Does the Hybrid Work-From-Home Model Work?

Summary

• A new NBER working paper by Stanford University economists and Trip.com analyses a randomised control experiment of the hybrid work-from-home model.
• The NASDAQ-listed company of over 35,000 employees found a hybrid working arrangement significantly reduces quit rates and improves job satisfaction.
• Importantly, there is no discernible detrimental impact on productivity. If anything, worker output increases despite sitting at the desk roughly 2.5 hours less when working from home.

Introduction

The pandemic caused a monumental transition to work from home (WFH) as offices and shops closed during lockdowns. Back in November 2020 and May 2021, we reviewed the transition. The consensus then was it went better than expected and that longer-term adoption was probable. A year on, hybrid working-from-home (HWFH) models have become popular. But has home working hindered productivity?

No, according to two Stanford University professors. In their new NBER working paper, they evaluate the performance of 1,612 engineers, marketing and finance employees at a large technology firm, Trip.com, that has adopted HWFH. The impact has been largely positive, with productivity, retention and job satisfaction all increasing relative to full-time office workers. The study’s outcome was so well received that all 35,000 employees at Trip.com now work under the HWFH model.

Hybrid Working From Home

By the end of 2022, around 30% of workdays will be WFH, more than double pre-pandemic levels but half the April 2020 peak (Chart 1). HWFH is becoming common, offering workers two to three days at home and the remainder in the office.

HWFH models have four main benefits. Firstly, workers want to WFH. Around 70% of employees that can WFH say they want to work at least one day a week remotely. The main reason is the reduction in commuting time, which saves the average worker 70 minutes a day.

Secondly, individual activities like coding or writing are usually easier in a quieter space. Indeed, three-in-five employees believe they are more productive working from home. Thirdly, and lastly from an employee perspective, home working gives workers greater time flexibility, helping them balance childcare constraints.

Combined, employees see WFH as equivalent to roughly a 5% pay rise – a large cost saving to businesses. This is the fourth main benefit of HWFH models: cost savings for businesses, either indirectly from salary equivalents or directly through reducing space costs. This could be one reason why employers plan to allow between 1.6 to 2.4 days per week to be worked from home.

The main drawdowns from home working are increased home-specific distractions and a lack of access to the right technology. However, according to the survey we covered in May 2021, the average employee has already invested 1.2% of their annual salary on improving home working conditions. And, with children back at school, the drawdowns may be more limited than some critiques expect. This is what the paper tests.

The Experiment

In the summer of 2021, Trip.com, a NASDAQ-listed global travel agent with 35,000 employees headquartered in Shanghai, decided to run a trimmed-down HWFH system to evaluate its long-run potential for the firm.

For it, they took 1,612 engineers, marketing and finance employees in the Airfare and IT divisions, randomly assigning them into two groups. The first would WFH on Wednesday’s and Friday’s (HWFH group) while the other would continue to come into the office full time (control group).

To be assigned to the HWFH group, your birthday would need to be on an odd day of the month. Those with even-day birthdays would be allocated into the control group. This way, the firm created a perfectly randomised control experiment, which they ran for six months.

The Volunteers

Of the 1,612 employees, 70% were in technical jobs that required writing software code for the website, and front-end or back-end operating systems. The remainder worked in business development, market planning and in business services. Across the groups, employees spanned all seven pay grades, from graduates to employees.

Only 518 participants were volunteers. The lack of volunteers they believe was over concerns that volunteering would be seen as a negative signal of ambition and productivity. So, one month after the volunteers began their trial, a further 1,094 non-volunteers were recruited.

In terms of characteristics, most employees were in their mid-30s. Two-thirds were male, and one half were married with children. All had a university degree. Volunteers tended to be younger and less experienced, with longer commutes. Employees with managerial roles were less enticed by the experiment.

The Results

Four results stood out from the experiment:

1. Those randomly assigned to the HWFH group were less likely to leave their jobs and reported higher work satisfaction scores.
2. The same group reported working shorter days at home, but longer days in the office compared with the control group. HWFH employees were also more prepared to work at the weekend.
3. Remote worker employees became better than their peers at communicating, even when in the office.
4. The productivity of HWFH employees increased relative to the control group, especially IT professionals.

Attrition Rates

Consistent with earlier survey results that show individuals want to WFH, those Trip.com employees who could were much less likely to leave their job (Chart 2). During the six-month experiment, just 4.7% of HWFH workers left, compared with 7.2% in the control group. This was also noticeably higher for those who volunteered to be part of the experiment, i.e., ones who really did want to WFH.

To contextualise these findings, the firm also ran an anonymous survey. It showed that HWFH employees were more likely to recommend the firm to friends, and have higher levels of work satisfaction, life satisfaction, and work-life balance than control employees.

Work Patterns

The company monitored work hours through an employee’s VPN. The data showed that the treatment group worked fewer hours on Wednesday’s and Friday’s – those days they worked from home – by roughly 2.5 hours a day. However, the same employees increased working hours on office days and outside of core working times, such as weekends, by 6.6% relative to the control group (Chart 3).

On balance, work-from-home staff worked around an hour less a week, but for a similar level of performance (measured by code written by IT staff, and promotions for non-IT staff). Over a full year, this equates to two days off. However, working from home reduced the number of sick days taken, because staff were more able to continue working with a mild illness. So overall, labour supply increased with the ability to work remotely.

Communication

Perhaps unsurprisingly, the treatment employees saw a 14.1% increase in their use of the company’s internal messaging chat (TripPal) relative to the control group. Most of the increase was on WFH days, but there was also a 10% increase on office days. According to the authors, the main driver is treatment employees starting to increase their overall level of messaging even in the office.

The increase in internal messaging communication was mirrored for group video calls too. The firm saw the overall increase in communication, even when in the office, as positive. Greater communication leads to more ideas, improving productivity. Graduates may also feel more comfortable asking questions via messaging than face to face.

Employee Performance and Productivity

The firm uses two measures of performance – a six-month performance review and promotion rates. However, for 729 employees whose primary job is coding, the authors also collect the number of lines of code they have written to measure productivity.

For the first two measures, the authors find no discernible performance difference between the treatment group and the control group. However, in the firm-run survey, WFH employees self-reported 1.8% higher productivity than the colleagues who worked in the office full time.

On the coding front, the authors found the lines of code written by treatment employees rose 8% over the six months, compared with the control group. This is despite working fewer hours.

So, there appears to be no negative productivity impact from remote working. And some of the evidence suggests productivity even improved under a hybrid remote work environment. As a result, in February 2022, the firm rolled out HWFH for all employees (although this was paused during the March Covid lockdowns in Shanghai).

Bottom Line

Clearly, the ability to WFH depends on the role. However, growing evidence supports an HWFH model for employees that can. It reduces costs, improves communication, provides flexibility and improves work-life balance. All this comes with limited, if any, cost to productivity.

Given the significant rise in the cost of living, as well as wage inflation and slowing economic activity, the HWFH model may be here to stay for at least a while longer!