Who Really Benefits From Remote Work?

This podcast is supported by Progressive, a leader in RV insurance.

RVs are for sharing adventures with family, friends, and even your pets.

So if you bring your cats and dogs along for the ride, you'll want Progressive RV Insurance.

They protect your cats and dogs like family by offering up to $1,000 in optional coverage

for vet bills in case of an RV accident, making it a great companion for the responsible pet

owner who loves to travel.

See Progressive's other benefits and more when you quote RV Insurance at Progressive.com today.

Progressive Casualty Insurance Company and Affiliates,

pet injuries, and additional coverage

and subject to policy terms.

I'm Hannah Rosen.

This is Radio Atlantic.

And today I have in the studio with me

Jerusalem Demsis,

who is the new host of a new Atlantic show

called Good on Paper.

Jerusalem, welcome to the show.

Hi, thanks for having me again. I'm so excited.

Sure. Okay, Jerusalem, I have a very Jerusalem question for you.

Have you in your personal life ever had a good on paper moment?

Like a thing you thought was good on paper that when you actually executed it or told your friends about it, it just was not. Yeah.
Well, when being asked this question, I'm realizing like how hard of a question it can be to answer. Yeah.
The thing that comes to my mind is I was very young. I must have been like 11 or something or 12 years old.
And I was really into rollerblading. I feel like anyone who was a kid anytime in either the 90s or the early aughts I mean they were roller skating rink parties all the time I don't even roller rinks were just like common and so I love doing that and then I was watching like that tv show rocket power which is really popular and I got a skateboard and in my head because I was young I didn't like think about like this is an entirely.
I was just like, these are the same things. This is the same family of sport activity.
I'm amazing at this one, so I'm going to be fantastic at this other. No problem.
And my mom was like, I remember vividly, she's like, just start slow. Do something.
Don't do anything weird. And I was like, in my little kid brain, I was just like, she doesn't get it.
I'm a star a star. There's like no problems here.
The first thing I do is I go to like, I mean, it's probably a relatively small hill now as an adult. But at the time, it felt like kind of like a big hill.
And I just like take my skateboard. I stand on it and just go down the hill and immediately break my wrist.
I immediately break my wrist. And, you know, like I remember it's so funny.
I have this vivid memory of being in the backseat of the car. I'm like crying.
My head, I think, must have been in my sister's lap because my mom is driving. And I'm just like in shock.
I'm like, how could that have happened? I'm a rollerblader. Like, how did I break my wrist here? And yeah, no, I never set foot on a skateboard ever again.
So low resilience. Amazing.
Because all of the good on paper concepts are in that story. It's like truly because it's not you didn't do anything wrong.
Like you had really good intentions. You had some amount of expertise and knowledge.
Like you weren't a total idiot. Like you kind of knew what you were talking about.
You made a leap of faith, which seems understandable. You're like, OK, I can do this one thing.
So I'm going to be able to do this other thing. But to any brain, and yet, the whole thing is a disaster.
Oh, yeah, yeah. It's very funny because it's also just like, I have not actually broken a belly outside of this one moment, and I think of myself as someone who liked to take risks, but after that I was like, no, you're not.
You don't take risks. You take very, very calculated steps that may not be the steps other people would take.
But like, you know, you're not getting on a skateboard again. So, yeah.
Well, I mean, you know, your show Good on Paper, I feel like in other hands, it could be arrogant. It could be like you idiots who don't know what you're talking about.
Yeah. But as from this story, it actually is okay because you know when you did that too.
So it's not just pointing fingers at other people. You're like, we all do this.
Like we all have these ideas we think are amazing and then they're not. It happens.
And I also think it's like in many ways too, like part of what the show is trying to do is why is it that we thought this was good in the first place is a huge part of the show because it tells you something both about how people or, you know, scientists or politicians think about a problem. And also like it helps you revise in the future because everything that we've tried, well, some things have been bad on paper, but most things people try are good on paper for a reason.
And so you're going to make this mistake again, like even times in the show where we're like, okay, well, now, now we feel like we know the answer because of this research or that research. In 20 years, there could be some other, you know, Atlantic journalist continuing in the 20th, you know, iteration of this show going, you know, actually, that was also something that was good on paper at the time.

So I agree.

I think a lot of it has to do with, like, how to develop an intellectual humility without losing the ability to make arguments. Right.
Well, listeners, Jerusalem has an amazing new show called Good on Paper, and I'm going to let her introduce it to you. My name is Jerusalem Demsis, and I'm a staff writer here at The Atlantic.
And this is the first episode of Good on Paper. Good on Paper is a policy show that questions what we really know about popular narratives.
Narratives do a lot to drive what our world looks like, whether they exist in the broader media ecosystem or as a consensus within a specific group of people, like economists or policy wonks.

