Evidence from a Chinese Experiment on Working from Home

09/01/2013
Featured in print Digest

Home workers increased the minutes they worked on each shift by 9.2 percent.

CTrip is China's largest travel agency, with 16,000 employees and a market capitalization of about $US 5 billion. Beginning in December 2010, it conducted a randomized experiment that measured the productivity of employees working at home versus those working in one of its call centers.

In Does Working From Home Work? Evidence from a Chinese Experiment (NBER Working Paper No. 18871), co-authors Nicholas Bloom, James Liang (who co-founded CTrip and helped to make its productivity data available), John Roberts, and Zhichun Jenny Yang present the results of that experiment. They find that the home worker group increased its productivity by 13 percent over the trial's nine month lifetime, with no measurable difference in the quality of calls.

Thanks to reduced breaks and sick days, home workers increased the minutes they worked on each shift by 9.2 percent. In a workforce with an average commute of 80 minutes per day, the employees attributed their reduction in sick days to the fact that on some days they felt well enough to work at home but not well enough to commute.

Their calls handled per minute also increased by 3.3 percent. When asked, the home workers attributed that improved performance to the quieter home environment. Work satisfaction also was higher for home workers, and attrition was half that of the control group working in the call centers. Interview results suggested that the home workers were less likely to quit because no other comparable home-working jobs existed.

CTrip estimates that the improvement in performance was worth about $375 per employee per year, that working from home reduced its office costs by $1,250 per employee per year, and that reducing attrition saved about $400 per employee per year in training costs and lost productivity. The only negative result from the experiment was that rates of promotion for home workers were almost 50 percent lower, possibly because they were not visible to office-based managers.

Of the 994 employees in the airfare and hotel booking departments at CTrip's Shanghai call center, 503 volunteered for the experiment. Of the volunteers, 249 met all conditions, such as working for the firm for at least six months, being willing to work from home for nine months, having home access to broadband internet, and access to an independent workspace at home during their shift. They were invited to apply for the experiment. Applicants with even-numbered birth dates worked from home as part of the 131-member treatment group. Applicants with odd-numbered birth dates continued working at a company call center as part of the 118-member control group. The home workers worked from home four days per week, returning to the call center for one shift on one day.

At the end of the experiment, CTrip expanded the work from home option to the entire firm. Two-thirds of the control group opted to remain in the office. Half of the home workers elected to return to the office, citing "concerns over the loneliness of home working and the lower rates of promotion." As a result of the self-selected exodus of people, a large fraction of whom had performed relatively badly at home, the performance gains by home workers nearly doubled after the experiment ended.

--Linda Gorman