Steven Davis (Hoover Institution) on Job Flows, Policy Uncertainty and Work From Home

Jon Hartley: This is the Capitalism and Freedom in the 21st Century podcast, an official podcast of the Hoover Institution Economic Policy Working Group, where we talk about economics, markets, and public policy. I'm Jon Hartley, your host. Today, my guest is Steve Davis, who is the Thomas W and Susan B Ford Senior Fellow and Director of Research at the Hoover Institution and senior fellow at the Stanford Institute for Economic Policy Research, SIEPR.

He was previously on the faculty of the University of Chicago Booth School of Business for more than 35 years, including serving as deputy dean of the faculty. Steve is also a research associate at the National Bureau of Economic Research, a visiting scholar at the Federal Reserve Bank of Atlanta, a senior advisor to the Brookings Papers on Economic Activity, and an advisor to the Monetary Authority of Singapore.

He's been elected a fellow of the Society of Labor Economists. He's an IZA research fellow and a senior academic fellow of the Asian Bureau of Finance and Economic Research. He hosts Economics, Applied, a video podcast series sponsored by the Hoover Institution. Steve is also the co-creator of the Economic Policy Uncertainty Indices, the Survey of Business Uncertainty, and the US Survey of Working Arrangements and Attitudes, which tracks Work From Home, The Global Survey of Working Arrangements, and the Work From Home Map project, as well as the Stock Market Jumps project. He also co-founded and co-organized the Asian Monetary Policy Forum, held annually in Singapore. Thanks so much for joining us today, Steve.

Steven J Davis: Well, thanks for having me, Jon. It's a pleasure to be here.

Jon Hartley: I want to start really on your early life. When you grew up, how did you first get interested in economics? Were there any particularly formative moments?

Steven J Davis: Well, as I recall, I took an economics course as a senior in high school.

I don't remember anything about it except that I found it interesting. So when they asked me to write down a major when I went to college, I picked economics, started there. And I dabbled a bit in political science, sociology, law, and international relations, sometimes in my reading, sometimes in courses.

Somewhere along the line, I don't remember exactly when the idea crystallized in my head that I really wanted to understand human societies and the human condition. Midway through my collegiate career, I think I came to the conclusion that economics had a better set of concepts and tools for doing that than other social sciences.

And so I decided I would go to grad school in economics so I could learn how to think about these matters. And that was the level at which I thought about it at the time.

Jon Hartley: Were there any particular recessions you experienced growing up or anything like that?

Steven J Davis: Well, we were in the middle of the 70s inflation, the oil price shocks.

So those were kind of formative external experiences, I would say, early in my development as an aware human being, say as a senior in high school and early years of college. So those were very much in the background. And so I had an early history in macroeconomics and unemployment, probably because I was living during times when inflation was high and unemployment was high in the mid-70s.

Jon Hartley: I guess a very formal time for a lot of economists, certainly which influenced the rational expectations revolution coming out of that time. A lot of macroeconomic models came out of that time.

Steven J Davis: There are lot of macro economists around my cohort as well because of the turbulent macroeconomic times in the US and elsewhere.

Jon Hartley: Well, great. I mean, how did you settle on sort of macro labor unemployment dynamics as an area of interest? You graduated from Brown with your PhD and you spent 35 years on the faculty at the University of Chicago. How did you stumble across labor, job flows, and later, economic policy uncertainty and work from home?

How did you stumble across labor jobs, and firms? How did you get interested in that?

Steven J Davis: I was interested in unemployment pretty early in my graduate school career, maybe even before I got there, I don't recall, but I definitely had an early interest in unemployment. My dissertation was inspired in part by David Lilien's work on sectoral shifts.

So that was part of my thinking. I read pretty much on my own a lot of the early search theory work by Diamond and Mortensen and Pissarides and their various co-authors. And I found it super interesting. It definitely influenced my thinking. There was no labor economics course at Brown until my fifth year when Robert Moffatt arrived.