But sometimes these narratives are built on shoddy ground.

One fact, or a set of reasonable facts, spins out of control and is woven into a tale that goes well beyond what we actually know.

This show came about as an extension of my own writing and reporting here at The Atlantic.

And I think that about a bunch of things from why it's so hard to build a wind farm in Alabama to why a bunch of people had babies during the pandemic, I'm struck time and again by the strength that certain narratives have. There are overly broad and often overly simplistic claims about the world that play a huge role in how our political system works.
And I'll be completely honest, there are plenty of times where I've realized those kinds of ideas are playing a role in my own thinking. That's sort of my beat.
I dig in when I see something that seems off or under-theorized

or at least not super fleshed out.

And while there's no one right answer,

the goal of the show is to figure out what we really know about a topic

and use research to get a deeper understanding of the truth.

This episode of Good on Paper is about the messy economics of remote work. Behind the scenes in this whole debate is a presumption that remote work is good for employees and bad for employers and bosses.
But is that true? You know, for my part, I've been a bit disillusioned by the remote work experiment. There are, of course, amazing benefits to remote work.
For those with disabilities or dependents, remote work can be more than just convenient. It can open up opportunities that hadn't been possible.
But at the same time, there have been some serious costs. Missing out on the social part of work.
Sure, there's some annoying water cooler chit chat, but I have a nagging feeling that I've lost out on important learning and connections by being remote. Most of all, it's not really clear to me how you make these decisions fairly.
Can my desire to work in person with my colleagues trump another person's desire to work from another city? It's still something I'm working out. A few weeks ago, I talked with Natalia Emanuel.
She's a labor economist working at the New York Federal Reserve Bank, and she wrote a really interesting paper that helps unlock the varied impacts of remote work. All right, Natalia, welcome to the show.
Thank you so much for having me. I'm really excited to be here.
Before we begin, I do note that the views I would express today are my own.

They don't reflect those of the Federal Reserve Bank of New York or the Federal Reserve Bank. me.
I'm really excited to be here. Before we begin, I do note that the views I would express

today are my own. They don't reflect those of the Federal Reserve Bank of New York or the Federal

Reserve System at all. So they're simply mine.
Yeah. So, I mean, you were finishing your PhD

when COVID hit, right? That is correct. How was that? Did remote work feel that different to you?

Like, I kind of imagine academics sort of like siloed off in their offices, like never speaking

to each other. Well, my co-author on two remote work papers, her name is Emma Harrington, who is now an awesome professor at University of Virginia.
She and I were randomly put into the same office in a second year of graduate school. And then partly because of that, we ended up becoming co-authors because before that, we actually hadn't really known each other particularly well.
So there is an element of, yes, we were kind of siloed. Yes, we were in the basement with almost no light at all.
But by being in the same windowless office together, we did form a sort of a nice bond that way. This feels like an econ paper that's like come to life.
Isn't this

like a finding? Yes, exactly. Academics that like sit near each other, tend to co-author or something?

Correct. Yeah.
In terms of the actual COVID during the job market, it had a very important impact on

us, which is that all of our job market was done remotely. So we were doing interviews remotely.

We were doing fly outs to visit the potential places we might take jobs. All of that was

not a actual fly out. That was a Zoom fly out.
And so that was the place where it had more impact

Thank you. fly outs to visit the potential places we might take jobs.
All of that was not a actual fly out. That was a Zoom fly out.
And so that was the place where it had more impact, perhaps, on the actual paper writing. Did you think it affected the interviews or anything? So purely anecdotally, I would say the people who I have given talks to remotely remember me and remember my findings less than when we were in person.
Wow, just because everyone's like kind of doing, I don't know, the New York Times Connections games while they're listening to you. That makes sense.
I imagine it was email, but I think you have a more enjoyable thing. Maybe they liked my talk more because at least they were doing something fun.
Yeah. So I feel like before we get into the meat of your study, there are very different estimates about how many people are actually remote working right now.
And it led me to realize, like, you know, how do we actually know what's happening? Like, do you have a sense of how many are remote working? Like, why does it feel like we're getting different answers from different data sources? There is a big difference among different ways that you could ask this question and sort of exactly what you mean by remote work.

