Where do you imagine I might be writing these words? Am I at the FT? Or crunched up on a train? Perhaps I’m at the kitchen table or by a swimming pool. I may be wearing a suit, swimming trunks or pyjamas. Once you start to imagine some of the possibilities, you may find them difficult to banish from your mind.
For most occupations – from chef, to surgeon, to security guard – telecommuting is impossible. But an increasing number of jobs are becoming feasible to do from home. Perhaps it’s not surprising, then, that telecommuting is catching on. In the US, 10 per cent of people sometimes work from home, and more than 4 per cent of people do so predominantly. The curious thing is, we know very little about whether home working is effective – and within specific industries, different companies sometimes have very different practices. Swathes of management thinking are still akin to medieval medicine: there are lots of ideas floating around, but nobody really knows what’s effective and what’s superstitious nonsense.
My eye was caught, then, by a new study conducted by Nicholas Bloom, James Liang, John Roberts and Zhichun Jenny Ying, all economists at Stanford University. Liang also happens to be the founder and chairman of Ctrip, a Chinese travel agency, listed on the Nasdaq and currently valued at about $2bn.
This confluence of interests created a remarkable opportunity. Ctrip employs a lot of call-centre staff, and call-centre jobs – just like columnist jobs – turn out to be perfectly amenable in theory to working from home. Staff can be monitored through digital means, and all they need is a computer, a phone and a quiet space. But Liang simply didn’t know whether introducing telecommuting would work well. So, with Bloom and colleagues monitoring the results, Ctrip introduced a carefully designed randomised trial. From a pool of 255 qualified volunteers, Ctrip used birthdays – odd or even – to assign about half to nine months of home working and half to continuing to work from the office.
The telecommuting was a big success. Home workers on the same shifts as their office-bound colleagues were online and available for more minutes because they took fewer breaks and fewer sick days. They also took slightly more calls per hour logged in because, in the quiet home environment, they could hear customers more clearly. Each call was just as productive. Meanwhile, the home workers reported more job satisfaction and a better mood in general, and they were less likely to quit their jobs.
Totting up the results, Ctrip reckoned that each home worker, on average, produced better performance worth $375, saved $1,250 in office rental costs, and also saved $400 because of reduced turnover and training. That saving of more than $2,000 looks particularly impressive given that the average salary was $3,000 over the course of the same nine months. Telecommuting works, at least for Ctrip – and quite possibly for call centres all over the world.
It’s worth pausing, too, to contemplate the importance of running a proper randomised controlled trial. After the experiment was finished, some people who had not enjoyed working from home came back to the office; while eager would-be home workers replaced them. Because it was the less productive workers who tended to give up home working, this self-selection promptly made home working look twice as productive. We should conclude that home working is likely to be more effective when self-selection is allowed – but also that evaluations based on self-selection instead of randomisation are likely to deceive us.
Now, if you’ll excuse me, it’s time to get a coffee from the staff canteen. Or do I mean a cold beer from the poolside bar?
First published in the FT Magazine.