And by then I was seriously working on my thesis and getting ready for the job market. So I never took a labor economics course in grad school, oddly. But as I said, on the macro labor side, I had done a lot of reading on my own, and my dissertation was motivated by David Lilien's work.

So that's how it started. And then sometime between I got my job at Chicago and I arrived at Chicago, it sort of dawned on me. I'd read some work by Darby, Haltiwanger, and Plant, and I'd started working on my own with what we now call the CPS, gross flows data, gross flows across different states of the labor market.

Unemployment, not in the labor force, in employment. And I said, wow, these flows are huge. This kind of net sectoral shift, which was net movements of employment across industrial sectors that were the focus of David Lilien's work, is not really where it's at. Especially if you wanted to understand unemployment as a frictional phenomenon in the mold of the Mortensen, Pissarides-type models that came to play such an important role in the profession.

So it's the combination of the very early empirical work. And I was that at that time somewhat steeped in the early theory of job search. And if you think back to these Mortensen and Pissarides style models, they typically feature some kind of matching function where the inputs are job seekers and open job positions in the form of vacancies.

And at least in the early versions of those models, the job seekers were typically people who had lost their jobs via layoff or sometimes via quit. So you start to ask yourself, well, what are the central inputs to the matching function, sort of a production function for producing matches or new employment relationships?

And it wasn't the intersectoral shifts that David Lilien was focused on. It was something closer to what was happening at the level of individual employers. So that led me down the path of thinking, and I soon started talking with John Halterwinger about this. Well, if we really wanted to be serious about this in terms of the empirical foundations or underpinnings of this frictional view of unemployment, we should go out and try to actually measure the loss of jobs and the flows of workers from other labor market states into the unemployment pool, into the pool of actively seeking work.

And so that's kind of how we started early on and how I thought about it.

Jon Hartley: That's fascinating. And I'm curious, did your time in Chicago shape your journey in any way? I feel like you're spending 35 years in such a storied department. It certainly, I'm sure, leaves an impression on anyone. And also you leave an impression on the department, of course, and the business school.

Steven J Davis: At least the first one's definitely true. I'll let others speak to the second one. Look, I came to the University of Chicago which I think of as one of its golden ages. So just first, it's interesting, the way that I learn changed when I went to Chicago.

Brown was a small economics department and maybe two or three seminars a week and maybe one in macro. I did an awful lot of learning at Brown by just reading and thinking and arguing with my fellow grad students as well. But that was usually in the context of homework problems.

But I really did spend a lot of time just reading on my own and going through articles, trying to understand them. And so when I got to Chicago, there's this, I don't know how many, 15, 20 seminars a week. And this tradition of very vigorous oral discussion argumentation was still quite alive, at least in some of the seminars in Chicago.

I went to a lot of workshops, and I often read the paper in advance and go to the workshop. I went to three or four a week. And I didn't just sit there like a bump on the log. I was intensely engaged, at least in a listening capacity and sometimes in a speaking capacity.

And so I shifted from learning about reading journal articles, which I continued to do, but all of a sudden, I'm learning tons orally by talking to people in seminars and out of seminars over coffee. So the way that I learn shifted, and I remember that. And then the people who really had a profound impact on the intellectual environment for me in those days.

So my favorite workshops were the applications workshop and the labor workshop, which was essentially the same cast of characters in both workshops. So we had Gary Becker, Sherwin Rosen, Eddie Lazear, Kevin Murphy, Robert Topel, and many others. I think Bob was away, maybe didn't come till my second year there.

He was at UCLA. So if you just think about the people I listed, there were five or six others who were, broadly speaking, in the labor economic space. It was just a tremendous opportunity to build my human capital in labor economics as well as macro. I mean, I went to the macro money workshop as well.