Does that mean that there is no place you have to go to for your work? Does it mean that you have to be in your workplace's office as opposed to a cafe shop? Does it simply mean that like you have to have left your bedroom? You also can get different answers when you're asking, are you fully remote versus are there certain days of the week when you are remote versus how many hours a week are you remote? And so those two dimensions can give a lot of variation in terms of exactly what number we're getting. So the one that I'm going to just try to use in my head and for listeners is what the BLS, Bureau of Labor Statistics, is doing.
So in 2024 in February, their survey data shows that 77% of people did not telework at all. Around 11% of people teleworked all hours, and roughly 12% teleworked some hours.
So that's kind of, it feels like this is a really big conversation for like 12% of the population to be fully remote working. Like, do you feel like that's like an outsized conversation that we're having about remote work? Well, I think the 77 number of people who are not working remotely, that makes a lot of sense insofar as some jobs are just really hard to do if you're not on site, right? Being a car mechanic, very hard to do if you are not actually at the car.
Similarly, you know, trauma surgery, maybe one day it'll be done by robots who are, and you know, the robots are controlled by people who are far away. That's not how trauma surgery is happening right now.
Similarly, we're not thinking about sort of occupational therapists or, you know, nursery school teachers. So many of those jobs, like there just isn't a possibility of them even being remote.
And so what we're thinking about here are the jobs where, you know, there is a possibility of being remote. You can imagine sort of sales, customer service, consultants, software engineer, many jobs that are sort of more computer-based.
Those are the ones where we should be thinking about the sort of remote work is a possibility. And, I mean, the quintessential people who can work from home are probably like software engineers and coders, which brings us to your study.
So you have a working paper at the National Bureau of Economic Research and BER, and it came out last November. Can you tell us about it? Sure.
So we are looking at software engineers at a Fortune 500 company. And this is a sufficiently large company that they have on their main campus two buildings where the software engineers sit.
And those buildings are about 10 minutes apart. Well, 12 minutes if you're on Google Maps, 10 minutes if you're me.
We found that some of the people who were on teams where everybody could be in one building, whereas because there's sort of not as much desk availability, some teams actually had to be separated across those two buildings. And so the teams that were separated across the two buildings had most of their meetings online.
Because if you're only having, you know, a 20-minute meeting, you're not going to spend exactly the length of your meeting walking there and back. And so we can sort of see beforehand what happened to those particular teams.
And then once the pandemic forced everybody to work remotely, we can see what happens thereafter. And so we can use the teams that were already meeting remotely, and they're sort of our control group, that they're remote before the pandemic, they're remote after the pandemic.
Whereas the people who are on one building teams, they were sort of with the rest of their colleagues, And then after the pandemic, they're working remotely. That's sort of an interesting context to look at from our perspective, because it allows us to understand there is a measure of productivity.
And then there's also a measure of digital collaboration. And so we were trying to understand sort of what remote work does for the pieces that you might learn from colleagues, right? There's another study that finds that sort of a sixth of all skills that one acquires over their lifetime are coming from colleagues.
And so we are very interested in the impact of remote work on this sort of collaboration and on the job training. And so we also think that software engineers are particularly interesting because in many ways, it's sort of the best case scenario for remote work.
So for one, all of their output is digital. Also, software engineers have established mechanisms for giving each other digital feedback on their code.
And that was something that they had sort of industry standard and has been for decades before the pandemic. So what are your main results, right? You're observing these software engineers.
And as you say, like these software engineers are basically just coding full time. They're just writing a bunch of code and they're getting comments on that code.
And that's kind of like how you're looking at feedback. So what are the findings of that observation? Yeah, so we're finding that the folks who were in person with their teams,

they were in the same building, we're going to call them one building teams,

they were getting about 22% more feedback from their colleagues on their code. So they were

just getting more skills, more mentorship when the offices were open. And then when the offices

closed and everybody was going remote, pretty immediately we see that gap closes.

And so then sort of everybody is getting less feedback than they were. And this is sort of useful as a counterfactual because if you imagine you're saying, oh, well, they're getting 22% more feedback.
Well, maybe that's just because they tend to be chattier. Or maybe it's because they really actually need that feedback a little bit more, the people who are on one building teams.
If that were the case, then even after the offices close, that would still persist. Whereas if this is something really coming from, you know, being in person with your colleagues, then that gap would close.
And that's exactly what we find. So there are 11.5% more people commenting on engineers' work if they're in one building teams than if they're in the multi-building teams, right? So there are a lot more people commenting on the work.
There are a lot more people commenting on your work if you're on a one building team. So what is happening there? Why is it that someone who's in a one building team is seeing more comments? So we look at this sort of in terms of the exact type of comments.
So part of this is they're just getting more comments on the initial go, but then also they're asking more follow-up questions and then getting more replies to the follow-up questions. And so we're seeing sort of the depth of conversation is partly driving this.
We additionally see that this is happening in terms of speed, that they're getting faster feedback as well. And so sort of there are many dimensions here.
I would also put a small asterisk here, which is that we're measuring this in terms of the digital comments that they're getting. But people who are in person, it is much easier to just sort of turn to your neighbor and say, hey, can we just talk about this for a quick second? And so if we think that that's sort of happening more among the people who are sitting next to each other,

then the estimates that we're getting are actually lower bounds.

