interviewed firms is confidential, it is easy to find similar examples from publicly available accounts in other firms. Consider GE during Jack Welch’s tenure, Trimble Navigation under Steve Berglund, or IBM under Lou Gerstner. In all three cases, our analysis shows, there was a decline in R&D investment followed by a decline in the returns on that investment (RQ).
We also know some details about how R&D strategy changed in these cases: GE shifted to strategy of divesting businesses in which they were neither number one or two in their markets (televisions, semi-conductors, and aerospace) and expanding into businesses that didn’t rely on R&D (NBC, GE capital); Trimble shifted from a strategy of developing its own technology to one of acquiring other firms for their technology; and IBM shifted to a strategy of reducing R&D while patenting the stock of existing innovation (increasing patents almost 500%). The shift in patenting policy was not to protect innovations, but rather to license them and/or to use them as chips to gain access to other firms’ technology.
Why do these shifts occur? We believe, and now have correlative evidence, that it’s because outside CEOs are less likely to possess the technological domain expertise necessary to drive growth from R&D. When CEOs lack this expertise, they are more likely to manage R&D “by the numbers,” despite the fact that those numbers are more elusive than those for capital and advertising. Indeed, we found that companies with outside CEOs have lower innovativeness as measured by RQ, and that those effects become more pronounced the more R&D intensive the company is and the more technologically different it is from the CEO’s prior company.
Note the solution is not to avoid outside CEOs. There are many reasons companies benefit from hiring an outsider, such as to effect change. Moreover, not all outside CEOs lack domain expertise (e.g., CEOs from rival firms); conversely not all inside CEOs have it (CEOs promoted from finance). Rather, the solution is to ensure that companies whose growth derives from R&D hire CEOs with technological domain expertise.
Instead of decentralizing R&D, recentralize it.
One of the changes we learned outside CEOs make is decentralizing R&D. Decentralizing is a natural consequence of managing by numbers because it shifts R&D investment decisions to divisions, where R&D investment can be more readily linked to outcomes. The problem with giving division managers control of R&D, is that their compensation is typically based on division profits (which they largely control), rather than on the company’s market value (over which they have little control).
The distinction between current profits and market value is important because market value takes into account future profits, so in principle it captures the long-run returns to R&D. In contrast, current profits penalize R&D, because accounting rules require R&D to be expensed. This means all R&D is subtracted from operating income in the year it’s expended, while the payoffs to R&D don’t occur until future periods. Thus the further out the fruits of R&D, the less likely operating divisions are to conduct it.
My research indicates that companies in which R&D is decentralized have 40 to 65% lower RQ than companies with centralized R&D. This means they generate less revenue, profit and market value per dollar of R&D. Perhaps more problematically, a study by Nick Argyres and Brian Silverman found that decentralized R&D produces innovations that have a smaller and narrower impact on subsequent innovation.
Instead of over-focusing on “development,” shift the portfolio back toward “research.”
The logic underlying the push for decentralization and greater relevance of R&D is that R&D directed by divisions will be more responsive to what the customer wants. While responding to the customer sounds completely unobjectionable, its vulnerability is best captured in the Steve Jobs quote, “A lot of times, people don’t know what they want until you show it to them.”
When R&D is directed by divisions, the company fails to invest in the early stage technologies that open up new opportunity. Again, this occurs because division manager compensation is tied to division profits. So in addition to causing R&D to disproportionately favor development rather than research, it also causes R&D to be parochial to the funding division. This is because there is little incentive to conduct R&D that benefits multiple divisions.
A real-world illustration of this parochialism comes from the “Organization 2005” initiative at Procter and Gamble (P&G). The initiative decentralized R&D in an effort to make the big company “feel small” (e.g., to provide greater managerial control and break bureaucratic inertia), as well as to more closely fit the needs of customers.
The result was a dramatic shift at P&G from 90% centralized control of R&D in the 1990s to 90% decentralized control of R&D by 2008. In the words of then-R&D chief Bruce Brown, making business-unit heads responsible for developing new items inadvertently slowed innovation by more closely tying research spending to immediate profit concerns. Relatedly it led to smaller, more incremental innovation. While the number of innovations doubled, the revenue per innovation decreased 50%.
One reason revenue per innovation decreased is that early stage research dwindled. Prior to the decentralization, P&G was known for creating entirely new product categories: first synthetic detergent (Dreft in 1933), first fluoride toothpaste (Crest 1955), and more recently: Febreeze odor fresheners (1998), Swiffer (1999) and Crest Whitestrips (2001). Following the decentralization, and the shift away from research, P&G failed to introduce a single blockbuster.
This is not a surprise. In a study with my colleague, Carl Vieregger, we found that across the five most common configurations of R&D allocation, the configuration with the highest RQ allocates twice as much of its total R&D investment to basic and applied research than does the average company.
While a long horizon is necessary for innovation and growth of companies (as well as the economy), a call to resist the forces of short-termism is unlikely to yield results. This is because short-termism is now built into company structure, due to the rise in outside CEOs, a trend toward decentralized R&D, and a move away from basic research. Each of these trends on its own is associated with lower RQ. Moreover the trends are interrelated: Outside CEOs tend to decentralize R&D, and decentralization tends to decrease basic research.
But fortunately, these trends can be reversed. Boards in firms whose growth derives from R&D can begin taking domain expertise into account when hiring CEOs. Further, firms can begin recentralizing R&D, and increasing the levels of basic and applied research. Prudently applying these prescriptions should increase a company’s RQ, and accordingly its revenues, profits, and market value from R&D. This will generate fundamental growth of firms, rather than merely growth in their valuations.