The New Geography Of Labor Markets

>> John Taylor: We're very happy to have Steve Davis speak to us today. He's both at the Hoover Institution in SIEPR. S-I-E-P-R. Anyway, it's the title is the New Geography of Labor Markets, it's a jam packed paper, lots of interesting things happening and we're anxious to hear what you have to say.

 

So go ahead Steve.

>> Steven J. Davis: Thank you John and thanks everybody for coming. So this is a talk they'll actually draw on several sources. Mainly the paper that's circled there in the middle, same title.

>> Speaker 4: Why did you have your name there?

>> Steven J. Davis: Why is it my name there, well, I'm an author on all those papers but as you can see I need lots of help to get anything done.

 

So two of my co authors or collaborators on this worker here, Nick Bloom, which I think everybody knows, and Mert Akan, which you will know. He is an incoming PhD student at Stanford, Stanford Econ, so he'll be joining, he's been working with Nick and me for a while and he'll be joining the Econ program here.

 

It's gonna be empirically-oriented paper but before we march off into the forest and get lost in the trees, let me give you an overview of the main points I wanna leave you with. The first one you already know and I've spoken about in this forum before and that is there's a great deal more work from home than there was before COVID.

 

But I'll remind you of the magnitudes and also I wanna counteract any media narratives that may have contaminated your thinking about what's happened to work from home. And then I wanna turn to one important consequence of that whose ramifications are, I think haven't received much attention yet and that is that many more workers now live far from their employers work sites.

 

Just to give you an idea, one metric of this is the fraction of workers, at least in our data set, which I'll tell you about, who reside more than 50 miles from the employer's work site. That was 4% before the pandemic. It's about 9% by 2024 if you look at people who were hired since March 2020, which is probably a better indicator of where we're headed, it's more like 12%.

 

Okay, so big increase in the fraction of people who are not very, very close to their employer's work site. Related to that is that work from home facilitates relocation to states with lower tax rates and areas with lower housing costs. So we will provide some evidence on that and I'll show you that rates of net migration away from high tax states rose significantly in the wake of the pandemic, at least among the affluent.

 

And something similar played out in housing markets as well. These out migration pressures are most acute for cities with high housing costs situated in high tax states and that have an industry mix that lends itself to remote work. And there's a city not very far away from here that is a poster child for all of those characteristics.

 

Which gives you one reason why, one sense of why San Francisco has had a tough time in the wake of the pandemic. It's not just bad city politics. I'll offer a very rough estimate of what the aggregate state level income tax revenue consequences are of this relocation that's been triggered by the rise in remote work after the pandemic.

 

There's some strong assumptions built into this exercise, but it's the best we are able to do thus far. But the numbers are non trivial. So we estimate that something like 7 to 8% of aggregate state income tax collections were lost because of the systematic migration of people, mostly affluent people, away from high tax states.

 

Yeah.

>> John Cochrane: The high housing costs remain, so somebody else must be moving in or blowing up apartments.

>> Steven J. Davis: There's an overall increase in the demand for space that happened as a result from work, from home. And Nick has written extensively about this, but there's been kind of a change in the shape of the rent gradient.

 

So how real housing costs have declined in many urban centers, including painful for me in Chicago, which I moved out of, to come to this glorious location. So, that the residential housing story is, it's a complex one. Overall price is up, but the spatial structures shift quite a bit.

 

Well, also provide some evidence on the magnitude of the gains in tax taxes and housing costs that are enjoyed by the people who moved in response to this newfound locational flexibility after the pandemic. The numbers are stunningly large. So the main exercise I'll run through in this respect is focused on people with annual earnings more than $250,000 a year and among those who switch states, okay, so we're looking at the switchers now.

 

The average across the whole country. So that's people switching into and out of these high tech states. The average reduction in the top state level income tax rate marginal rate is 5.2 percentage points in 2020. There are also sizable reductions for people who moved in 2021, 2022 and 2023.

 

Yes.

>> Speaker 4: If you earn money in California, it doesn't make any difference where you live?

>> Steven J. Davis: Yeah, yeah, let me. Hold on. You're getting into the trees now. I told you I would. I told you I would.

>> Speaker 4: There were some very large drop outside of making outside.

 

 

>> Steven J. Davis: Yeah, I'm sure it's not news to you, but if you move from California to Texas, your top marginal tax rate is gonna drop by 12 percentage points, okay?

>> Speaker 4: Yes, I get that, but you really asked the question.

>> Steven J. Davis: Yeah, okay. I'm not quite sure what the question is now.

 

I'm just telling you, on average, what happened to the top marginal tax rate integrating over the people who moved in our data set in 202, okay? And there were somewhat smaller numbers through 2023. Now, if you look at people who switch zip codes and slightly different threshold here, 150k who switched zip codes but stayed with the same employer, which has fewer.

 

The measurement problems you were just describing, they experienced a 16% reduction in local housing costs on average, which is also huge. So the point I wanna leave you with here, this is something that has not received much if any attention in the literature thus far. There are many sources of private and social welfare gains, or at least changes associated with the shift to work from home.

 

And they've been the subject of many papers, some that I've written with Nick and which is involved productivity effects, commuting, the value of personal autonomy, flexibility and time use over the day, and relaxation of the joint location constraints on two or households. So these tax and housing cost savings that I just outlined are on top of all those other things, okay?

 

And if you think about that for a minute, you get some sense of why there is so much resistance in some quarters to return to office mandates after you've already relocated your household and so on. Okay, so another thing that's happened is that I think this is the first paper to provide any evidence on this point.

 

Separation in hiring behaviour differs between more and less distant employees and by distant employees or far employees, I mean those who reside in a location which is far from their employer's work site. Okay, and the short story of what you see here is from the perspective of employers, they treat labour inputs on both the recruitment margin and the separation margin as more flexible when those people live farther away, okay?

 

That may not be too surprising, but I think it's an important consequence partly because feeds into the last point that I wanna talk a little bit about, which is that the firm level labour market footprint has become more geographically dispersed. This is a statement on average, across firms become more spatially diffused, okay, and that's an ongoing process.

 

That process hasn't fully played out yet. And that's because workforces are continuing to turn over and newly recruited employees tend to live a lot farther away from their employer's work site than people who are already working for that employer before 2020. I'm gonna show you that, I'm just giving you the overview here.

 

And this is important because it has implications for the disruptions and the costs that are associated with job loss in future labor market downturns and future labor market restructurings. They will be more spatially diffused than what we have experienced in previous recessions. And I've written on the costs of job displacement with Til Von Wachter.

 

I think one of the reasons that certain types of job loss are so painful for individuals and for families and for communities in some sense is they tend to be highly concentrated in space and time. Okay, and what this is saying is that going forward, they'll be less concentrated, at least in space, if not in time.

 

Okay, so that's the overview, John.

>> John Cochrane: It would also imply better matching, which should show up in higher salaries and higher productivity.

>> Steven J. Davis: Yes, yes. I won't have any evidence on that today, but I do think that's buried in the illusion I made about productivity effects of work from home, one way in which these productivity effects can show up is in better matching.