And so what's the effect of all this?

What's the effect of getting more comments?

There are a number.

The first is that, as you might imagine,

if they're working on building skills and responding to these comments,

their actual output is a little bit lower.

So they're producing fewer programs overall.

And accordingly, because they are producing fewer programs, they also are less likely to get a pay raise. But once the office is closed and sort of that level of mentorship has now equalized, the people who had been working on building their skills, they're actually more likely to be getting pay raises, and they're actually twice as likely to be quitting to go to a higher paying job or a job at a higher paying company.
And so it really depends on the timeframe that you're thinking about this. In the short run, it looks a little painful because they're not doing as well.
But in the long run, it really is sort of you're seeing the fruits of their labor. So I find this really interesting, right? Because what it indicates is that there's sort of like this investment that happens early on in someone's career.
And then when they go remote, the people who had that kind of investment are able to still sort of capitalize on it. But like, in the time, they're going to look less productive than their more remote peers.
Like those remote peers are like just banging through code. They're not having to respond or engage with their mentors or with the older engineers.
you know it's a strange finding, right? Because it would indicate that managers would really prioritize and see that remote work was doing well in the short term. Totally.
And I think that sort of that is consistent with what we saw at Meta, right? Early in the pandemic, Mark Zuckerberg was like, yeah, this sounds great. People seem to actually be more productive when they're remote.
And then sort of three years in, that's when Mark Zuckerberg was like, actually, let's come back to the office. It seems that people actually are more productive when we have some amount of in-person time.
And so it does seem as though there's sort of, it does take a little bit of patience to be able to sort of realize these different effects over different time horizons. Wait, you mentioned meta.
Is this meta? So I'm actually not allowed to share what company we're studying. Okay, great.
Well, I will just, in my head, imagine a giant campus in Silicon Valley that has multiple buildings where software engineers work far apart. That sounds like a perfect thing to imagine.
Their own conclusions. And so do these findings contradict earlier findings in this space? Existing literature about remote work and productivity, as I've mentioned before, it's kind of mixed.
But there's like the seminal 2015 study from Nicholas Bloom where he looks at a 16,000 employee company in China. And, you know, the study designed there, it's like, you know, it's employees that volunteer.
They then randomly assign those to either be work from home or in the office. And they find that work from home leads to a 13% performance increase in productivity.
So both more minutes per shift and more calls. It's a call center.
So it's more minutes per shift that they're making calls and also more calls per minute. And so that feels like very different than what you're finding here.
Yeah. So first of all, I think that it is totally possible to have different findings in different settings.
One of the things that makes Nick's study particularly interesting is there, it was, as you mentioned, all volunteers, right? These were existing workers who had been at the company already and they volunteered to go remote. So that's not necessarily the case when we're thinking about the pandemic.
Not everybody volunteered to be remote. Also in that context, everybody had to have their own room to work in as a specific workspace as separate from people who are like working on their bed.
And so that also could change it. And so you do see potentially different outcomes there.
Also at a travel agency, that is pretty siloed work. Whereas at software engineers, they do sort of need to understand what this code base is doing, how people have been thinking about that particular function already.
And so there is a little bit more of a collaborative nature there. The other thing I would note is that eventually,

remote work unraveled in that context because there were fewer promotions happening among the

remote workers. And so people ended up wanting to come back to the office because that's where

they got the visibility to be able to sort of get the promotions that higher performance