I just found them, they were a little less no-holds-barred in the style of discussion. It's a style of discussion which now is kinda gone out of style. It's no longer really accepted in the economics profession. It was very vigorous. People interrupted each other a lot. Especially in the applications and labor workshops, the speaker often spoke less than half the time.

And that was not necessarily a bad thing, because Gary may have spoken more than anybody else, and that was probably the right allocation. And so I learned a tremendous amount in these workshops. And I think back, those two workshops in particular, and the people I mentioned really made for a profoundly rewarding intellectual experience.

And so I kinda grew up in that culture and learned a lot about labor and about other things, but especially about labor. So my education during the first five years in Chicago as a labor economist, that's where it happened, even though I never took a course in the subject.

Jon Hartley: That's terrific. I know many people have gone through that labor group at Chicago under Gary Becker, either as faculty or students. I think of certainly folks like Glaeser, Steve Levitt, folks like that have gone through there. I'm curious, just getting back to firm dynamics and job vacancies, a lot of your work is hugely seminal in this space.

I'm curious, what, in your mind, have we learned about job creation and destruction in the recent decades as we've gotten better data on labor and firms? We often hear things like small businesses or small firms that are responsible for the majority of job growth in the US economy. “Small businesses are good.” That's kind of an often repeated thing. I mean, to what extent is this true in your mind? And I know some of your research says maybe otherwise, maybe somewhat younger firms. Do we see smaller firms being the most innovative or small businesses being the most innovative?

Is that necessarily true either?

Steven J Davis: No, so let me say a few things first. One of the big things we learned, which is now so widely accepted, is kind of just everybody's internalized it. But I think my early work with John Haltiwanger and also with Scott Schuh was instrumental in establishing this fact in an unambiguous, pervasive way, is that every market economy at all times and almost every sector has lots of job creation and destruction going on all the time.

Okay, and so that was, at least in early receptions of our work in the late 80s and early 90s, viewed as somewhat controversial. And then when accepted for the US, okay, well, the US is different. That's how the US operates. Everybody knows it's kind of a highly flexible labor market.

But the story we've learned since then is, as I said, outside of centrally-planned economies, which really do in the state-owned sector, often have very little gross job creation and destruction. The norm of a market economy is that every period, there are jobs that disappear and there are many other new jobs that open up.

That's job creation in our lexicon. And so what you should think of is the vision you should have of the economy, the labor market, at least a market economy. Yeah, there's a modest net change of employment from one period to the next. So maybe in the United States, we might have employment growth by 1%, so that might be 1.5, 2 million new jobs in a year.

But underlying the extra 2 million jobs that we gained on net, there might be 17 million newly created employment positions that weren't there say, a year ago and 15 million of the positions that were there a year ago are now gone. So that's a very basic fact, but it's important to internalize your thinking for many reasons.

First, back to the original motivation of why I went down this research path. It helps you understand how frictional unemployment can be a significant phenomenon, even in a well-functioning market economy because lots of jobs are disappearing and those people either have to leave the labor force or find a new job.

They often have an intervening spell of unemployment while they're seeking a new job. There's a lot of resource reallocation happening in the economy all the time through the labor market, directly from these statistics. But often, capital and other intangible forms of factor inputs are also being reallocated at the same time.

And there's a strand of literature which I played a modest role in, but John Haltiwanger and others played a much bigger role. That goes off and looks at the connection between that resource reallocation process, which is ongoing in the economy all the time, and productivity growth. And I think the central message from that literature as I take it, there are many nuances.

But the central message is part of the way that the economy revitalizes itself and pushes productivity advances is by continuously reallocating labor and other inputs from less productive, less valued uses to more productive, more valued uses. So that's really two, I kind of give you two big messages so far.

Now, the small business piece that you asked me about. So let me back up. There are so many myths and half-truths around the role of small businesses in the economy, it's kind of hard to cut through them all. And at some point in the mid-90s, I just got irked and sick of hearing all these ridiculous claims and misleading claims about the role of small businesses in the economy.