 

Okay, so let me dive. Let's march into the forest now and show you some of this, so this is work from home rates over time. This is our favorite survey, survey of working arrangements and attitudes, which Nick Jose Maria Barrero and I run. It's been running monthly since spring 2020.

 

You can see the long time period there. There's a government survey overlaid, which it doesn't cover quite as long a period. Tells a very similar story. The bottom line, thing to note that I want you to take away from this picture is two things. One, about four times as many workdays, full workdays, are happening remotely now as before the pandemic.

 

So it's somewhere around 28% now according to our estimate. And the second point is not much has really changed in this respect since early 2023. So at least in terms of the overall work from home rate, things have kind of settled down, okay?

>> John Taylor: We've kind of reached an asymptote.

 

 

>> Steven J. Davis: We've asymptoted. Now not all of the consequences have reached an asymptote, okay. But at the level we-

>> John Cochrane: All employees in the US?

>> Steven J. Davis: There are some restrictions. This particular series screens out. So it's people 20 to 64. We've screened out those who earn less than $10,000 a year.

 

So people who are working part time in a convenience store or a fast food outlet or just in the summer, they're kind of screened at it. And those jobs are just before.

>> John Cochrane: This is an amazingly high number still, okay, it is an amazingly high number.

>> Steven J. Davis: So that's one reason I'm showing it to you.

 

The work from home rates reported by executives we've fielded in the survey business Uncertainty, a survey that Nick and I design and field in cooperation with people at the Atlanta Fed House. Very similar story, once you do your best efforts to harmonize the frame of these two surveys and I can show you other evidence, non survey evidence, typically like what you get from Castle Security office swipes.

 

Those are very good data, but they're more narrowly focused on office buildings in cities that also kind of tell you a similar story. So that's the US picture, I wanna look now globally and the two main points I wanna make about the global thing is, and here I'm drawing on another survey that Nick and I field with a team of several people.

 

This is a survey, it's modeled after the US survey, but we field it about once a year and we're up to about 40 countries now. Two points I want you to take away from this one in other countries as well. The evidence suggests that the extent of work from home has kind of settled down to a reasonably stable level since 2023.

 

That's sort of what this picture is showing you. The second fact I want you to have in your head, which Nick and I and others have written about elsewhere and which we have another working paper in progress about. I'm not gonna try to explained today, but I just want you to appreciate this fact.

 

To put in context, what I'm telling you about is that work from home rates, the levels differ greatly across countries and they're basically highest in the Anglosphere and lowest in Asia with Europe and Latin America in between. Okay, again, I'll just put that back on the table, my job today is not to explain that back.

 

 

>> Speaker 5: Why is China so?

>> Steven J. Davis: I'm not gonna try to explain that today. Look, it's basic, there's several reasons, but let me just not go into it. There's a multi-fold explanation. I'm happy to talk with somebody offline, but there are multiple elements in the overall picture and I think I'll just defer that.

 

You can have me back here another next fall and I'll tell you all about that. Okay, so here's a picture. This is something you could have guessed, but it's another piece of background that's important to understand that in the cross section there's an important correlation between the extent to which you work from home and how far away you live from your employer.

 

So this picture is pretty flat until you get to people who are living more than 20 miles away from their employer. And then you see that the percentage of days worked from home rise significantly with distance from employer. Again, not a huge surprise, but the magnitudes here are quite important.

 

By the time you get to people who are living more than 250 miles away from their employer, according to our data, and this is from our survey data, again, something like 60% of their work days are being done from home. Not too surprising. 250 miles is a long ways to commute.

 

Okay, so now the main data source that we rely on for this paper on the new geography of the labor market is from a company called Gusto. Okay, Gusto is a payroll processing firm like ADP, but not nearly as big. They provide tax and payroll and other HR type services to mostly small and mid sized employers.

 

All right, so we have anonymized matched employer employee data so we can follow the two over time. They're linked to each other. We know, among other things, the zip code of the employee's residence and the zip code of the employer's work site. And that is going to be a key input into much of what we do.

 

And when I talk about distance, it's basically haber sign distance between those zip code centroids, that's what we're using here. We'll reweight these data to kind of match the CPS, but still they can be unrepresentative in some significant respects.

>> Speaker 5: Like a multiplant firm, the work site is where that worker goes or is it the corporate headquarters?

 

 

>> Steven J. Davis: Here it's basically the level of granularity that happens in the reporting to the state level employment security agencies. So for larger companies, they'll have different reporting unit numbers for different facilities. For smaller companies, if you're running three coffee shops in the San Francisco area, you're probably just gonna have one location for all of them.

 

So there is some noise there. I've left it out of my main deck here. But there is some cleaning of the data in which we basically, in an effort to be conservative, throw out firms that even in 2019 had a surprisingly large share of their employees living more than 50 miles away from the alleged work site.

 

Okay, so these data have both great strengths and some weaknesses. The great strength is we've got this matched employer employee data set. We can follow both longitudinally. We have pretty precise information about where the worker lives, where the employer's work site is, the weaknesses. We're kind of stuck with what's in the Gusto universe, okay?

 

 

>> Speaker 5: What do you do about airlines?

>> Steven J. Davis: Airlines? Well, each airline employee will have a base typically tied to one of the base employment sites. And so they'd be assigned to that most cases cuz that's where their paycheck would come from. But I don't think there are any airlines in our data set.

 

 

>> Speaker 5: They punted on that.

>> Steven J. Davis: No, we didn't. Remember what I told you? We cleaned the data and if in 2019, let's say you're an airline company and you discover that half of your employees live more than 50 miles away from the alleged work site, you're out of our data set.

 

We've taken you out.

>> Speaker 5: These are workers who don't work at their work site.

>> Steven J. Davis: That's the point. So I just told you, those would get screened out of our data set in the data cleaning stuff that we did, okay? So I'm not gonna make use of those data.

 

They're in the universe. But a company that looks like an airline company won't be in our analysis data set. Okay, that's what we did.

>> Speaker 4: They aren't gonna hire Gusto either, right?

>> Steven J. Davis: No, I don't think they're in Gusto. That's why my first response to Bob was I don't think we have any Gusto.

 

I don't think we have any airlines in our data set. But even if we did, they're not in the analysis data set that I'm gonna make use of. So, given the strengths and weaknesses of the data set, I'm gonna make use of three different subsets or versions of it, yeah?

 

 

>> Speaker 4: A little bit of a worry there cuz you're throwing out companies and looking at changes. You're thinking about sales. Any company in sales is likely to have that same airline. Like if I have a region, I don't necessarily have to live near that.

>> Steven J. Davis: Look, the data are imperfect, okay?

 

I'll stipulate that, nonetheless, we can learn a lot from them and five years from now we can go to the Social Security records so we can get access to them and we can do a very clean version of this. But, I don't wanna wait five years. And I spent a lot of time working with large scale government data sets in my life and I grow increasingly impatient with the increasingly large cost of using them.

 

So I may never do that. Maybe somebody else can do that. But for now I got these data and this is what I'm gonna use and they're informative. So for some of the exercises where I'm worried about potential composition shifts over time distorting my analysis, I'm gonna look at a balanced panel of firms, all of the employees in those firms.

 

But I'll look at a balanced panel of firms. In other analyses, all I really need to do is be able to follow employees from one year to the next. And so I'll sometimes look at all continuing employees, which means those who are employed by the same firm from one December to the next.

