Working from home increases productivity, but is much more nuanced than you might think
Inspired by a flurry of interesting new research on working from home (see below), I couldn’t resist looking at the data myself. In the UK, the Understanding Society survey* has been asking people about their experiences throughout the COVID-19 pandemic, and I was curious to find out what this might tell us about working from home. [* University of Essex, Institute for Social and Economic Research. (2021). Understanding Society: COVID-19 Study, 2020–2021. 11th Edition. UK Data Service. SN: 8644, DOI: 10.5255/UKDA-SN-8644–11]
I had two strong in-going hypotheses when I started exploring the data. First, I had found some of the positive findings about working-from-home productivity compelling. These included research showing that — for the 30% or so of people whose job can reasonably be performed remotely — working from home had many benefits including increased productivity. Second, I was convinced that the effects are highly heterogeneous — in other words, very different for different people, jobs, teams, and organisations.
Personal productivity is, of course, hard to measure in most modern jobs. Using the Understanding Society data, we are relying on people’s self-reported sense of how much more or less they were able to do in an hour, compared to a January/February 2020 baseline. We are also relying on people’s self-reported statements on how much more or less they worked from home at the time of each survey, again relative to the baseline in early 2020. The exact question asked on productivity was formulated as follows:
Please think about how much work you get done per hour these days.
How does that compare to how much you would have got done per hour back in January/February 2020?
1. I get much more done; 2. I get a little more done; 3. I get about the same done; 4. I get a little less done; 5. I get much less done.
For my analysis, I focused on people who said they worked at home more frequently than during the baseline period, to see what they thought had happened to their productivity. The raw results certainly seem to support my hypothesis about heterogeneity: experiences were mixed. First off, around 50% of 1,900 people for whom we have data actually said the amount they got done in an hour was “about the same”. As for the remainder, around 22% said they got “much more” done in an hour, 19% said “a little more”, while 2% said their output was “a lot less” and 7% “a little less”.
What explains these different experiences? A lot of different things, ranging from a person’s age, sex, role in the household (e.g., responsible for home schooling), communting distance, occupation, job autonomy, and so on. [Note: I’ve taken this list of factors from the excellent “Home Sweet Home: Working from home and employee performance during the COVID-19 pandemic in the UK”, by Sumit Deole, Max Deter and Yue Huang].
Previous McKinsey analysis found that among women with children working from home, those with scheduling flexibility were three times more likely to have a positive view of their work effectiveness, compared to similar women but with no schedule flexibility. Intriguingly, I also found various correlations* in the UK data suggesting that simple things, like access to a desktop computer, were predictive of someone’s positive working-from-home productivity. [* I did not perform causal inference analysis.]
I had also suspected that even someone’s personality would have an impact on their productivity at home. The idea was initially sparked by a report by Global Workplace Analytics which suggested that “self-discipline” was a key enabling factor for successful remote work. Since the Understanding Society dataset also contains personality traits (from an earlier wave), I thought I’d have a look at whether any of them were associated with higher self-reported working-from-home productivity.
Unfortunately, because the personality traits are also correlated with each other, and with many other consequential factors (such as sex and occupational choice), it’s not easy to draw any strong conclusions. I did want to illustrate the influence personality can have, though, so chose to display the results for one of the “big 5” personality traits, extroversion. [One definition for extroversion says, “extroverts tend to be gregarious, assertive, warm, active, excitement-seeking, and positive”.]
The chart shows that people with a higher tendency for extroversion** who worked from home more frequently were much more likely to state that their personal productivity had improved. Among more extrovert people (right hand column), almost half said that they were “much more” or “a little more” productive, compared to the baseline period. In contrast, only around 30% of those with more introvert tendencies considered their productivity to have improved. [** The categories were based on a 1–7 scale, where those scoring 6 or 7 were considered “more extrovert”, those scoring 1–3 “less extrovert”, and those in the middle classed as “average extroversion”.]
It’s plausible that more extroverted people might have an inherently more optimistic take on their own productivity. But it also seems that extroverts found the work environment at home to be less distracting. For example, in the same data set, we can see that more extroverted individuals who had increased their working-from-home frequency were also slightly more likely to say that their concentration was “better than usual”. [This question was in the mental health part of the questionnaire and was phrased as “Have you recently been able to concentrate on whatever you’re doing?”]
The main point to take away is this: whether productivity at home is higher, lower, or the same really depends on a multitude of factors. Generalisations are, therefore, potentially dangerous and unhelpful. Tailoring remote working approaches to both business, team and individual needs is likely to be much more effective than blanket one-size-fits-all policies. High quality management — and managers—that pay attention to individual needs and circumstances have never been as important.