So I thought I'd just write a paper, which seemed to me to be stating a bunch of obvious points about the misleading nature of many of these empirical characterizations. Don't get me wrong, I don't dislike small businesses. I think they play a hugely valuable role in the economy in many respects, but there were so many outlandish claims about their role in the economy.

Sometimes, often coming from the SBA, the small business advocacy or agency, whatever it's called which particularly irked me because they're a taxpayer-funded organization, part of the government-

Jon Hartley: Small business administration.

Steven J Davis: Small business administration, right? So I decided, look, I'm gonna write a paper and just set the record straight.

So I'm just tired of these things and then everybody will understand what's wrong with these characterizations that we can go on. And this paper which to me, I really wrote for an undergraduate. I was kind of hesitant about writing. It was still early in my career and I was a little concerned about, aren't my colleagues going to think that I'm just writing these papers that state the obvious?

That paper, for whatever reason got an enormous amount of attention and I led to a picture on the front page of the New York Times business section with me in some factory, and so on. I think it was a more glorious thing to be on the New York Times in those days than it is now, but let's go back.

Here's the substance, there are many of these misleading characterizations made about small businesses. Let me take up a few of them. So one thing to understand, there's a statistical sleight of hand that comes up all the time, and not just with respect to small businesses. The scope for this mischaracterization arises particularly because the gross job flows that we talked about before, the newly created positions and the lost old positions are so much larger than the net changes from period to period.

Net employment in the United States, if it changes by 2 million in a given year, and you might then go look at, well, how much did the net position of small businesses change over that one year? And you can suppose it just happened to be, it was 2 million as well, and there was no net change in the number of jobs at larger businesses, however, you wanted to find small and large, then there's a sense in which it's true to say, it's correct to say that all of the new jobs were created by small businesses over this period.

Now there's a sense in which that's correct, but there's also a sense in which it's extremely misleading. And so why is that? Well, because within, among the set of small firms, there are lots of jobs being created, maybe 10 million, and lots of jobs being destroyed, maybe 8 million.

And among the larger businesses, there are lots of jobs being created. And I guess if I wanna make my numbers add up right say there's a 7 million new jobs created and 5 million that are disappearing, okay? So okay, it's true that in some accounting sense, you can say the number of net new jobs created by small businesses is equal to the number of net new jobs.

But you could also go find a set of large businesses because I just told you there are 7 million new job positions created, large businesses, you can go off and carve off a set of those and say, you know, those five industries, large businesses, and those five industries accounted for 100% of all the net new jobs.

Or you could do something that governors who run for presidential in presidential elections are very fond of doing. If they come from a state that's had a lot of net job growth, you can go say, you know, from 19 x to 19 x, 85% of all the new jobs created in the United States were in Texas or Florida or New York or California, whatever it happened to be.

The point I'm trying to make is all of these statements, in some sense, in an accounting sense, can be true. They're just very misleading in terms of sometimes the way they're said, but certainly is the way they're heard by their intended audience. So we wrote in this paper, we wrote in the mid-90s.

We call this netting out reality. So if you give me the micro data and especially if you're looking at a period where the net change is small, I can go find 100 different groups defined by size, by age, by geography, by industry where I could make this kind of statement that 100% of the jobs were created or accounted for by this group.

Okay, so when you explain it that way, you see why this statement is. It can be both factually correct, but essentially nearly meaningless and quite misleading in the way it's often received by the audience. That's one of the problems. But as I tried to make with my geography examples, this statistical sleight of hand or mischaracterization Extends well beyond the small business issue.

Jon Hartley: It's fascinating, I mean, a huge contribution. When we normally think about jobs, a lot of macro people tend to follow the BLS employment situation report. And you get the monthly non-farm payroll numbers, and jobs numbers, and you hear 200,000 net new jobs. But people forget that that's just net, and there's plenty more gross job flows moving around.