 

And then for some analyses, I'm not so worried about these kinds of selection or compositional shifts. And I'll just look at the whole data set. Okay, so this is looking at the balance panel. Okay, so it's a balanced panel of firms, not a balanced panel of employees, a balanced panel of firms and everyone who worked for those firms month by month.

 

And I'm just showing you over time, these are monthly data. The mean distance to the work site in the left hand chart and the means here are windsor rise to 250 miles. So that's kind of another response to Bob's question about airlines and the like. And you can see it's gone up from about 15 miles before the pandemic to about 26, 28 miles in 2024.

 

I think the more interesting metric, for a variety of reasons, in part because it's just a more robust measure, is the percentage of people who live far away. Far for present purposes, means 50 miles away from their employer's work site. That number has gone from about 4 to nearly 10% since the pandemic struck.

 

Now, if you take the right chart and you break it down a little bit, you get this. So the gray curve here is just what I showed you on the previous chart, showing you the fraction of people who live more than 50 miles from their employer. It's gone from about 4 to 10%.

 

But if you look at the bottom, you can see among people who were already employed by those firms, not much has happened. So this notion that people actually stay with their same employer but move far away, there are some cases like that, but it's not a big deal.

 

The real action is happening on the new hires margin, okay? And you can see that since 2023, 2024, about 12% of the new hires are of people who live more than 50 miles away. And obviously that process will take a long time to play out and take 20 years or more before the entire workforce turns over.

 

 

>> Speaker 6: In this sample for the period, you're looking at about what fraction of the employees were hired after March 2020?

>> Steven J. Davis: Good question. I should have calculated that, calculated that, but I haven't. Do you have a sense of that, Bert?

>> Speaker 7: He did, but I don't remember off the top of my head.

 

 

>> Steven J. Davis: You're right. It'd be a good number, and it'll be a sizable number. You can sort of infer something about it from what's happened to the all employees. So it's gotta be a large fraction, otherwise it wouldn't have gone up so much. All right, so now, this phenomenon has played out very differently across earnings groups and across industries.

 

The left chart here is looking at the earnings groups, and you can see that for kind of people in the lower and middle rungs of the earnings distribution. And not much has really changed in this respect in terms of the percentage of people who live more than 50 miles from their employer's work site.

 

But when you look at the affluent, and I'll just focus on the top bucket here, individuals who are earning more than $250,000 in a year, it's gone from something like 6% living more than 50 miles from their work site to 16%. Okay, so that's a huge increase. And it already tells you something about who is in.

 

Enjoying most of the benefits from the big shift to work from home. The right chart is the industry breakdown. And again, it's a very heterogeneous story. There's lots of mainline industries like retail trade, manufacturing, accommodation and food services and the like where things really haven't changed that much.

 

People's residential locations relative to their work sites, it doesn't really look all that different than it did before the pandemic. At the other extreme is the information sector. Okay, yeah.

>> Speaker 8: Question just on the last slide. This is surprising to me given the relocation story. How does this fact square?

 

I would have thought that it would be people who were already at the firm and now are able to work remotely that are relocating.

>> Steven J. Davis: So it's true that relocation is often tied to job change. But we can't follow those events in our data set because people, when they change jobs, change employers, I should say, they leave the gusto universe.

 

So, when I look at the effects, the tax and housing costs effects of people who moved, I'm going to restrict my attention to those who stayed with the same employer, which is probably causing me to greatly understate the phenomenon for the reasons you're suggesting. But that's, that's all I can do with these data.

 

 

>> John Cochrane: Cuz you mentioned a lot of out migration in your early slide. But this isn't screaming out migration.

>> Steven J. Davis: Well, they're important out migration patterns. These data set is not well suited for me to quantify the overall extent of out migration. What I can use it for is for people who continue to stay with the same employer where they're moving in terms of housing costs and tax rates.

 

And that's how I'll use it.

>> John Cochrane: So people who migrate out would maybe quit one of these jobs and then sign up with somebody.

>> Steven J. Davis: Exactly, and I can't follow those people.

>> Speaker 4: Tax things are really worrying because if I live in California, if there's a great job in California, somebody from another state takes their job remotely, they're paying California taxes.

 

 

>> Steven J. Davis: California is unusually aggressive in this respect. So, look, look-

>> Speaker 9: One issue depends where the firm has, you're correct with the firm of it smart, but set up in the longer run a subsidiary and say in Nevada.

>> Steven J. Davis: Yeah

>> Speaker 9: And pay you from Nevada. And then you're still working for Google, but your paycheck comes from Google Nevada and then therefore entirely avoidable.

 

 

>> Steven J. Davis: And we can follow. And we do follow those events. So we will see somebody if they go from Google California to Google Texas, we'll see them.

>> Speaker 10: Are you able to see Steve and Nick and, and track mobility among firms within the Gusto data set?

>> Steven J. Davis: Within what?

 

 

>> Speaker 10: The Gusto?

>> Steven J. Davis: Yeah, we can see mobility as long as you stay with the same firm. Well, we could in principle try to do that. The Gusto universe is not comprehensive enough that that's a very useful enterprise. If you had ADP data, you might wanna try that.

 

Okay, there's another question. Yea, John.

>> Speaker 11: I'm just curious how we're thinking about distance and remote jobs, headquarters. I'm also curious, the 50 mile threshold, how much of it's just moving maybe the suburbs? You could kind of figure this out with having the. How much of it being hired in the suburbs now of the same metro area of the same employer?

 

 

>> Steven J. Davis: Well, we chose 50 miles for a couple reasons. One, it encompasses the suburban rim in most metropolitan areas. But second, just kind of introspection. 50 miles seems like for most of us, the outer limit of the daily commute. Nothing more serious than that. We could have chosen 100, 250 or so on.

 

We've looked at that. All of the patterns are qualitatively the same. Of course the precise numbers will differ.

>> Speaker 11: This is mostly people being hired in different MSAs.

>> Steven J. Davis: Yeah, or at least different cities. It might not even be an MSA. They just might move from Chicago to downstate Illinois and commute into Chicago a couple times a month or something, yeah.

 

 

>> Speaker 12: I wanna ask a question about the next slide, the one that you. So is the graph on the left, is that true? Does it look as stark within firms? So that gradient of pay, is there that much of a difference within companies?

>> Steven J. Davis: Well, some of this is within companies, okay?

 

 

>> Speaker 12: Across end within? That's all the variation.

>> Steven J. Davis: Yes, this, this is all the variation in the balance panel of firms, okay? That's what I'm using here. And the reason I've used the balanced panel of firms here is the Gusto itself is a rapidly growing company over this period.

 

And I don't want to let changes in the mix of Gusto's customer base drive my outcomes.

>> Speaker 12: Okay, but if you had firm fixed effects, would it look that extreme?

>> Steven J. Davis: People, it's a great question. I don't fully understand.

>> Speaker 5: It could be the rich people work for Google and the poor people who work for Kentucky Fried Chicken.

 

 

>> Speaker 12: That's right, yeah.

>> Speaker 5: Or it could be that within Google. The rich people live-

>> Speaker 12: Are starting to work remotely and that.

