Work from home or go to an office: Which kind of workers are happier and more efficient?

For the first time–surprisingly–there’s some solid academic research on the subject, led by a Stanford economics professor.

We won’t hide the ball. Yes, the study found that workers who were allowed to work from home reported higher satisfaction, and they did their jobs more efficiently. But there’s also a big asterisk on that conclusion–one that might make you hesitate before going to a 100 percent work-from-home model.

Below: the study, the results, the caveat–and what it means for you and your work.

First, get 500 volunteer workers

First, the research, which was reported recently in a working paper published by the National Bureau of Economic Research, and led by Stanford economics professor Nicholas Bloom and his graduate student James Liang (more on Liang in a minute).

Their paper was born from a big hole Bloom had found in previous research.

“Everything I saw was ‘pro working from home,’ and put out by people who were for it from the outset. The people against it have stayed quiet,” Bloom told the TED Ideas blog recently.

(Bloom also has a TEDx Stanford talk called “Go Ahead, Tell Your Boss You Are Working From Home.” It’s embedded at the end of this article.)

So Bloom set out to conduct his own, rigorous study. Challenge number one was how hard it was to find a large number of research subjects working in similar jobs, who could divide into work-at-home and work-at-the-office cohorts, and who were willing to cooperate with the researchers over a long enough period of time.

But the solution was sitting right in front of him: Bloom’s graduate student Liang, who was also CEO and co-founder of China’s biggest travel agency. He recruited 500 volunteers from his company’s call center in Shanghai for the study, half of whom were accepted.

Work at home wins*

Here’s how the research worked. Accepted volunteers had to have been with Liang’s company, CTrip, for more than six months, and they had to have a private room at home to work from, with decent broadband internet. Then, the volunteers were randomly divided into two groups–the ones who would work from home four days a week, and the ones who would remain behind, working in the company’s offices.

The experiment would run for nine months–and their managers were concerned. Would their employees be able to avoid “the three main pitfalls of being at home: the bed, the TV, and the fridge?” as the TED Ideas blog’s Ari Surdoval put it?

In a word: Yes. Bloom, Liang, and their two other co-authors reported the following results in their working paper “Does Working From Home Work? Evidence From a Chinese Experiment”:

  • CTrip saved about “$2,000 per year per employee” on office space.
  • The employees who worked at home were 13.5 percent more efficient than those who worked in the office.
  • Work-at-homers also were 9 percent more engaged  than their in-office co-workers (measured by the percentage of time each hour they were actually logged into the company’s call-taking system and doing their jobs).
  • Work-at-home employees “reported shorter breaks and fewer sick days and took less time off.”
  • Attrition rates were 50 percent better with the work-at-home employees.
  • And while it’s hard to measure happiness quantifiably, the employees who worked at home “reported higher job satisfaction.”

Sounds great, but there’s just one catch …

So far, it sounds like a big win for the work-from-home crowd, and in fact CTrip was so convinced that it decided to offer almost all its employees the option to work from home.

But when it did so, something surprising happened: Half of the study participants who had been allowed to work from home decided to return to the office.

They cited two reasons, by and large. First was the isolation that the work-at-homers felt. Second was their (accurate) perception that employees who worked from home were less likely to be promoted and receive bonuses.

Also, many of the employees who volunteered for the study were among the youngest in the company. Many reported that they lived with their parents–and that the experiment made them realize they counted on work as a daily escape from their families!

So with that human factor in mind, here are Bloom’s recommendations for companies and workers who want to consider work-from-home policies.

1. Experiment with making it condition-dependent.

Perhaps start by offering it “on a contingency basis when severe weather events are forecast, or for summer days when people’s children are out of school,” for example, he suggests in the TED Ideas blog article.

2. Treat it as a bonus or reward.

“it could be given on an individual, probationary basis; it could be part of a promotion; or it could be granted in lieu of a raise or a bonus. And if productivity falls, an employee can return to being in the office full time.”

3. Don’t jump straight to 100 percent.

Maybe one or two days per week working from home is the best solution, Bloom theorizes. “You don’t want to go much higher, because you risk jeopardizing the cohesion of your team.”

4. But give it a try–seriously.

“The need to go into a workplace five days a week started because people had to go to a factory and make products. But companies that still treat employees like that are increasingly finding themselves at a disadvantage,” he says.

Keep in mind: Bloom and his team put their research into action themselves. Their report ends by noting that “much of the research for this paper and its writing were done by the authors working from home.”