Time to Embrace Hybrid Work Arrangements
Research keeps showing hybrid work is a superior form of work organization.
Thank you for reading our work! Nominal News is an email newsletter read by over 4,000 readers that focuses on the application of economic research on current issues. Subscribe for free to stay-up-to-date with Nominal News directly in your inbox:
Our updated goal for 2025 is to hit 10,000 subscribers:
If you would like to support us further with reaching our subscriber goal, please consider sharing and liking this article!
Work from home (WFH) has recently come under fire from JP Morgan CEO Jamie Dimon, as well as Elon Musk and Donald Trump with their recently announced return to office push for US government workers. These individuals believe that WFH is inherently a bad work model. Fortunately, we have better analysis than just beliefs about the impacts of WFH – let’s look at why the latest economic research disagrees with them.
State of WFH
Since the Covid pandemic, WFH has persisted in the US:
Based on surveys from WFH Research, about 25% to 30% of work days are currently being done from home.1 The trend appears to have stabilized. Clearly, WFH has remained a popular form of work organization. How does it affect work?
Productivity of Public Sector Workers
Police Work in the UK
Fenizia and Kirchmaier (2025) (“FK”) studied the impacts of working from home by public sector employees. Public sector employees are a good sample of workers to investigate since, unlike private sector roles, most public sector roles do not have performance based compensation in the forms of monetary bonuses. Moreover, public sector workers typically also have stronger job security. This means that the performance of these workers is less influenced by outside factors (bonuses, job security), unlike for private sector workers.
FK looked at workers at the Crime Recording and Resolution Unit (CRRU) in Manchester, UK. This unit is tasked with recording the details of cases stemming from incoming calls, making outgoing calls and reports, and making decisions about whether the case falls under the purview of the unit. Each worker in the unit does all three tasks on a rotating basis.
When working in one of the three work streams, the specific task within each work stream is assigned to the worker by an algorithm. Lastly, workers also get assigned which day they can work from home – the workers themselves do not get to choose this day.
The main variable of interest for FK is the number of cases recorded per day. FK, therefore, look at the number of cases workers record per day.
Work from Office vs Home
On average, each staff member records approximately 6 cases a day, spending around 93 minutes in the computer system recording all these cases. Note – the time may seem low but it only captures the time actively inputting into the computer system. Workers spend most of their time gathering information, as well as speaking on the phone with victims/informants.
Workers assigned to work from home end up recording about 9%-10% more cases compared to workers in the office. Moreover, workers do not actually spend more time recording cases – the total number of minutes spent on the computer system is the same. This means workers are actually more productive, as they do more cases per hour.
Beyond just comparing worker groups (work from office vs work from home), FK can actually investigate whether a particular worker is more productive at home or in the office, since each worker ends up working in both settings. It turns out that, on average, being assigned to work from home increases productivity by 12%, which is higher but similar to the group estimate. The chart below shows the productivity of an individual working from home vs working in the office.
An observation (red diamond), which is an individual, that is above the line is more productive at home than in the office. As can be seen, most workers are above the line.
Nominal News Comment
This study seems to be quite convincing due to the fact that almost every element of work, such as what work stream, which task, and in which location (home vs office) is assigned quasi-randomly2, like a natural experiment. Since workers cannot self-select into whether they work from home or the office, and cannot choose what they do at each location, the effect of location is isolated. Thus, the study appears to capture the ‘causal’ effect of location (home or office) on productivity.
It is worth noting that productivity here only captures how many cases are recorded in a system. The inputting into a system, which on its own can be deemed to be a manual/routine task, requires a lot of non-routine work first – following up on calls, understanding if the case falls under the purview of the unit, etc. One of the main benefits of working from home, as stated by the workers, was having fewer distractions and a more conducive atmosphere to this kind of work (for example, doing a call without background noise).
Readers of Nominal News may contrast this finding with research in our previous post on work from home. One of the referenced studies – Emanuel and Harrington (2023) – found that productivity, when working from home, dropped by 4%.3 In that study, Emanuel and Harrington looked at the performance of call center workers at one of the largest companies in the US and found that the main cause for the fall was an operational reason – it was tougher for workers to connect to their managers or receive assistance from others in dealing with challenging calls, resulting in more time spent per call.
In contrast, the work of the police unit does not require much interaction with others. It appears, therefore, that work that can be done independently might be better done when working from home than in an office setting.
Hybrid Working
Choudhury, Khanna, Makridis and Schirmann (2024) (“CKMS”) used a post-Covid policy in Bangladesh to study the impact of different amounts of hybrid work. The government policy restricted the number of people allowed in the office. CKMS looked at a company impacted by this policy over nine weeks. At that company, there were 130 workers, some of whom were allowed to work in the office, while others worked from home. A daily lottery decided who worked from the office or from home.