Thank you. happening among the remote workers.
And so people ended up wanting to come back to the office because that's where they got the visibility to be able to sort of get the promotions that that higher performance really warranted. And so they weren't getting promotions because they were doing worse work or they weren't doing promotions because kind of managers had this attitude that like people who are in person, who they're talking to in the office, like those people are just more worthy of promotions.
Well, I wouldn't say that they were doing worse work as according to the next paper, it seemed as though they were actually doing better work. They were overall more productive, but it does seem as though there is a sort of disconnect between pure productivity metrics and sort of how the sort of human component of promotions.
And so, I mean, you have a 2023 study where you look at a call center. It's a U.S.-based call center, and I'm not sure how else it may differ from Bloom's study, but you find that pre-COVID, remote workers were answering 12% fewer calls per hour, and that feels like, you know, there's something going on that's stably less productive about remote work, even kind of in the same sort of work context.
So what's going on in understanding the differences in your findings versus Nick Bloom's? Yeah. So in our study, we are finding that before the pandemic, the people who elected to work remotely, at least in this company, which again, as you mentioned, we were thinking about a Fortune 500 company and their customer service workers.
And there we found that the people who chose to work remotely tended to have lower productivity on average than the people who chose to be in person. And so that's sort of

what economists would call, you know, negative selection. But that is also consistent with if you

anticipate that the people who are going to get promotions are those who have closer connections

to the managers and are those who are going to be in person. And you know that you're sort of, you might be, not to use a horrible pun, but you might be phoning it in a little bit, then that would make sense that sort of you would be more willing to be remote.
Now, of course, I have no idea what was in each individual person's mind, but that is consistent with sort of understanding that there is a promotion penalty to being remote. So returning kind of like to your original new study about also about remote work.
But so I think the thing that's really interesting about the research you find is also the sort of junior versus senior benefits to remote work, right? So like, I really want to talk about how different it is if you're an early career software engineer versus a late career software engineer. What happens to people early career versus late career when it comes to remote work? Like how does that affect their productivity? How does it affect how they do their jobs? Whether what research they're getting and their long-term outcomes? So in general, it's the people who are most junior who have the most to learn and are getting the most comments and sort of therefore having to do the most learning.
And who's giving this feedback? Well, that's the more senior people. Those are the people who have been with the firm a lot longer.
And we see that sort of the hit to productivity is actually happening both among junior people, but then particularly it's concentrated among the senior people who then have to be really understanding somebody else's code and thinking deeply about it and giving them feedback to try to think, oh, how can I help this person grow and how can I help make sure that this code is doing well? And so that meant that for the senior people, there was a cost in their productivity from being in person and providing all of that feedback. And so that means when they go remote, particularly the senior people's productivity actually increased.
And so again, for them, like you could see a sort of boost in productivity right at the beginning of remote work. And then from the firm's perspective, you could imagine that that might not persist forever if you're then getting your junior engineers who aren't getting as upskilled as you might hope.
So like senior folks are just like, thank God I don't have to answer all these comments all the time. I can just do my job and that benefits them.
I wonder though, I think this is really interesting, right? Because popularly understood is that people who are young really want to work remote and that older people are more willing to come back to the office for whatever reason. Why is there this sort of disconnect if it is the case that young people are really missing out on this both productivity enhancing, but as you said, wage enhancing and promotion enhancing benefit of learning from senior engineers? Like why aren't they clamoring to get back in the office? One hypothesis is that they simply don't know, right? Maybe they are not aware of the benefits of mentorship from being in the office.
Maybe they're not aware about sort of how that mentorship and the skill building actually translates into future jobs, future earnings. So that's one possibility.
Another possibility is maybe they have a different value system, right? Maybe they're willing to say, look, my job is not the top priority for me, and it's much more important for me that I am spending time with my roommates, my neighbor, my friends, my loved ones. That's a possibility.
I think another possibility, and our paper gives a little bit of evidence, is that if you have even one colleague who is remote, that that yields about 30% of the loss from having everyone be remote. Wait.
So if just one person on your team goes remote, you lose all of that benefit of being in person? Well, a third of it, yeah. A third of it.
That's huge. Right.
It's huge. Does it scale up? If it's a second person, is it like 40? Did you find anything there? We didn't actually look at that.
But it is a huge impact. Really, in some ways, that's validating.
It means every single person really matters. But if it's the case that when they come into the office, not everybody is there, and so they're still doing some remote Teams meetings or WebEx or whatever it is while in the office, then it's possible that they're not actually getting the whole benefit of being in the office.
And so perfectly rationally, they're saying, maybe it's not so much. Maybe I'm not getting all of this mentorship.