And that's, I think, a huge contribution of, I think, perhaps culminates in your 1996 book, Job Creation and Destruction with John Haltiwanger and Scott Schuh. And also I guess, it really speaks to, I guess this Schumpeterian point about creative destruction and provides a lot of empirical backing for, I guess, that sort of underlying theoretical kind of idea.

Steven J Davis: On that point, I sometimes think of our work as “Schumpeter with data”.

Jon Hartley: That's a great way to frame it. Yeah, I think you couldn't have put it any better. And it's amazing too. I mean, Schumpeterian creative destruction gets cited a lot, I feel, and it's amazing that you were really the first researchers to really emphasize that point in the data.

Just shifting gears here a little bit, I'm curious what you think about the concept of full employment. It's this old Keynesian labor market concept that suggests that there's slack in the economy when the factories are empty and the workers are on the sidelines.

But I feel like today in 2024, we don't really have a good way of say, measuring it, particularly given how dominant the service economy is now. The CBO potential GDP full employment measure, in my mind, is kind of a way of cheating or in part updating in a somewhat backward-looking way. We discover that growth doesn't necessarily get back to its old trend like it didn't after the Great Recession. The CBO potential GDP full employment scores get updated in this backward-looking way to account for the fact that there actually was hysteresis. But some researchers say it's important to look at vacancies on top of unemployment.

The Beveridge curve, I think is historically a bit more of a pretty reliable, stable relationship. I mean, Covid shifted things out quite a bit. But historically, I think the relationship between unemployment and vacancies has been a lot more stable than say, the Phillips curve relationship between unemployment and inflation.

Steven J Davis: It's a low bar.

Jon Hartley: A low bar, but I'm curious, have we learned anything in your mind about what full employment is in recent decades, or in your mind, is it kind of an outdated concept now?

Steven J Davis: Here's how I think about it. Both full employment and in some senses mirror image slack, they are useful concepts.

They're also difficult to measure. And I think it's a mistake to rely on any single indicator of either full employment or slack. And because in almost any time period or episode, you can tell me about, I'll tell you what was wrong with what indicator in that episode and why it gave you a misleading answer based on how it behaved then compared to the historical pattern.

So I think we ought to look at a whole bunch of indicators, and this is typically what I do. We want to look at the employment-to-population ratio, probably adjusted for the age mix of the workforce, the unemployment rate itself, which comes in many varieties, as you know, it's not just a headline number.

Vacancies, yes, but also the vacancy to the unemployment rate. How long does it take to fill vacancies? I also look at matching efficiency that pops out when you plug your favorite measures of job seekers and open job positions into a matching function. I tend to prefer here, I'm of the same view as Bob Hall that I find the matching function to be actually a more useful construct than the Beveridge curve itself, although I've used both.

And you could go on in this vein, I don't think there's a single measure of full employment or slack that you want to hang your hat on. Now why are these useful? Broadly speaking, we want to get some handle on how close we think the economy is currently operating at close to, I left off, by the way, capacity utilization measures.

We can also measure the utilization rate of the physical capital. So there are others as well I should mention, that want to get some sense of how close is the economy to operating at its full capacity in some sustainable way. That's interesting all by itself, because if we, in the extreme cases when there are millions of workers, or tens of millions of workers who are out of work, who don't seem to have any near-term options to find work.

So we're not talking about a situation of just frictional unemployment, then something is amiss in the economy. So we want to know whether the economy is operating close to its capacity. And then in many, many theories, Keynesian theories in particular, and I put some weight on this idea, although not too much.

But some weight on the idea that if labor markets are extremely tight in that there's not much slack, or that we're at full employment, or even above various measures of full employment, we might expect wage pressures to be pretty strong and wage inflation, if not already, then in the near future to tick up.

