Harvard Business School

Browse By:

The noncompete clause is the basis for a love-hate relationship, depending on which side of the desk you're seated.

The noncompete is meticulous legal phrasing that forbids inventors or other professionals from offering their talents to an organization's competitors before a set period of time has passed, often twelve to twenty-four months. Such a clause, for instance, sparked a public legal battle two years ago when Google hired away a top Microsoft executive, Kai-Fu Lee.

But disputes and tensions over noncompetes are not isolated instances. As most managers know, tensions like this affect careers and businesses all the time.

The power of the noncompete clause has led to a unique Harvard Business School paper with implications for day-to-day behavior, careers, business, and policy. Authored by Matt Marx, a doctoral student, Deborah Strumsky, a research associate, and Lee Fleming, an associate professor of technology and operations management, their paper—available for download—is titled "Noncompetes and Inventor Mobility: Specialists, Stars, and the Michigan Experiment."

What they found: While noncompetes do constrain mobility in a demonstrable way, more important was that these effects were amplified for inventors with particular characteristics. As Marx explains, the results suggest that "star" or "specialist" inventors wishing to explore career opportunities may need to look outside a state that enforces noncompetes.

The findings are important for regional development as well: "Given the many and mostly unsuccessful attempts to recreate Silicon Valley in various areas, understanding whether noncompetes play a role could be of help to policymakers," he adds.

Before entering the doctoral program at Harvard Business School, Marx—who holds six patents—spent ten years as an inventor and executive in the field of speech recognition. In this e-mail interview, he discusses the impact of noncompetes for inventors and those who would tie their hands.

Martha Lagace: I am guessing that your professional experience showed you a lot about the world of noncompetes before you embarked on this study. How did you get interested in studying them vis-à-vis inventor mobility? What gap in the research did you want to investigate?

Matt Marx: You're right; this topic drew directly on my experience. After working at a speech recognition company in Boston, I was recruited by another speech recognition firm in Silicon Valley. At first I thought I wouldn't be able to take the job given the noncompete agreement I had signed in Boston, but I was nonetheless able to make the move because the California courts generally refuse to enforce such agreements. Ironically, when I returned to Boston a few years later for business school, the founder of the first company invited me to do some consulting on the side while a student, but I wasn't able to because the Massachusetts courts would enforce the noncompete that the California company made me sign.

This experience got me thinking that my employment opportunities had been geographically circumscribed by differences in the enforcement of noncompetes between states—that with my highly specialized skills, the only way I could change jobs would be to move to one of the ten states that doesn't enforce noncompetes. (It's a bit of a continuum, where some states are stricter than others, but the ten states considered significantly more lenient are California, Nevada, Arkansas, Washington, Montana, North Dakota, Minnesota, Wisconsin, Connecticut, West Virginia, and Oklahoma.)

But not everyone is affected identically by noncompetes. Our research shows that those with specialized skills (such as speech recognition engineering) are impacted more than those with more generally applicable skills. For instance, C++ programmers can probably find work more easily at companies that don't compete with their current employers. We also found that "star" inventors—those whose patents are highly cited in other patent applications—are also more strongly affected by noncompetes.

Q: What did you observe about the pluses and minuses of noncompetes from the perspective of employer and employee-specialist/star?

A: From the employer's perspective, one thing to keep in mind is that noncompetes are far from ironclad. Even if one operates in a state that enforces noncompetes, employees may explore their options at competitors in states that do not enforce, especially during liquidity events such as initial public offerings and mergers. For example, when speech-recognition supplier ScanSoft acquired Nuance in October 2005, twelve of Nuance's thirteen research and development engineers left for Yahoo. The Santa Clara County judge refused ScanSoft's request for a temporary injunction based on the Nuance engineers' noncompetes.

The flip side of the coin is that employers can exploit these differences by locating in areas such as Silicon Valley where noncompetes do not block one's ability to recruit talent. For example, one legal scholar claimed that Novartis opened its "intellectual property factory" in La Jolla, California in order to take advantage of the free(r) flow of talent. I'm surprised that more Massachusetts companies haven't opened offices just over the border in Connecticut, which is more lenient on noncompete enforcement.