>> Speaker 5: Yes.

>> Steven J. Davis: Yeah, we can look at that.

>> John Cochrane: What kind of tasks can this be done? If you work at an auto body shop and you're spraying cars, you can't do that from home.

 

Doing this, you do that-

>> Steven J. Davis: And the rightmost chart tells exactly that story. And there's lots of other evidence.

>> Speaker 12: But it could be related to seniority or something in addition to tasks.

>> Steven J. Davis: We've looked at that elsewhere and the Gusto data aren't maybe the best suited for that.

 

But there is some of what you described. It is the more senior folks who are often the ones taking advantage of the opportunity to work remotely.

>> Speaker 5: Do you have any information on their jobs are within the firm? Because-.

>> Steven J. Davis: Not much, not enough to make use of.

 

We basically know sex, age, earnings, industry. I think that's basically-

>> Speaker 14: Aggregate, There are more information technology workers outside of tech firms, inside of tech firms.

>> Steven J. Davis: That's right, but we don't have occupation. So for sure. So, okay, then let me push ahead and just show you. So now we're gonna look at continuing employees.

 

I no longer have the balance firm panel requirement. I'm now looking at all employees who were with the same firm from December of one year to December of the indicated year on the horizontal scale, okay? So larger sample. And we've divided it up into earnings bins. What they're listed there, you can see.

 

And here I'm showing you the change in the state's top tax rate as you move from state A to state B. So to go back to my earlier example, if you move from California to Texas, then I would code you with a 12 percentage point change. Whatever it is, if you don't move, you code it with a 0.

 

Say it again.

>> Michael Boskin: 13, 3.

>> Steven J. Davis: 13, 3, okay. Maybe. Are you on this margin, Michael?

>> Michael Boskin: Unfortunately, yes.

>> Steven J. Davis: Okay.

>> Michael Boskin: Unfortunately you probably are too.

>> Steven J. Davis: Okay, I did tell my wife about this, about the picture coming up last night and say, wow, you know, it really does make you wonder why you're living here.

 

But let me go ahead. So what you see here is a couple things. There wasn't any discernible pattern before the pandemic struck. A little bit of something going on in 2019. The pandemic hits and then these numbers are in percentage points on the vertical scale. So it's 15 basis points is the integrating over everybody in the indicated earnings 250k earnings buck in 2020.

 

So that's kind of a hit to net aggregate state level income tax collections subject to all the provisos, you know the data sets not perfectly representative. I'm ignoring the distinction between average and marginal tax rates here. Not everybody pays the top marginal tax rate in the state even if they're in the 250k and so on.

 

So okay, that, that said, you can see something's clearly going on here. Okay, in a big way. And now you can make all your complaints. I've already kind of highlighted some of this. This is my very crude effort to get a handle on what this might mean for state level tax collections.

 

I don't know if it's Josh Rao here. I'll probably better he's. So this is very crude. But it tells you, it gives you a sense that this is a non trivial issue. So if I take these, I'm gonna focus on the clicker here, the 250k category. I'm gonna take these reductions and I'm just gonna accumulate.

 

If I accumulate them over this four year period, that adds up to 48 basis points of net cumulative tax rate reduction from moves in this most affluent group over this four year period from 20 to 2023. Persons in that group are about 40% of aggregate US labor income as of 2022.

 

So it gives me a number to multiply the 48 basis point reduction. That's how I get the $25 billion per year of lost. You wanna call it lost state level income tax collections for this particular earnings group. Yeah.

>> Speaker 4: Just to understand, it's a select example. So those people obviously had the flexibility at the same company to move which is a very small fraction of people in this group.

 

So you can't multiply by that number to all the people in the world?

>> Steven J. Davis: No, yes I can.

>> Speaker 4: And that's a gross over exaggeration of how big the effect is.

>> Steven J. Davis: No, no, it's not an exaggeration for the reasons you just said. This is an unusual group.

 

Most people achieve this flexibility and a much larger number of people achieve this flexibility by changing employers, okay? And I've missed that. So on this score at least this is probably an understatement, okay. So look, I'm just trying to put out a number. It's the best I have now, I know of no better number on the subject.

 

Some number is better than no number. So that's where we are on this.

>> John Taylor: What you might wanna do is take Josh has numbers for California, for example. You might want to apply your methodology and see how this squares. Also I think State Legislative Analyst Office, Department of Finance.

 

 

>> Steven J. Davis: Yeah,

>> John Taylor: New York and California are the big losers here.

>> Steven J. Davis: They are.

>> John Taylor: And they have the most progressive highest income tax rates. New York City even higher when you add in the state income tax.

>> Steven J. Davis: Yeah, there's more we can do here and in particular we can try to.

 

The obvious thing to me it seems is to take something like the NVR tax system tax sim and try to impute an average state level tax rate for these individuals based on the limited information we have about them. We haven't done that yet. All right.

>> Speaker 5: One other number I can give you actually.

 

I spoke to the controller's office in Washington DC turns out they're also here, they're actually even worse because they're so small and they reckon they were losing a couple of billion dollars a year. So DC that was their estimates. I don't know how they did. It was nothing to do with our statistic to their tax returns from state tax.

 

 

>> Steven J. Davis: What fraction of their revenue was that?

>> Speaker 5: That was big. It was so large that. They were having to change business taxes and retail taxes. So if they're 2 million. 2 billion and you can figure out that.

>> Steven J. Davis: Yeah, so that suggests our 25 billion number is too small.

 

Okay, so let me turn to housing for which these issues are less pressing because it's much easier to move within the state across zip codes. Stay with her same employer. So going back to John's question earlier, this would be somebody who just moves to the suburbs here as well.

 

You see, there's something big happening that wasn't happening before the pandemic. If I again focus on the top earnings group that's covered in this chart, here it is. The average reduction averaged over people who moved in, those who didn't move is already 1.5 percentage point reduction in housing costs.

 

These numbers are still big but declining over time. And it looks like this process hasn't fully played its way out. If you look at the relationship between-.

>> Speaker 17: They could just go into a bigger house at the same price.

>> Steven J. Davis: They could, it's just what we've done here is we've taken the average zip code level Zillow price from 2017 to 2023.

 

So we've conditioned out price changes that were driven by the pandemic itself. You're quite right that you can get more house per dollar if you move out of the most expensive markets. And we haven't taken that into account. Okay, so here's the relationship between medium home value in the employer's zip code and the percentage of employees that live more than 50 miles away in the cross section.

 

So this is a different kind of exercise. Doesn't immediately give you a bottom line number, but it kind of sidesteps some of these measurement issues that you might otherwise run into. And you can see that there's a very powerful relationship. People who work for companies in housing markets with high housing costs are living elsewhere.

 

The ability to do that is much greater now for many people than it was before the pandemic. That's the central point. So, if I turn now to those who actually move, get a sense of the numbers there. These are as we discussed. This is a selected group. Not everybody has this degree of flexibility cuz they don't necessarily work for a company that operates across multiple state lines or will let them operate across multiple lines if you're a remote worker.

 

But these numbers are huge. That's the one I wanna leave you with. So again, in 2020, the people in the top earnings group here who switch states, the average across all the switches in the data set is a 5.2 percentage point drop in their top marginal tax rate.