So that's why I think these measures are useful, but they all have their weaknesses, and hence it's most useful to look at several of them. In recent years, in particular during COVID, but in the wake of COVID and partly related to the shift to work from home and what that did to the workforce, the usual relationships in the Beveridge curve and vacancy durations, I think these measures differed quite a bit in terms of what they were telling you about the tightness or slack in the labor market at that time relative to what you would have inferred from the same statistics before the pandemic.

Jon Hartley: It's interesting to think, too, that concepts like NAIRU were very much Milton Friedman's concepts, but are Keynesian in many ways. But it's interesting just to think over the past few years how unemployment's fallen to such a low level that essentially it's forced people to revise their estimates of what NAIRU (u*) is.

And so people thought NAIRU or u* was 5%, then unemployment falls below 5%, pretty meaningfully, closer to 4 or 3%. And yet inflation didn't really start rising at that point. At least in the pre-sort of pandemic period. I feel like the Fed has really moved away from thinking about concepts like, say, u* or r*, and is kind of more just looking at the actual prints themselves and just trying to shift in the direction against the winds of inflation.

Or just operate as to whether unemployment to levels that are well above, the historical means. Forget about what some sort of idea of model-driven u* or r* is. In my mind, they've just sort of assumed now that u* is just the long-run historical mean.

Steven J Davis: Yeah, that's not a very satisfactory approach, in my view. And even when moving beyond that, much of the literature about the evolutions of, say, the natural rate of unemployment is really focused on just the demographic mix of the work of the working-age population. I don't think that's very satisfactory.

And I can explain briefly a couple of reasons why that's so, and this predates the pandemic. One has to do with both back to our job, and our earlier discussion of job creation and destruction flows. Those flows have trended downward. The rate of gross job creation and destruction and the rate at which workers lose jobs to vacancies has been trending downward for a long time.

And part of that has to do with the aging of the workforce. Part of it has to do with, even conditional on the age of the workforce and the age of existing employers, there's been a reduction in the rate at which people get laid off. Okay, so other things equal, you would expect that to reduce the extent of frictional unemployment.

I've written a few papers in this space with others. I think there's a pretty compelling case that there's been a long-term downward drift in the natural rate of unemployment, partly for that reason, even over and above any contribution you would get from demographic shifts alone. So I think part of the low unemployment rates that we've had for quite some time now, apart from the pandemic episode, but in the years since the global financial crisis fully played out, we've had pretty low unemployment rates.

And many people have often, I think, misinterpreted that as an indication that labor markets were really tight and that inflationary pressures would be high. One more point on this part of what's happened, and not the whole story at all, is the way in which matches formed is changed over time.

There's less ex-post experimentation. That is, we're going to hire some person and see how they work out. And if they don't, work out, we're going to fire them. So there's less ex-post experimentation, there's more rigorous ex-ante selection. I think that has happened both for regulatory reasons, that the costs of getting rid of somebody who didn't work out have grown higher for regulatory and legal liability reasons.

But also the technology for evaluating people before you actually hire them has probably also increased. We live in a much more data-rich world. You can acquire more information, you can rely on algorithms to help you evaluate people if you'd like. That has its own drawbacks. But there's a whole set of reasons, and I don't think I confidently can parse it out.

But there's a whole set of reasons why we've shifted from ex-post experimentation, which tends to generate a lot of frictional unemployment, to more ex-ante selection and longer-term employment relationships. That also feeds into some extent to the reduction in the extent of frictional unemployment. So these are just some of the reasons why a simple statistical approach to characterizing the evolution of u* just as a function of demographics, would lead you astray, in my view.

Jon Hartley: Absolutely, it's a fascinating concept and a fascinating debate. I want to shift toward economic policy uncertainty. This is an area in which you've been very active in the recent decade. You, along with your co-authors Nick Bloom and Scott Baker, developed economic policy uncertainty indices based on various uncertainty keywords in the news and the frequency at which they appear.

These measures have been hugely influential in both academia and the private sector as well. How much, in your mind, does economic policy uncertainty matter for firms and growth?