Q: Until I read your paper I just assumed that noncompetes were an invention of modern—especially high tech—companies. But as you make clear, noncompetes have enjoyed a long history. How and why did they arise?

A: The earliest recorded noncompete case was brought in England in 1414, when a dyer of clothes tried to enjoin a former assistant of his from setting up shop in the same town. But judges in the fifteenth century did not look favorably upon noncompetes, as the Bubonic plague had largely decimated the European labor supply. Legislation of the time essentially placed labor in the public domain, which was only strengthened by the rise of craft guilds in the sixteenth century. Legal scholars claim it was not until the Industrial Revolution that courts began to routinely enforce restrictions on employee mobility, though they generally held that such had to be "reasonable" with respect to duration and geographic reach.

Q: How widespread are noncompetes today? And what are the consequences for those who sign them?

A: These days, noncompetes seem nearly universal, and not just in technology companies. I recently spoke with an eye surgeon who will essentially be unemployed for the next eighteen months after resigning from a practice over a dispute with the owner. While his noncompete, like many in the medical and dental fields, was confined to a radius around the practice (in his case, twenty miles), he realized that there were no major hospitals outside that radius but within driving distance where he could perform surgery. The consequences can be equally severe in technology companies: for instance, in 2001 Doron Kempel was forced to resign as the CEO of SANgate Systems after just three months when a judge determined that he had violated the noncompete he had signed at EMC.

My own experience got me thinking that my employment opportunities had been geographically circumscribed by differences in the enforcement of noncompetes between states.

While no one has conducted an exhaustive survey, several scholars' results suggest that noncompetes are ubiquitous or nearly so, particularly in venture-funded companies where investors seek to protect their intellectual property. It might seem that Non-Disclosure Agreements (NDAs) would guard against the loss of trade secrets and other IP, but such agreements are notoriously difficult to police. Moreover, employers have alleged that even those with the most noble intent of upholding their NDAs will "inevitably disclose" what they know in the course of everyday work.

As I noted earlier, states differ in their enforcement of noncompetes. California has quite vehemently refused to enforce noncompetes, even denying Merrill Lynch the right to specify that its noncompetes signed in California would be governed by the laws of another state (with more stringent enforcement). That said, even the anticipation of a drawn-out court case can have a chilling effect on mobility. Although the California courts would likely have rejected Microsoft's claims, Microsoft and Google settled out of court regarding Google's hiring Microsoft executive Kai-Fu Lee to head Google China. And the following year, Microsoft evangelist Vic Gundotra decided to take a year off after accepting a job offer from Google in order to wait out his noncompete and avoid similar entanglements.

Q: How did you frame and carry out this study, and why in Michigan?

A: Other studies had looked at rates of firm foundings, but we wanted to track the mobility of individual scientists, engineers, and other creative professionals. We used the U.S. patent database, which gives the hometown for each inventor on each patent. We developed methods to determine whether the two patents by "Lee Fleming" and "Lee O. Fleming" are from the same person (they are) and whether the patents by Matthew Marx in Boston are the same as those by Matthew Marx in Connecticut (they're not). Thus we were able to observe inventors' mobility over time, at least to the extent that they filed patents.

We were initially interested in whether the differences in enforcement between states might help explain the agglomeration of talent in regions like Silicon Valley. Our first attempt was to compare rates of interstate movement—Boston to California and vice versa. In comparing across states, however, we were unable to rule out confounding factors—how did we know that people weren't just moving to California for the weather? What we wanted was to compare mobility with and without noncompetes, controlling for other factors in what economists call a "natural experiment."

States differ in their enforcement of noncompetes. … That said, even the anticipation of a drawn-out court case can have a chilling effect on mobility.

Fortunately, we found just that in Michigan. At the turn of the twentieth century, Detroit in many ways resembled the Silicon Valley of the last few decades. Growth of the nascent auto industry was explosive, with more than 500 firms entering before 1915. Just ten years before, the Michigan Legislature had passed a statute quite similar to that in California, barring the enforcement of noncompete agreements. This law governed all such disputes until March of 1985, when it was repealed as part of the Michigan Antitrust Reform Act.