And so there are, there you go, three hypotheses. I am partial to the last two things you said.
I

don't really buy hypotheses usually where they're like, someone's just being dumb and like they're

doing something that's bad for them. I usually buy like, they either are prioritizing something

else. Like, you know, not everyone wants to be a productivity maximizing machine.
Like they may

want to just not have a commute. They may want to live near their family, whatever it is.
And

I don't know. are prioritizing something else.
Not everyone wants to be a productivity maximizing machine. They may want to just not have a commute.
They may want to live near their family, whatever it is. And I think also this last thing that you said is really important too, because the Atlantic offices are open, but there's a lot of hybrid work.
And so you're coming in on a day where there might be 10 people on your team, or they're coming in on a day where you're like, wow, I'm the only person on my team here. And those are very different days, and they're very different things you might get out of that.
So that hits stronger for me. One of the things that's pretty interesting is that we find even when you're in a building with colleagues who are not on your team, we still find a sort of bump in the mentorship and the feedback that one gets.
And it's not from your teammates. Then of course, it's from the non-teammates, but that there still is an element of enhanced mentorship, feedback, collaboration, simply by being around people.
We're going to

take a quick break, but more with family, friends, and even your pets. So if you bring your cats and dogs along for the ride, you'll want Progressive RV insurance.
They protect your cats and dogs like family by offering up to $1,000 in optional coverage for vet bills in case of an RV accident, making it a great companion for the responsible pet owner who loves to travel. See Progressive's other benefits and more when you quote RV insurance at Progressive.com today.
Progressive Casualty Insurance Company and Affiliates Pet Injuries and Additional Coverage So I think that probably the most interesting angle in your piece is the angle on gender.

And can you tell us a little about this?

Like, what is different about how women in this firm receive feedback on their code? Yeah. So before the pandemic, we find that female engineers are receiving about 40% more comments on their code than our male engineers, giving us an effect that's roughly twice the size as it is for male engineers overall.
And so we're finding that this mentorship is particularly important for female engineers. And to sort of unpack where that's coming from, we find that the female engineers are much more likely to ask questions when they are in person.
So, I mean, when I first heard this, I was just like, okay, are they getting more feedback because people are just nitpicking women's code? Like, how did you decide whether or not this was actually actionable feedback or if it's just people being sexist? Yeah, this was one of our first concerns. One of the first people we presented to said, are we sure this isn't mansplaining? And so what we did is we took a subset of the code of the comments and we gave them anonymized to other engineers and said, is this comment helpful? Is it actionable? Is it rude? And we then took their reviews back and we found that they are equally actionable, not sort of differentially nitpicky for female engineers.
And so it does really seem as though these are substantive, meaningful comments, but not simply mansplaining. And interestingly, not differentially rude either.
That's great to hear, actually. And sorry, these external reviewers, they were blind to gender when they were looking at the code, right? They were blind to gender.
They were blind to seniority. They were blind to whether you were proximate or not proximate to your colleagues.
All they saw was the comment. And what that raises for me, though, is this question, like, if women are disproportionately getting actionable feedback, is the claim that women's code is just worse than men's? So we don't actually see the code itself, but we can see that sort of we're not finding they're sort of more problematic overall.
It's not as though, like, we're seeing, oh, there's bigger issues brought up in the comments of sort of they will always break or something like that. So I find this interesting, and I also think it's interesting because this is not the only plane on which women are affected differently than men in your study.
So you have this finding that junior women are receiving a lot more code and a lot more actionable feedback, and it's benefiting them potentially down the line. But you're also finding that the people who are giving them all that feedback tend to be senior women engineers who themselves, as you said before, giving all that feedback takes time.
That's something that hurts your productivity. So that cost seems disproportionately borne by senior women.
Yeah, I think you said it exactly right, that the feedback that's going to both female and male junior engineers, a lot of that is coming from female senior engineers. And so that the giving of the mentorship is also coming from female engineers.
And so we see exaggerated effects on both the benefit sides for females. So junior women are getting the benefits, but also senior women are paying the price.
And so when they go remote, do senior women get more productive? Yeah. That's really weird, right? Because I feel like the dominant frame for the pandemic and gender was mostly around this idea that women were really, when they were going remote, were being now doubly burdened, right? So you go home and if you're a mother, you have to do more childcare.
Often you're finding you have to share space with your male partner in a lot of cases. And it was kind of like the sense that like potentially women are now having to be doubly burdened by the responsibilities of home and responsibilities of work.
Obviously, it's not contradicted by your evidence, but like, were you surprised by this finding? So I guess I would make one technical point and then one overall comment. The first technical point is that this is sort of why it's really, really helpful to have a control group, right? Because in both of our groups, both the people who are working in one-building teams beforehand and the people who are in multi-building teams and therefore a little remote beforehand, both of them would be sort of similarly burdened by the pandemic.
And so we can sort of difference out the impact of the pandemic and really just zero in on the effect that's only coming from working remotely. So that's sort of one component there.
The other piece that I would mention is that in our sample of engineers, only 16% are parents. So that doesn't seem to be sort of the main component here.
And in some ways, I think that that's, while not sort of helpful in terms of thinking about the impact of the pandemic, potentially helpful in terms of thinking about remote work long term, where when we're thinking about remote work post-pandemic, we're not really thinking about, oh, but you will also be trying to supervise your fifth graders, you know, language arts exam. It's hard for me to know how generalizable these findings are.
And, you know, basically every single study I see, major study I see on remote work, it's mostly done in the context of either software engineers or call center employees. And those feel like just two very specific types of jobs and are not maybe similar to a lot of other jobs that are potentially work remote, whether it's someone you're working in HR or or you're working in media, or you're working as a lawyer, or anywhere in the legal profession.
It's hard for me to know how much you can take away from this and apply to other contexts. How do you think about that? So, yes, you are totally right that the existing literature feels as though really focuses on sales, call center, and software engineers, partly because those are places where we have really good measures of productivity.