 

 

>> Michael Boskin: You see their pay after they move?

>> Steven J. Davis: We see their pay before and after, yeah.

>> Michael Boskin: Was there any change?

>> Steven J. Davis: We didn't look at that. We could see whether they work hard.

>> Michael Boskin: Well, you have a lot of small and mid sized companies, a lot of large companies will base their pay on where you're living.

 

 

>> Steven J. Davis: Some do, but so we've asked about this. A smaller share than I would have guessed and in the survey evidence I've seen. But you're right, some do. So, and these are the housing cost changes. Again, these are enormous. So what this is saying is among the set of people who've chosen to relocate zip codes in the wake of the pandemic, and you can see that's gone up a lot.

 

They're enjoying very large reductions in their housing costs and probably increases in the value, the actual utility value of the home too, that we're not capturing in this calculation. So these are huge effects. All right, I wanna switch to the last.

>> John Taylor: Move out of a tiny apartment in New York.

 

You can actually have a home office.

>> Steven J. Davis: You can have a home office. You can have a backyard.

>> John Taylor: I'm just saying the ability to work from home in a tiny apartment in New York isn't so easy.

>> Steven J. Davis: Right. Yes.

>> Speaker 18: Question about the previous two graphs. There doesn't look like there's a big structural break in 2020 in either of these, especially the previous one.

 

People were, if they had the chance to move, were also still moving. Like that one doesn't look like.

>> Steven J. Davis: Well, there's some. You can see the data for 2017 are quite noisy. And that's because I said Gusto has been growing a lot. You just have a lot less data for 2017.

 

So it's hard to read anything, anything clear out of 2017. There's something going on in 2019 before the pandemic struck. But there's a pretty clear break, it seems to me, in the overall pattern here. Despite the simplicity of the calculation and the same thing here, there's clearly something different happening in 2020 than it happened before.

 

Okay, but you're right in the sense that there were some indications that these pressures, these out migration pressures were present before the pandemic. Yeah, just that there were fewer people who have the opportunity to act upon them before work from home. Yeah, Nick

>> Speaker 19: Is the tax stuff driven a lot by Florida and Texas?

 

I mean, two states without income taxes.

>> Steven J. Davis: Yeah, we looked at that. We didn't go very far down that path that we start pushing the data too hard. We've thought about doing a whole bunch of bilateral comparisons. The data don't seem quite. You can see already the standard errors on some of these numbers are pretty large.

 

So we haven't really pushed. If we can get access to the ADP data, we'll make the entire map of what this looks like, but that hasn't happened yet. Okay, last topic. So gonna shift gears here and think about a different issue. And let me describe to you how this picture is constructed.

 

Now we're gonna use the full data set. There's not really a reason to be so concerned about composition or selection bias for this exercise. So for every employer in the data set, every firm, we can construct its monthly growth rate. Okay, and we can also compute its higher, its gross hiring rate and its gross separation rate.

 

This is just a non parametric portrayal of what that looks like. Similar to things I've done with Faberman and alterwinger and other settings. What I'm trying to show you here is as. Remember, whether we've got employer fixed effects in this regression or not. Maybe it says in the notes.

 

No, it doesn't look like we. Okay, we don't. Thanks, Mark, so what we're looking at here is the thought experiment is, think about it. As an employer expands more and more rapidly, to what extent does it adjust its labor input by gross hires of near employees? That's the solid red lines there versus gross hires of far employees.

 

That's the lightly colored red line. Same thing on the separation side. As the employer is contracting, to what extent is it contracting? By getting rid of its employees who live nearby versus employees who live far away. So one thing you see immediately. And this is something that I've documented extensively in previous work.

 

These relationships are extremely nonlinear, okay? But for the present purposes, the important thing to note here is that the lightly colored red line and the lightly colored blue line are always above and somewhat more steeply sloped, at least on the right. The right side for the red and the left side for the blue than the dark line.

 

What does that mean? That means on the margin, as firms shrink, for example, they're more likely to separate from their distant employees than from their near employees. And as they expand more and more rapidly, they're more and more likely to draw on distant new hires than nearby new hires.

 

I don't think either of those things are surprising. There's other reasons to expect that. The first evidence I know of the document. And the way I would summarize it in just very basic economic terms is firms treat distant employees as more flexible labor inputs on both the recruitment margin and the separation margin.

 

 

>> Speaker 4: Ignoring the employees themselves.

>> Steven J. Davis: Thank you. This is conditioning out an exhaustive set of job tenure effects, age effects, sex effects, because I had exactly the same issue. And this is on the separation margin here. And you can see that the same pattern emerges.

>> Speaker 4: No, but I'm worried that employees have more opportunities.

 

If you're in remote, you have both.

>> Steven J. Davis: Not ignoring that. I'm telling you a fact about how employers adjust and you're telling me there's other concepts.

>> Speaker 4: Lawyer treats them with more flexibility. Could be the employees are more flexible.

>> Steven J. Davis: Okay, I'm assuming other knowledge, not in evidence that I've established elsewhere.

 

So let's. Let's talk about the left side of this. Not surprisingly, as employers shrink more and more rapidly, they turn from quits to separations. So, as we move out this, I mean you're right, that was an implicit assumption in my head. I should have stated it. And interpretation I gave I think holds up.

 

 

>> Valerie Ramey: Steve.

>> Steven J. Davis: Yeah.

>> Valerie Ramey: You know how seniority always led to last in, first out. What you have is a new fact which is far out first out.

>> Steven J. Davis: Okay, that's great, Valerie. I think we're gonna borrow that, far out first. That's exactly right. And I did I think this back to this composition point when my first reaction to this chart was the left hand side is all about we already know.

 

Remember what I told you already. There's been a huge expansion in of work from home and hiring of more distant employees. So we know in our data set the far employees have less tenure, extensive body of work that Valerie just alluded to that higher seniority people are less likely to separate.

 

So this composition issue is a real one. And what this chart says is even after we control for that in an exhaustive manner, you get the far out first out that Valerie described, okay, so

>> Speaker 10: One more question too.

>> Steven J. Davis: Yeah.

>> Speaker 10: So the idea that everything has to be priced in so you have greater flex.

 

We were thinking about utility welfare workers wise, you have greater flexibility. But anecdotally friends at McKinsey tell me, well it comes with a cash in terms of the support, admin support. It's a non pecuniary aspect of the job. Benefits for people who are not exactly the highest, highest under the pay range or even agreement that you have to do your task composition changes.

 

You kind of have to do a little bit more than you use, manage your position more of admin or supervisory. Which remotely do you have a sense. Can you tell from the data?

>> Steven J. Davis: Not from these data, but we've looked at some of these

>> Speaker 10: Degradation. A dimensional quality degradation.

 

 

>> Steven J. Davis: Yeah, look, we've looked at this in survey questions in various ways and the bottom line which. Comes through at least multiple surveys is people do spend a little more time working for their employer when they work remotely, but it's only about 40% of the commute time savings they enjoy.

 

So that's the bottom line. We've seen this in surveys for the US, we've seen it in other countries. That's not very far off from a Nash bargain, right? All of a sudden, this new technology of working has opened up. There's extra surplus on the table. And at least in terms of just the time dimension, obviously the compensation matters too.

