When is a company not a company? A modern-day Zen koan, maybe, but the evidence is all around us—Airbnb, Elance-oDesk, Handy, HourlyNerd, TaskRabbit, Uber. These businesses seem much more like conglomerations of independent professionals that connect to customers through a common platform, rather than traditional businesses that employ their workers.
And so it seemed until June, when the California Labor Commission disrupted the sharing economy when it declared that an Uber driver was an employee, not an independent contractor. Far from being a "neutral technological platform," the ruling argued, the company was "involved in every aspect of the operation"-from how many rides drivers had to accept in an hour to how they had to interact with customers. Two separate class-action lawsuits against Uber and Lyft are still pending, and the lawyer responsible has brought new lawsuits against similar on-demand companies Shyp, Washio, and Caviar.
"I think it would be crazy to all of a sudden say that all Uber drivers are employees, because we would be effectively killing the business model"
What to make of this brave new world of organized chaos? On the one hand, companies have a clear interest in enforcing consistency in order to build their brands. On the other, their business models are predicated on the ability to empower professionals to deliver services independently.
The key word to understanding this brave new world is "control," says Andrei Hagiu, an associate professor in the Strategy unit at Harvard Business School, who wades into the controversy with a new working paper, Enabling Versus Controlling, co-written with National University of Singapore economics professor Julian Wright.
Uber may have very good business reasons for controlling certain parameters of the service—price, for example—but being flexible in other areas, such as letting drivers choose the car they drive, says Hagiu. "This places [Uber] somewhere in between pure platforms and pure employers. While everyone agrees that the key defining characteristic is the extent of that control, there is no clear sense from the courts of exactly where the dividing line between platforms and employers might be."
These issues are hardly new. Businesses ranging from hair salons to real estate agencies have long organized themselves as associations of individuals, with the company itself exerting more or less control over the interaction between service provider and customer. Some hair salons employ their stylists, provide them with product lines and equipment, and conduct advertising. Others salons simply rent out chairs to independent stylists and allow them to cut hair and cultivate new clients as they see fit.
"Companies have played with this degree of freedom for a long time," says Hagiu. Nevertheless, platform models have become much more prevalent recently as "technology has made it easier to build marketplaces for services and fine-tune the degree of control exerted over service provider-customer interactions." HourlyNerd allows consultants (typically MBA students) to hire themselves out individually to companies on an hourly basis, promoting themselves with custom-based profiles and setting their own fees, in direct contrast to the traditional consultancy model of companies like McKinsey & Co. that trade on the power of their corporate brand.
A key question for service marketplaces today is when they should choose to operate as a traditional business (employing and controlling professionals), and when they should choose to operate as a platform (enabling professionals to interact with customers on terms they can choose). In their paper, Hagiu and Wright use an economic model to explore this question.
The model features two types of decisions that affect the joint payoffs of the firm and the professionals: nontransferable and transferable. Nontransferable decisions are always completely controlled by the professionals (e.g., how friendly to be to customers) or by the firm (e.g., the quality of a salon's interior or of the ride-hailing app). In contrast, transferable decisions can be made by either party—the type of car an Uber driver uses; the kinds of details an Airbnb host lists about an apartment offered for rent; or how a hair stylist at a salon will advertise her services.
If the transferable decisions are controlled by the firm, then the latter is functioning as a traditional business; if the transferable decisions are controlled by the professionals, then the firm is functioning as a platform.
Follow the Incentives
According to Hagiu and Wright's model, choosing who controls transferable decisions is all about incentives. Whichever party's nontransferable incentives are more important should be given control over transferable decisions. "If the way employees behave is super important to revenues, then you should move to a platform model," says Hagiu. "That means you should also give them a higher percentage of the revenue in order to motivate them."
Things get more interesting with the degree that the actions of one professional affect the revenues generated by other professionals—what is known in economic terms as spillovers. If professionals are competing with one another and are negatively affecting each other's revenues—drivers poaching each other's fares, or stylists advertising against one another—then that would seem to tilt the balance in favor of the traditional business mode. "If the spillovers are negative, and the professionals only care about themselves and don't coordinate, then there is an inefficiency in functioning as a platform," says Hagiu. In that case, it's incumbent on the firm to "internalize the spillovers" by coordinating its employees' actions.
A certain amount of competition among employees, however, can be good for a firm even in a platform model by incentivizing employees to put forth greater effort even without paying them a high commission. "Individual hairdressers may be advertising too much because they compete against one another," says Hagiu, "but that has the advantage of also incentivizing them to exert more effort without requiring a higher bonus." By contrast, in the traditional mode, where the salon employs the stylists, the competitive pressure is absent, so the salon would have to pay a higher bonus in order to motivate them to the same extent. Thus, this mechanism can actually push the optimal arrangement back toward the platform mode.
So what does this have to do with the degree to which Uber and other companies control their employees? The reality is most of these companies fall into a poorly defined gray area, says Hagiu, where creating the optimal incentives to maximize revenues for contractors and companies alike is a delicate balancing act of incentives and control. Yet, the law as currently written creates a highly inefficient either/or situation.
"The way the law is today, it forces you to make a binary decision: You are either a platform enabling independent contractors or an employer that has to pay full employment costs," says Hagiu. "It is very inadequate for these marketplaces." Oftentimes, the optimal model for a company might be somewhere in the middle—controlling some aspects of contractor performance but not others, a situation that current regulations do not yet account for.
"I think it would be crazy to all of a sudden say that all Uber drivers are employees, because we would be effectively killing the business model," says Hagiu. After all, Uber only takes a 20 percent commission from drivers—so the company is likely thinking of them mostly as independent contractors, and giving them the lion's share of revenues in order to incentivize them. "On the other hand, I also don't think it's right to say that Uber is a pure marketplace and that it doesn't have any obligations. The answer is obviously in the middle."
New Category of Worker Needed?
Some legal scholars have recently proposed an intermediate category of workers called "dependent contractors" with some legal protections and benefits without the full perks of employees, similar to classifications in other countries such as Canada and Germany. Hagiu sees this as at least a step in the right direction, short of providing a continuum of degrees of control but at least offering a pit stop on the way between full autonomy and full control.
"There are many degrees of control, and regulations should assign them different levels of cost," says Hagiu.
The real debate, he argues, shouldn't be over whether workers in the shared economy are employees or independent contractors, but about how to devise new forms that would accurately reflect their unique in-between status.