I would love to be able to think about this for other occupations, but I do think that we have a bit of a quantification problem.

As I mentioned earlier, I think one of the things that's sort of useful in our context is to think that software engineering is probably most amenable to remote work and that other contexts that sort of don't have these established ways of giving each other feedback online, that don't have very structured systems for how to meet. So software engineers often work on the agile system of meeting where they have daily stand-up meetings that happen regardless of whether you're in person or not.
They have very structured ways of exactly when they are going to be doing sort of a sprint on exactly what type of work. And they have a lot of coordination around who's doing what when.
And so for occupations that don't have either of those things, of digital means of giving feedback and that sort of meeting structure, that you can imagine remote work is likely to work less well for them. But that feels like something that a lot of different industries could innovate on, right? I mean, I think one of the things that I've heard pointed out is how many more patents there are now on remote work technologies.
It just feels like not even just those sort of technologies that help make it possible for any, you know, a lot more firms to work at home, but also just like the cultural technology, like the fact that you can just ping someone on Slack, the fact that you can just, you know, huddle quickly. Clearly, I use Slack way too much.
Or you can, you know, figure out a way to have a stand up with your manager. Like, I mean, in a media environment, usually kind of like walk over a newsroom, but people know how kind of standing meetings that they will just have with their manager.
So how much of that is not portable to other workplaces? Oh, I totally think many of these are portable. And I do think that we're going to have some growing pains as people realize, oh, I could just have a standing meeting and then realizing that, oh, but now I have a standing meeting with 15 people and it's taking up half of my Friday.
And so I do think there will be some growing pains, but that there is quite a lot to learn from other organizations that have already done remote work pretty effectively. And so zooming out a bit.
So Adam Ozmik, he's also a labor economist and he's also a longtime booster of remote work. He once half jokingly said that skeptics of remote work could basically be described as either extroverts, urbanists, workers in obviously non-remote occupations, and downtown office building owners.
And so, you know, a Venn diagram of labor economists and urbanists has like significant overlap. And so I wanted to ask you if you think your background as labor economists sort of biases you against remote work or thinking that it's positive.
Do you feel like that you're coming into the work kind of feeling like it's not going to go well? Or how do you think about that? Well, I'm definitely not an extrovert, so we can cross out that one. I would not say I had strong priors going into this.
It was one of those topics that I was genuinely extremely excited to see whatever the results would be and could totally have spun a story that it could go in either direction. But I mean, I guess, like, do you think that you would be surprised if long-term remote work was viable at large scale across these firms? I mean, even what you said at the beginning of when we started chatting about your ability to meet Emma and your co-author and work with her.
I mean, those kinds of findings are often really strong underlying belief systems for labor economists. I do think there's a totally a world in which remote work really takes off and we can have massive productivity gains.
I think that this comes with a lot of growing pains that we were discussing of trying to figure out exactly how we can still make sure that we form deep connections, have a lot of mentorship. And I think we see a lot of firms doing some incredibly creative things, whether that's sort of quarterly off-sites or sort of teams coming in at regular intervals and trying to do sort of a round robin of who's meeting with what.
And so I do think we're in a period of experimentation while we're trying to learn how this is going to work. But yes, I would definitely say that there is a world in which this does work and that we have to figure out exactly how it's going to work.
So, you know, we've talked a lot about productivity here, but life isn't just about productivity. There are lots of reasons why someone may or may not want to work remote.
What's your sense of the impact of remote work on individual well-being? This is the question in many ways. That on the one hand, maybe it allows folks to live, you know, close to their family, their community.