 

But at least in terms of the time dimension, the employees give some of it back to the employer and they keep a little more than half for themselves, okay?

>> Speaker 10: The pay wise, pay structure wise, you don't see a change.

>> Steven J. Davis: Why is it what?

>> Speaker 10: In terms of the pay structure within the firm, when firms-

 

 

>> Steven J. Davis: The pay structure question is still very much an open issue. There's work on this. The work that Nick and I and others have done basically says the following. But this is my own view, and this is what our work says. I don't wanna present this as the consensus view cuz there is no consensus on this question yet.

 

But our evidence suggests that when the pandemic initially struck, a bunch of people found themselves enjoying the benefits of working from home at their old wage, and they essentially got 100% of the benefits over time. Employers had clawed some of that back essentially by offering slower, slower wage growth than otherwise, which is my preferred explanation for part of how it is that we unwound that inflation surge with so little difficulties in the labor market.

 

There was a source of slack there. Real wages did grow very slowly in the kinds of sectors that, that enjoyed the biggest, biggest increases in remote work, information, finance, business and insurance. Those are sectors that before the pandemic had seen consistently higher real wage growth than most other sectors that was interrupted for a period of time.

 

I think it's this clawback phenomenon, but I think there's not a consensus view on that yet. All right, so let me take stock of a couple things that we've talked about now and draw up this last implication. This picture here, but also the evidence I showed you earlier that the spatial footprint of the average employer has become more dispersed.

 

Both those things say that in a future downturn, when an employer lays off a bunch of workers, for example, the spatial impact of that will be more spread out than we've seen in previous recessions, downturns, labor market restructurings. So think about what that means that means, you don't have quite as much concentration of people with the similar skill sets losing jobs at the same place in the same time.

 

I think this is quite an important development.

>> John Taylor: Absolutely.

>> Steven J. Davis: And it's one of the reasons why one optimistic is not the only reason, but one of the reasons why I'm somewhat optimistic that the economy will absorb coming AI driven layoffs that some people worry about much more.

 

I don't wanna say it's gonna be easy process. It won't be nearly as hard as the deindustrialization process was because that was the opposite of this. Manufacturing jobs highly concentrated in certain locations, highly cyclically sensitive. So you've got a huge number of people with similar skill sets being laid off at the same time in the same places.

 

That's less likely to be the pattern going forward, John.

>> John Cochrane: But it depends, cuz it's also likely that the tasks people are doing farther away, are different than the tasks. If you're running an auto body shop, the guy who paints the car is there, people who do the accounting are 50 miles away.

 

So when the recession comes, are they gonna be preferentially laying off people in some kinds of tasks versus other kinds of tasks?

>> Steven J. Davis: Yeah, I take that point. But with respect to AI, the kinds of tasks that seem to be most at risk have a lot of overlap overlay with jobs that are remotable.

 

 

>> John Cochrane: You mentioned two things which are different. One is a recession when you have concentrated plans. The other is a technological change which is a longer term thing. Yes, the factory town is not gonna be in trouble like it was before.

>> Steven J. Davis: Exactly.

>> John Cochrane: Those are longer term, slower moving things.

 

 

>> Steven J. Davis: Right, so I mean the metaphor I have in mind or the example I have in mind is think about the, the loss of bank tellers on ATM machines. That was a gradual thing over time and it was spatially.

>> John Cochrane: But those banks increased, employment banks increased.

>> Steven J. Davis: Well, that's yet a different response.

 

Yeah, but the bank tellers got reallocated to doing different functions.

>> John Cochrane: Since banks were cheaper, they had more branches.

>> Steven J. Davis: I'm just another example. Telephone operators. Okay, they were kind of spatially diffused. It happened over time. So I think those are better metaphors for future job loss than steel mill shutdown secretaries, yeah.

 

 

>> Speaker 5: For the word process.

>> Steven J. Davis: All right, so I guess, yeah, I got time for this case study. So I'm gonna switch gears and talk about a different page paper briefly, but this speaks to one aspect of what's going on. So, we are cooperating with a firm called Tempo.

 

I think it's the largest customer service company in Turkey. We're doing a couple things with them. But one thing that I'll talk about here briefly, they gave us access to all of their data on their call center employees back to, I think, 2017. Is that right, Nick? And they have extremely detailed data about what these people are doing and what kind of calls they're taking.

 

They got, obviously, it's easy to track productivity for these people in various ways. They have pretty good customer service, quality measures and so on. What makes this company interesting in the context of the current talk is they were entirely on site work for the call center employees before the pandemic hit.

 

They had about 3,500 employees in call center roles. So they have a broad clientele, banks, mobile phone operators, food chains and so on. So that they're servicing other corporate clients for the most part. So there was a national lockdown in Turkey on March 11, 2020, precipitated by the pandemic.

 

Within two weeks, the company had to go from fully on site to fully remote because the law required them to do so. And so they had to scramble. They've had a sufficiently positive experience with this that after the lockdown ended. I think lockdown ended October or something. Is that right?

 

In 2020, I remember. But sometime later it's gonna come up on one of the charts I'm gonna show you. After the lockdown ended, they decided to stick with fully remote, okay? So I want to just briefly show you some of what happened, in particular, what happened to the spatial distribution of jobs.

 

So just so you know what this means, because I want you to understand what a Turkish office environment looks like. This is maybe different than you might have imagined, a US high rise. So on the left there, that's what the call center operations look like before the pandemic struck.

 

And the key thing I want you to take away, this is a very crowded, high distraction work environment. And then this is the work environment afterwards. People working at home. Not luxurious quarters by any means, but much quieter.

>> Speaker 5: Picture below.

>> Steven J. Davis: Picture to the right. Picture to the right is showing you four panels of people who work from home after the pandemic for the same company.

 

And the pictures on the left are what their offices look like before the pandemic. So they went from working in the office to working at home in this company. Here's the main findings that come out of this study. Remember, we've got fabulous data on these people. The shift to remote work led to a very dramatic change in the composition of their workforce in multiple respects.

 

More women, more older, more educated workers. And to the point of this talk, big increase in the share of persons who reside outside metropolitan areas. People who didn't want to commute into Istanbul, for example, but were perfectly happy to work from home. Workforce productivity rose a lot, 6%.

 

I showed you the previous picture, so you get some idea of why. And there's no apparent loss of service quality. Basically they're just generating more calls per hour.

>> John Taylor: Was there a sense in which there was any productivity increase trend procurement beforehand?

>> Steven J. Davis: No, there wasn't much. I don't think I have that.

 

Well, here, here we go. No, that's not the productivity picture, yeah.

>> Valerie Ramey: So it seems kind of idealized those right pictures cuz I know, say, staff at UCSD, some were zooming in from garden sheds and then they'd have bad Wi-Fi and so maybe the elderly relatives are there.

 

I mean, are you sure that that's.

>> Steven J. Davis: The company's not gonna hire. This is a high turnover company.

>> Valerie Ramey: Okay.

>> Steven J. Davis: So even by the time it gets 2022, 2023, most of these people were recruited and hired after the pandemic. They're not even gonna hire them if they don't have energy.

