SMITH BRAIN TRUST — Is the influential theory of "disruptive innovation" bunk? Or to put it in a less specific and blunt way: Are businesses — and business professors — too quick to accept as fact theories that aren't supported by rigorous data?
Both propositions are true, says Brent Goldfarb, a professor at the University of Maryland's Robert H. Smith School of Business, who made that provocative case in a major presentation at the Academy of Management conference last month in Vancouver, British Columbia. His argument is backed by a major new article in the Fall 2015 issue of MIT Sloan Management Review, co-authored by one of Goldfarb's sometime collaborators, Dartmouth's Andrew A. King.
The concept of "disruption" is everywhere these days, sometimes serving as shorthand for any competition (usually digital) faced by an entrenched company. But as sketched out by the Harvard professor Clayton M. Christensen in the 1990s, and later elaborated in such books as The Innovator's Dilemma and numerous articles by Christensen and co-authors, it describes a very specific process: Upstart companies adopt a new technology that appeals to a niche market rather than the market that dominant players focus on. (In Christensen's first studies, these were disk-drive companies that moved to disks smaller than the then-norm of 8 inches.) The dominant players have the technological savvy to compete, but, instead, they listen to their core customers and fail to invest in the new technology. The new technology then improves at a rapid rate, and the upstarts displace the older companies.
Whether those dynamics played out in the disk-drive industry has itself been a subject of dispute. (Goldfarb and Dartmouth's King do not believe they did: Far from experience leading to myopia, King and Christopher L. Tucci found in a 2002 paper, disk-drive firms with the most experience in existing markets were actually more likely to enter new ones.) But Christensen has in the years since suggested that the theory is a sort of universal explanation for technological shifts. Business leaders accept the theory as a given.
But Goldfarb says the theory has never been tested in anything like a rigorous way. "If you give Christensen all the benefit of the doubt, and say that his original study was factually correct — even if it was right, it doesn't make what he found a general fact about the world," Goldfarb says. "It makes it, at most, a fact about one industry. The next step would have been to replicate that study across many industries, with a reasonable sampling strategy that would tell us more about when it applies and when it doesn't. And under what conditions. And that's never been done. "
"Generalizing to the universe of industries based on the study of a single industry is simply wrong," Goldfarb said at the Academy of Management conference. A valid way to explore the theory's universality, he added, "would be to take several industries based upon certain criteria — say, high tech, low tech, manufacturing and services — and then explore the predictive validity of the theory of disruption across these industries. Instead, Christensen has assembled a series of case studies sampled on, as far as I can tell, incumbents losing to new entrants. When a case appears that does not comport, the strategy is to calibrate the theory to the data. This is a great strategy for lengthening the longevity of a theory but does not subject the theory to Popperian refutability." He goes further: It's a "fundamental violation of the scientific methodology." (The Chronicle of Higher Education explores rebuttals of disruption theory this week, quoting Goldfarb.)
The new Sloan Management Review paper, "How Useful Is the Theory of Disruptive Innovation?" by King and Baljir Baatartogtokh, a graduate student at the University of British Columbia, finds that only 9 percent of the 77 case studies that Christensen and his co-authors have presented as examples of disruption actually show all four of the features that supposedly define a disruptive event. (Those four qualities were: Incumbents were successful at sustaining innovation in their own target market; that innovation "overshot" the needs of a substantial number of customers; the incumbents had the capability of responding to the challenge from a newcomer; and the incumbents floundered afterward.) Fully 39 percent of the time, for instance, incumbents survived or even thrived after being disrupted, King and Baatartogtokh found. The authors' methodology was to interview experts in the field from which the case studies were drawn.
One notable and problematic assumption of the theory, according to Goldfarb, is that incumbent entities have the ability to respond to every threat. When email, backed by AOL, arose as a threat to traditional mail service, could the Post Office have plausibly reinvented itself as an email provider, as Christensen has implied? It seems doubtful. And Goldfarb points out that, contrary to one favorite disruption story, Fuji Film Holding Corp. did not outflank Eastman Kodak because it foresaw the disruptive impact of digital cameras on film. Rather, it recognized that digital cameras were on the road to becoming a commodity, and so it moved into other areas, including coatings, cosmetics and document processing.
When incumbents can compete with insurgents, they usually do. Tesla is often said to be disrupting the car industry, but Ford, Nissan, GM, Porsche and a host of other companies are all racing to win the electric-car race. "It's a really competitive space," Goldfarb says.
In recent years, a debate has raged in many areas of the sciences and social sciences about whether, in the quest for striking findings, researchers are pushing data too hard. In a forthcoming paper, Goldfarb and King suggested that 26 to 40 percent of the results in strategic-management journals would not reappear if the experiments were replicated. This is almost never conscious fraud, but rather a too-human desire to see patterns in complex data, where none exist (plus a smidgen of self-interest).
Cognitive biases have played a part in the widespread acceptance of disruption theory, too, Goldfarb says. "It's a general problem that once something is published, it gets an extra credence that it should not have," he says. "It becomes a fact."
Then there's the sociological problem that people who know the theory best become invested in defending it, rather than challenging it. In some quarters, defending disruption can "seem to be more of a religious stance than a thoughtful one," Goldfarb says. Meanwhile, people outside the subfield, who may be skeptical, "don't have time to investigate the evidence."
"I wouldn't at all state that industries don't get disrupted," Goldfarb says. "They do. It's just the mechanism does not work in the way Christensen has described."
Disruption theory serves as a "thinking shortcut" that obscures, rather than clarifies, why some firms thrive during periods of change, and others die, he says.