Steven J Davis: So I'll give you my overall assessment is that in most times and places, economic uncertainty and economic policy uncertainty are modest factors in explaining fluctuations in firm-level outcomes.

And certainly, I should explain the aggregate economic fluctuations. However these fluctuations in economic uncertainty are also highly skewed. And there are episodes, and I think of the period during the global financial crisis and in the years after the financial crisis as one example. The COVID period is another example, where there were extraordinarily high rates of both uncertainty and policy-related uncertainty that did significantly restrain investment in hiring decisions and mattered quite a bit at the macro level.

So I wouldn't put them in the same category as many of the first-moment shocks that are often driving even the everyday fluctuations, the more quarter-to-quarter fluctuations in income activity. But I do think that there are, are episodes in which they really matter a lot, and often episodes in which times are bad or you're being hit by bad first-moment shock, so to speak.

So these things aren't uncorrelated with other things happening in the economy. So the extra uncertainty is often being layered onto other stresses and strains that the economy is facing and in effect, amplifying those negative effects.

Jon Hartley: And it's interesting, and I wonder too if there's maybe an optimal amount of policy uncertainty.

Like obviously there's going to be more policy uncertainty in democratic regimes compared to, say, autocratic regimes where things are probably pretty low uncertainty. But I remember when talking about some of your research, which kind of came out of, I think, was heavily sought after during the deadlock congressional periods in the 2010s.

I remember working at Goldman Sachs and a colleague from China who said the problem with the US was they could never get anything done amidst all this policy uncertainty and things like that. I also think, well, having complete policy certainty might not be so great if the policies aren't so friendly toward markets in general, say in a totalitarian regime like China.

But I'm curious if you have any thoughts on that.

Steven J Davis: Yeah, a few thoughts. First, there's always going to be some level of economic uncertainty and even policy-related uncertainty with us, so let's just put that on the table. But much of the motivation, much of the reason, Nick and I, well, speak for myself, much of my motivation to get into the business of trying to measure policy-related uncertainty early on arose because there was a lot of, in the United States, at least politically, manufactured policy uncertainty that didn't seem to serve any productive purpose.

So if you remember, we had a lot of fiscal cliff-type episodes, tax code expiration episodes, and debt ceiling fiascos in which these are all things where the policy-related uncertainty wasn't built into the system the way, say, an election cycle is. It was a consequence of the way that policies were designed and the negotiations between the different competing political factions in Congress.

So perhaps the clearest example of this was the debt ceiling crisis, where Democrats and Republicans were essentially playing chicken because they were trying to get what they wanted with respect to other aspects of the fiscal legislation, and they weren't going to come to an agreement until the other side blinked.

And what was at stake then? What was at stake then was the capacity of the federal government to make timely payments on its various obligations, including even possibly, interest payments on Treasury securities. And so there was a non-trivial probability of some type of default event in US Treasury securities.

And given the central roles they play in the monetary and financial system in the United States and globally, that's not something you want to mess around with. I could go on and give you many other examples like this, but in this period in particular, in the wake of the global financial crisis, there were many such episodes involving fiscal policy.

The debt ceiling crisis. We talked about health care policy. Remember, we had at least two Supreme Court decisions where the Affordable Care Act, Obamacare, hung in the balance. That was partly a consequence of the way the legislation had been designed. So this was not an inevitable form of economic policy uncertainty.

This was built into the system consciously by policy design decisions and political decisions. So that's the second thing I want to get on the table. Third thing, I'm going push back a little bit on your claim that at least in democratic regimes, yes, there's policy-related uncertainty in democratic regimes.

We've got a presidential election coming up. It looks like a lot rides on that a lot road in other recent presidential elections. Certainly, there was a big fallout in financial markets in the cross-section of Trump's surprising victory over Hillary Clinton. So much of my work in the policy-related vein kind of draws out the role of election-related uncertainty as one source of policy-related uncertainty.