 

 

>> Valerie Ramey: So this wasn't a high frequency after picture, right?

>> Steven J. Davis: This is showing you monthly data. You can see that. I'll come to you just a sec, Pat, you can see that on the right, that's the picture I most wanna stress. So, these are our proxy for, we know where these people are by province and there's more and less populous provinces in Turkey.

 

There's a lot of people who live, you know, of course, in the rural and more conservative areas in Turkey. And you can see there's a big increase in the share of the company of those people. So these are just new hires. They weren't even working for the company before.

 

It goes from something like 5% before the pandemic or 4% to more like 20% two years later, okay. There's also been a rise in female employment, including married women, which is a big issue in Turkey, Pat.

>> Speaker 22: I was thinking of your tax calculations earlier. This makes it look in the stories you're telling, a big rise in participation by some Groups.

 

So if there's a big rise in participation, even though you're shifting away from the high tax areas, you're picking up all these extra people in the workforce. Well, total tax revenues, cuz new participants should go up. And is it hard to take account of that offsetting effect of the loss in tax revenues?

 

 

>> Steven J. Davis: That's a good point. And I think directionally I'm 100% on board with you. And there is evidence, Nick has a paper on this and there's. There is evidence that the shift to work from home is drawing some formerly somewhat marginalized, disenfranchised group into the workforce. That seems to be happening in Turkey, but even in the United States it's happening and we, we could try to quantify that.

 

We'd have to turn away from the gusto data set because gusto data set's not suitable for that. I'm thinking that is definitely a phenomenon.

>> Speaker 19: Small towns where they're sort of pass through towns, things have gone bad. But now if they have to work from home.

>> Steven J. Davis: This is another one of the underappreciated benefits, both economic benefits, but also social benefits of this shift.

 

Not necessarily tax benefits. Call centers writ large in some sense, yeah.

>> Speaker 5: I think it's as much like a macro. It's almost like monetary policy in the sense that you think y equals atimes l, forget every other thing. A is probably unchanged, but L is potentially up a lot like 2, 3%.

 

And if you spread that out over, and in five, ten years, even in a coefficient of two-thirds, that's material, put them in. Yeah, yeah, and it also major impact on them.

>> Steven J. Davis: They're still paying taxes though $50,000 don't pay state income tax.

>> Speaker 5: The margins, many of the margins are people close to retirement, people with disabilities.

 

 

>> Steven J. Davis: That's telling.

>> Speaker 5: But I think the l side.

>> Steven J. Davis: Let's let Bob Hall get in here.

>> Robert Hall: Have you looked at the history of JetBlue? If not, I'll tell you why you should.

>> Steven J. Davis: Okay, well tell me why I should.

>> Robert Hall: When they went public or when they were actually just created, they made a big deal out of the fact that all of their service workers gonna be working at home.

 

And it would all be 100% from time zero. Gradually they abandoned that and today they're about to be acquired by United. And there's no vestige in that chat room as far as I know.

>> Steven J. Davis: Yeah, well, there are certainly firms that run counter to the main trends that I've documented.

 

I showed you all the evidence at the outset is the aggregate Evidence is pretty clear that not much has changed since 2023 in the extent of work from home. Doesn't mean some companies aren't going the other direction some are.

>> Speaker 5: That was true in 2010. I actually visited JetBlue out in Salt Lake City.

 

The amazing thing was I visited multiple work from home folks that work in the call centers. Remember visiting one, there's a whole group of us and they said, what do you do? They're going around, and I said I'm an economics professor. And they said, this is someone that works in the call center.

 

The woman went micro or macro? And I was like, wow, how educated is that. You wouldn't get and said, well, I did in college. Basically they were getting college educated, particularly women that would never work in call centers unless they. Yeah, it's interesting, I didn't know that. Track back.

 

 

>> Steven J. Davis: But Nick was working on work from home before, it was cool. But no, somebody said this earlier, that part of what's happened in Turkey is they are getting. I actually didn't put it in the charts. I think they are getting better educated workforce. So to some extent it's these well-educated married women living in conservative parts of Turkey who aren't gonna commute in the big cities and are even perhaps somewhat reluctant to leave the home or an untapped workforce.

 

Okay, so there's the share of women. Turkey, as you probably know, has difficulty drawing women into the workforce, but not this company. Here's the productivity pattern somebody was asking me about how could this be so good? And you can see that there was kind of a transition period.

 

This is just calls. This measure here is calls per hour. We've got agent fixed effects in here in this picture we've got controls for the mix of calls, we know what kind of call they're taking and so on. So these are non trivial productivity gains. And we think it's mainly because you've taken these people out of this very high distraction office environment and put them into a quieter environment.

 

And we see some indicators of that, like just shorter break times and so on. Okay, so let me wrap up. There is an underlying theme that, that animates much of what I've said and as well as some of the other comments that came up earlier, including Patrick's remarks.

 

Work from home relaxes locational constraints for workers, for families, for employers. That's really a huge deal. Most of what I said flows from in one way or another. And it's why it's kind of the fundamental reason why I'm so bullish on the whole shift to work from home.

 

Not that it doesn't come with some downside in adjustment cost. I've listed some of the ways in which this locational flexibility plays out. There's a number of other consequences of this increased locational flexibility that we have not talked about today, that Nick and I have written about in other settings.

 

The labor supply issue that Patrick alluded to I think is a big deal. We've only scratched the surface of that into in today's discussion, the wage, the wage determination issues I think are largely unnsettled terrain in the research sphere. We haven't talked about what it means for business formation, business dynamics.

 

There's another literature that suggests that this is a big source of the boom in business formation which is now well documented in the United States beyond the initial relocation effects of the pandemic. So major, this is a major phenomenon. I tried to tell you, give you a little bit of evidence on some aspects of it today.

 

 

>> Michael Boskin: I have two quick points I'd like to make.

>> Steven J. Davis: Yes, go ahead.

>> Michael Boskin: One additional thing that you could look at people are moving across states is not just that they're getting lower housing prices and lower tax rates, but they're getting lower prices for everything else now that may eventually get longer and get equilibrated with slower growth in wages or something.

 

 

>> Steven J. Davis: Right.

>> Michael Boskin: Sort of, so I think that's one thing. The other thing is I think this flexibility point, maybe they're implying this is even broader because it's greatly expanded the optionality for lots of people in present and future and what they do, firms and employees.

>> Steven J. Davis: Yeah, we have one paper that actually tries to quantify some of these optionality benefits.

 

It's more on the combination of the expansion of broadband service and the ability to remote work and that does have important optionality effects at both the micro level and the macro level.

>> Michael Boskin: Yeah, absolutely.

>> Steven J. Davis: John.

>> Speaker 11: I kind of have a macro question for you because I think the FOCO results are far out, first out is really important.

 

So we still haven't had a post work from home recession which is why I keep predicting.

>> Steven J. Davis: Some people are working on that.

>> Speaker 11: Crisis yet but I'm curious given that there is a swedge and you have some pretty good data and pretty good numbers on it, what would be the implied like say we had a run of the mill 7.5% unemployment recession.

 

How would that given that wedge that you have, affect your headline takeaway graph on work from home? Or alternatively, how high would the unemployment rate have to go to kill all of work from home for most of it?