CKMS partitioned workers into three groups – high WFH (those that worked no more than 8 days/23% of days in the office), intermediate WFH (no more than 14 days/40% of days in the office) and low WFH (those who worked more than 14 days/40% of the days in the office). CKMS observed all the emails sent out by each worker and thus are able to measure the number of emails sent, to whom they are sent, as well as the ‘novelty’ of an email. How do they measure novelty?
Novelty was measured by using a neural network-based algorithm. The algorithm looked at all the emails in the company and would consider emails to be more ‘novel’ if the words and word combinations used in the email were rarer, suggesting that the email contained new information.
CKMS found that the intermediate WFH group did more work (as measured by email volume) and their work was also more ‘novel’ than either the low or high group. The increase in novelty is quite economically significant – with the intermediate group being approximately half a standard deviation more ‘novel’ than other workers.4 Additionally, CKMS looked at manager performance reviews, and although results are statistically insignificant, directionally, it appeared that the intermediate WFH group was the best performing cohort.
Work from Home – the Revolution Continues
There appears to be a disconnect between research findings and opinions of certain leaders. As a form of work organization, WFH, and especially hybrid work, has been shown to have significant benefits. The balance between office interactions and the ability to have a distraction-free work space along with the flexibility offered by WFH appears to be a great mix of both worlds.
It is worth adding that we are still in the infancy of WFH. Certain weaknesses of WFH, which are often operational (lack of access to managers, unclear project guidelines), can clearly be addressed with better work practices and potentially new technologies. Although offices currently provide these benefits, there may be other ways of getting these benefits provided by offices.
WFH also has tremendous well-being benefits (Angelici and Profeta, 2024) with people who have work from home arrangements feeling 30% to 45% happier. Workers are willing to sacrifice 8% of their pay for this benefit (Mas and Pallais, 2017), as WFH generates significant value for them.
Rather than fighting against WFH – which feels similar to the actions of luddites fighting the introduction of looms – the WFH method should be considered as part of any work organization structure. Maybe ‘work from home’ itself should be re-named (finding your ‘Optimal Work Environment’?) since the main findings from the research so far suggest that the current ‘100%’ office approach appears to be, on average, inefficient. The current implementation of WFH might not be perfect for every job, and it might never be for certain roles, but it is clear that it has many benefits and its development should continue.
For more research coverage on WFH – please read our previous articles that cover the benefits and drawbacks of WFH:
Interesting Reads from the Week
GDP Nowcast: GDP Nowcast estimates that US first quarter GDP (Gross Domestic Product) fell by 1.5% (annualized). It appears this is predominantly driven by a fall in net exports (imports much higher than exports). This would be in line with economic behavior that firms will want to import more things before tariffs kick in. This might result in a first quarter economic contraction.
- explains why egg prices in the US have went up, while prices for chicken meat have stayed flat.
- talks about the recent debate on low-birth rates and shows how in countries where the split of housework falls more on women, the fertility rate is lower:
“South Korea is a striking example of this. In 2024, it had the lowest fertility rate in the world — 0.75 births per woman. But South Korean women also spend nearly three hours more on housework and family care each day than men. In Portugal, the gap is even larger — women do four extra hours of domestic labour daily — and their fertility rate is barely above 1.4. Meanwhile, in Denmark and Sweden, where women do just under an hour more housework per day than men — a gap that still adds up to around 15 full days of extra work per year — fertility rates are notably higher at 1.7.”
For more additional insights, please see this presentation on the state of WFH in the US.
The difference between random and quasi-random is that random assignment occurs when the researcher assigns participants into groups, while quasi-random assignment is when it is done via some external policy, as is the case here with the CRRU using a scheduling algorithm.
At this firm, prior to the Covid pandemic, some call center workers worked from an office while others worked from home. Using call log data and workers’ schedules, Emanuel and Harrington were able to track how many calls per hour workers were responding to, while controlling for factors such as spending time on work training or other company tasks. During the Covid pandemic, all workers were forced to work from home. The workers that used to work in the office, were now also working from home. For the workers that used to work in the office that now worked from home, compared to their office level of productivity, the number of calls they took per hour dropped by 4%.
Although it is difficult to specify the exact economic impact, one way to think about this is that the work product of intermediate WFH workers would fall approximately in the 65th percentile of novelty rather than the 50th percentile.
These studies show that in some unusual circumstances a WFH arrangement might be neutral or somewhat positive. However, I do not agree that these instances are representative of most work done from home Therefore the conclusion that they say much about typical situations is dubious. To achieve an acceptable result, these workplaces have work that requires little interaction with co-workers, little supervision, and routine clerical work. They also require a reliable measure of both quality and quantity of work. The later is rare in most jobs. The measures of quality and quantity in the sample cases might suffice in routine clerical work when each unit of work is uniform and each unit can be verified. The cases that Musk has in mind do not fit these criteria.