And so there's a really wonderful gain in terms of people's well-being because they have these strong social connections. On the other hand, in many decades past, a lot of people found their friends at work and sort of many enduring friendships, many marriages originated in work.

And so if people are not making those connections at work, there has to be some other way that they are going to be able to make those social connections that are going to sort of fulfill their needs. And so maybe that substitution is happening.
I don't think we have a great idea yet. And so I think you, again, could imagine it going either way.
And I am extremely excited to see research coming out

that can give us insight as to which one will weigh more strongly.

I'm a little bit pessimistic about it.

And in part I am because I feel like the trend of work technology

has been to just eat into more and more of our leisure hours.

You know, email gets invented and all of a sudden you leave the office

and it doesn't mean that you've left the office.

And Slack gets invented. Now you have to be instantly available, even if you're like in the bathroom, like you know that your boss's message due.
And then now, I mean, there was a 2021 paper that looked at GitHub activity and found that users were more likely to work on weekends and outside nine to six hours when they went remote. And it feels to me that this is just another step in the machine of like, okay, remote work means now that there aren't even defined hours.
And in some sense, theoretically, that could mean flexibility. But in another sense, it can mean your entire life is now work.
I think that's totally possible. I would say that there's a world in which sort of that GitHub finding that you mentioned is actually a really good thing, right? So imagine the world in which I know exactly what my output has to be.
I know the product that I need to create, but I actually want to stop work at 3 p.m. so I can pick up my kids from school, hang out with them until, you know, 7.30 or 8 when they go to sleep, and then I want to put in my extra two hours that, you know, would have happened between three and five, but now can happen after bedtime.
So maybe that extra flexibility is actually welfare enhancing and the people who they're studying are actually really happy about that. And so I think simply based on that statistic, it's not obvious to me whether we think of this as a good thing or a bad thing.
I do think sort of work creeping and taking over one's entire life so that there's nothing else there and there's no time for anything else. I think that's almost certainly a bad thing.
But again, I'm not sure exactly on sort of how to think about the welfare implications there. Before we close things out, our last question, what's an idea that you've had that was good on paper? So I sew a lot.
I've sewn 17 quilts, several wedding dresses, only one of which was for me. And so one idea that I think tends to look good on paper is the home sewing machine that is computerized.
Oh, what is that? It's just a sewing machine that has a screen on it. And then you can say, oh, do this embroidery pattern and it'll, you know, output that.
And I would say for the type of sewing that I was doing, it was 100% useless. It meant that it was much harder to maintain, much harder to troubleshoot.
You can't do your own sort of oiling and maintenance in the same way that you could for a mechanical sewing machine. And at one point, the sewing machine actually just decided to only run in reverse and rather like, you know, driving in New York City in reverse.
It's possible,

but it's, you know, a little anxiety-inducing, not the world's safest thing. So I ended up reverting back to sort of this sewing machine that's fully mechanical, was made in 1910 by Singer's sewing machine, is actually foot-powered, hadn't been used in the entirety of my lifetime, but, you know, with a little bit of elbow grease was totally great.
So it was one of those things that sort of in the abstract seemed great and in real life was not.

Well, this feels like... with a little bit of elbow grease was totally great.
So it was one of those things that sort of, in the abstract, seemed great,

and in real life was not.

Well, this feels like a metaphor, a productivity-enhancing machine

that actually reduced your output.

Yeah, well, on that note, well, thank you so much for coming on the show, Natalia.

Thank you so much for having me.

Good on Paper is produced by Janae West.

It was edited by Dave Shaw, fact-checked by Anna Alvarado, and engineered by Erica Huang.

Claudine Ebade is the executive producer of Atlantic Audio, and Andrea Valdez is our managing editor.

And hey, if you like what you're hearing, please leave us a rating and review on Apple Podcasts.

It's how people hear about the show.

Or you can let a couple friends know on your own. My name is Jerusalem Dempsis, and we'll see you next week.
So that was the first episode of the Atlantic's Good on Paper podcast, hosted by Jerusalem Dempsis. Links to subscribe are in the show notes for this episode, or you can just search your podcast app for Good on Paper.
There are already several other great episodes in the feed about whether young men are really becoming more sexist, about who really protests and why, and many more to come. This episode of Radio Atlantic was produced by Janae West, edited by Claudina Bade, and engineered by Rob Smersiak.

Claudina Bade is the executive producer

of Atlantic Audio,

and Andrea Valdez is our managing editor.

I'm Hannah Rosen.

Thank you for listening.