>> Steven J. Davis: That second part, I think the premise is wrong.

 

High unemployment is not gonna kill work from home because firms are enjoying lots of benefits from work, from home as well, especially once they figure out how to adapt to it. So, I actually do have a little chart here since you asked about it, we actually did a survey using the, recently using the Atlanta Fed Survey of Business Uncertainty.

 

We asked all these employers, do you have any return to mandate, return to office mandates in the works now or are you planning any over the next year? If so, what do they actually entail? Are you gonna bring back everybody and so on, one day a week, five days a week and so on.

 

So this is just the bottom line. So we wrote this in reaction to all these newspaper articles, which every few months there's another wave of return to office mandates and work from home is over and so on. And we've been hearing that for about three years now. So we've decided to put some systematic evidence on the table.

 

And here it is. Even if all these announced RTO mandates and planned RTO mandates come to pass, it doesn't amount. The exact thing is it will reduce the overall work from home rate by 0.4 percentage points according to our calculation. So, I'm sorry, John, I'm just rejecting the premise of the second part of your question.

 

 

>> Speaker 11: That was a bit of facetious way of putting, I guess the point is the FOFO evidence would suggest that whenever that next recession comes, there is going to be some difference in work from home to some degree. Maybe it's not big, but you could put numbers to that.

 

 

>> Steven J. Davis: Well, we know, that's, I didn't put it here. We've also asked employers, what would you do in the wake of a recession with work from home? The answer is not much. And that's because many employers view it as a valuable attribute of their working arrangements. It saves on space costs.

 

 

>> Speaker 11: Just using your numbers, you said they're more flexible. Aren't you saying that the work will shrink in a recession?

>> Steven J. Davis: No, you're confusing, you're now going the opposite way. We were going the opposite way.

>> Speaker 11: I'm just saying that it's your credit.

>> Steven J. Davis: My point was how the downturn would happen okay at individual employers, but their equilibrium effects are going to play out as well.

 

 

>> John Cochrane: Concentration, take the town of Truckee.

>> John Cochrane: That's full of tech workers who are working from home. When they all get fired in the next recession, they are specialized in work from home. There cannot be ski lift operators.

>> Steven J. Davis: But you can work from home for any company that wants remote employees so you don't have to leave Truckee.

 

 

>> John Cochrane: But in the sense that there's a 10% of jobs that work from home and 90% are on site. You can only work those. You have to have to find another remote work from home job.

>> Steven J. Davis: So unless you, yeah. Unless you wanna relocate like everybody else had to in the past.

 

Okay, let's.

>> Speaker 10: Sorry.

>> Speaker 22: So I do think though, this is relevant to that interpretation of the separations and additions and whether it's employer driven or employee driven. Because I think the interpretation of employee driven is, look, these are employees that now have more flexibility. Their barriers to exiting jobs are lower.

 

So they're gonna bail out of the firms that are going down and then they're going to be attracted to the top firms. And that would be very consistent with the way that you talk, that you said that employers are talking about work from home and attracting the most talented.

 

So that's a different emphasis to interpret that data.

>> Steven J. Davis: And not surprisingly, better employees tend to leave sinking chips. Okay, and better in this context would be I got a wider opportunity set which other things equal. If I can work for companies anywhere, if I have the kind of skill set that lets me live anywhere in the United States I want and work for a company anywhere in the United States, then I'm not going to stick.

 

 

>> Speaker 22: Around with the same interpretation of what happens in recession.

>> Speaker 10: I have one question, the three of us have talked about it before, but I wonder whether you see that in the gastro data about the traditionally marginalized group, individuals who are statistically discriminated against or older workers or disabled workers for whom the flexibility is almost a precondition for participation.

 

 

>> Steven J. Davis: We don't see it in Augusto data set but.

>> Speaker 10: Do you see that traditionally-

>> Steven J. Davis: There's been a huge shift?

>> Speaker 24: I mean, this has a big fiscal effect. We see for disability, unless you're for disability, you see about half a million more. We have a paper about to come out, the AR Insights, you estimate about half a million more Americans with a disability are working because of work from home.

 

So that's like a causal statement. In fact, if you look at the employment of Americans with disability, it's gone up steeply. Some of that is more people have registered with disability, that's cognitively, you take them out, you look at physical disability, the numbers are flat, that the labor force participation was flat and then starts going up steeply.

 

 

>> Speaker 10: What about minorities? Because a lot of the discrimination that we have evidence for, I mean, you wonder if the contact is not as immediate.

>> Speaker 11: The disability, I think, is really important because also stop versus flow, if you look at most Americans with disability, they're post 50, 55.

 

And what tends to happen in the data is they just don't leave the workforce. If they're able to work from home two, three days, you get to 60, 70 of a package or 62, you would have retired, but instead you stayed. So the fiscal benefit. Yeah, we have much bigger fiscal problems.

 

 

>> Steven J. Davis: I'm aware. But yeah, this is one tiny positive.

>> Speaker 10: It's 10% of the.

>> Steven J. Davis: Yeah, Mallory.

>> Valerie Ramey: So I was thinking about this in the broader historical context and thinking about Joel Mulchier's wonderful discussions about going from putting out the cottage industry, you know, to the factory and stuff.

 

And it struck me that several things you need for this work from home. First of all, diffuse capital. And I love your case study because, right. They could send the little laptops home, whereas the factories had to have all of the capital in one place. And then being able to monitor the ouput.

 

So you had the piecework before during the putting out era, centuries ago. And then peer learning can't be too important. So the conversation we're having right now would just be too hard to have on Zoom. And so you can kind of look at all of those characteristics.

>> Steven J. Davis: There's some evidence in our TEMPO study, they put it here, that suggests the people actually started work for a few weeks on site before they went to promote work.

 

Actually, I don't know whether it's productivity effects in the narrow sense or it's just they don't have the kind of drop off in productivity six months later that we see very commonly. So we're not arguing there's no role.

>> Speaker 22: Have the drop off that you see for the people who didn't finish that sentence.

 

 

>> Steven J. Davis: Yeah, sorry. Thank you, Pat. So what we see is that if anything, productivity in the first six months on the job rises at least as steeply for the remote people as the other people. But the remote people, the ones who are working remotely, kind of have this big fall off after that in both attrition and productivity that we see much less of for the people who are working on site beforehand.

 

It almost suggests they're kind of getting detached.

>> Speaker 10: They detach.

>> Steven J. Davis: Okay, so the point I'm trying to make here, the substantive point, is the benefits of on site interaction is not just about learning, productivity, but also your attachment to the organization and the job. That seems to be important as well, aren't it?

 

 

>> Speaker 22: Aren't there a lot of work at home two or three days a week where you get a look.

>> Speaker 10: Yes.

>> Speaker 22: If you did it well, you get the best of both worlds. Like, you come in for a group meeting. You come in for that. And otherwise hang out at home two days.

 

Not for all those, but, yeah, that's what I mean. So that would be a subset of the people who are 100%.

>> John Taylor: Uber on a Friday afternoon.

>> Steven J. Davis: I do like Joel Mulcure's take on this, which I've heard him talk about, which is work from home is liberating us from the tyranny of the factory floor and the office.