Now, comparing conditions before and after a change in the law is a common method of establishing a natural experiment, but you have to be careful because sometimes the legislature is just "catching up" with common practice, rendering the comparison useless. For example, if the Michigan courts had started enforcing noncompetes in the late 1970s, and the legislature eventually reflected that in 1985, you would be hard pressed to say that the change in law was a completely external factor.

But the Michigan Antitrust Reform Act was about much more than noncompetes. When we read the legislative analysis, we didn't see noncompetes mentioned even once. We began to wonder whether the legislature might have repealed the noncompete prohibition inadvertently, which would make for a "clean break" well suited to this sort of study. In speaking with two authors of Michigan Bar Journal articles on noncompetes just after the change, we found evidence suggesting this in fact was the case. And although the legislature did amend the Reform Act two years later, retroactive to the initial act, it did not re-adopt the earlier prohibition on noncompetes but merely adopted the "reasonableness" test common to states that do enforce.

Of course, there are many factors that could have changed within Michigan before and after 1985, particularly given the turbulence in the auto industry. So the exact comparison involved the mobility of Michigan inventors compared to that of inventors in other non-enforcing states, observed both before and after the change in law.

Q: What were your main findings? What do the findings mean for individuals as well as companies and regions?

A: We found a sharp drop in the ratio of mobility of Michigan inventors to those in other non-enforcing states after noncompetes began to be enforced, a drop of about one-third. To be honest, that wasn't a huge surprise; noncompetes are intended to constrain mobility and appear to do just that. More interesting to us was that these effects were amplified for inventors with certain characteristics. In particular, inventors whose patents are highly cited in other patent applications were even less likely to change jobs following the change in law.

The effect for "specialist" inventors was even stronger. We determined the degree to which an inventor is a "specialist" by the number of technical fields in which they innovate, as recorded in the technical classes assigned to their patents by the U.S. Patent Office. Inventors whose patents are all in one class, or in just a couple of classes, are said to be more specialized. These results confirmed our initial hunch based on my own experience; more importantly, they suggest that "star" or "specialist" inventors wishing to explore career opportunities may need to look outside a state that enforces noncompetes. Hence, those with more specialized skills may want to think twice before signing an extensive noncompete and may even want to consider jobs in non-enforcing states lest they have no choice but to "go down with the ship" if their company does not perform well.

These results also beg the question of whether regions that enforce noncompetes may contribute to a "brain drain," particularly of highly-regarded or specialized technical talent. This certainly appears to be on the minds of members of the Greater Boston Chamber of Commerce, who in 2003 issued a report titled Preventing a Brain Drain: Talent Retention in Greater Boston. While it would be a stretch for us to claim that noncompete enforcement creates a regional brain drain, it is nonetheless likely that Silicon Valley headhunters recruiting talent from Boston worry less about noncompetes than vice versa.

Did California's prohibition on noncompetes help spur the growth of Silicon Valley? Did Michigan's similar policy in the early 1900s have a similar effect on the Detroit auto industry? Again, there are too many factors involved for us to make definitive conclusions. Nonetheless, given the many and mostly unsuccessful attempts to recreate Silicon Valley in various areas, understanding whether noncompetes play a role could be of help to policymakers.

Q: How would you like to extend this research? And what's next for you?

A: Fundamentally, noncompetes are a form of monopoly. Just as a patent permits a monopoly on a technique or tool for a limited amount of time, a noncompete (if enforced) affords a temporary monopoly of sorts on a person. The advantages of monopolies to private firms are well-documented and clearly sought after in practice. The implications of such monopolies are less clear, however, for non-corporate entities such as individuals and regions.

One of our next projects will look at whether employees behave differently given noncompetes. Will they take fewer risks for fear of making mistakes and thus losing their job? Also of interest to us is whether companies adjust their strategies based on noncompete enforcement. We might expect, for example, that unsanctioned spin-offs will distance themselves further from their parent firms where noncompetes are likely to be upheld. If so, we might see increased technological "crowding" in areas like Silicon Valley compared to areas like Boston.

Matt Marx can be reached at: mmarx@hbs.edu