But you don't wanna give autocratic regimes a pass. I mean, just think what Russia's been doing in Ukraine. You want to talk about a generator of policy-related economic uncertainty. That's an enormous one, and that may be the biggest one in much of Europe in recent years. So autocratic regimes generate uncertainty politically.

Policy-related uncertainty as well. It's just not so closely tied to the election cycle.

Jon Hartley: That's a great point, yeah, it's interesting. Even dictators can be very volatile and very unpredictable, and that's an excellent point. So I want to shift to your work on work from home. You have a famous paper with Jose Barrero and Nick Bloom titled “Why Working From Home Will Stick in the US”.

It does seem, according to the data that you track, that the percentage of full-time days working from home has stabilized around, say 30% or so over the past, say year or two. Now, that number, which if I understand correctly, only applies to service sector jobs. So we're talking about 30% of service sector jobs, roughly.

Steven J Davis: This number applies across the board for employees. Now, actually, I'm trying to remember whether we have the self-employed group in that or not. But most of the US economy is in the service sector now. So it's also true that most goods-producing jobs don't lend themselves to remote work.

Jon Hartley: Right, exactly, so it stabilized, this is the number of workdays, work from home, stabilized around 30%. It's still up from, say around 6% or so where it was before the pandemic, but it's also down considerably from where it was at the height of the pandemic, understandably so. When many people who are able to work from home are working from home becaise they can't go to work.

I'm curious, what in your mind have been some of the largest impacts of working from home, whether it's real estate or in the labor market? And how do you see the future of working from home? Do you see remote work becoming more common? I know your co-author, Nick Bloom, often talks about maybe a J curve where there is sort of a rebound in the share of work-from-home jobs or the number of days work from home.

Curious where you stand on that.

Steven J Davis: So I think this is the most profound shift in how we live and work. Certainly in such a compressed timeframe that's happened in decades. It's hard for me to think of another example, of such a big shift in how we live and work that happened so abruptly outside of wartime.

And there have been bigger shifts. The Industrial Revolution was a bigger deal than this. The transition from factories to offices was a bigger deal, but these things played out over centuries or decades. And this big shift in how we live and work with respect to remote work played out initially in response to the pandemic in a few weeks, and then over the course of about took about three years to settle down to where we are now and have been for some time, which is roughly 28% to 30%.

As we measured in the survey of working arrangements and attitudes, 28% to 30% of full paydays are worked fully at home or some other remote location. It could be a coffee shop, a library, even a friend's house or something, but somewhere not at the employer's work site and not at a client's work site.

So that's just an enormous shift. There are many aspects of it, some positive, many positive, some negative, some positive ones. The most obvious one, perhaps, is the time people save by not commuting. The time and the money they save by not commuting. That's a really big deal.

It's kind of a mechanical effect. But it's not to be taken lightly because the average American on the margin between working from home or working on-site, spends about 65 to 75 minutes in extra time on commuting. Plus a little extra time on grooming is a little extra time grooming. When you go into the work site, maybe you put on better clothes, make sure you shave, that kind of thing. So that's a lot of time. It means if you work from home three days a week, you're saving more than 3 hours out of your day. So that's a huge benefit.

There's also just the flexibility in time use over the day, which appears to be especially valuable to people who have young kids. So maybe you take 15 or 20 minutes out of your workday when your kids come home from school, or if you need to take your kid to the doctor or the dentist one morning, you do that.

And that's pretty, that's much easier to do if you can go from home rather than have to connect it to your commute, that's 30, 45 minutes away. So people value this flexibility and time use over the day. Many of them also value just the personal autonomy that comes.

I have before you before I joined this podcast, I had a little soft jazz going on in the background while I was trying to write a memo. I don't usually have music on when I'm writing my papers, but if I'm writing some administrative memo, I want to hear something soothing and calming.

I'm defining my background, I'm doing that here in my office, but for people who work in a cubicle, that's not so easy. Those are some of the benefits. We can get into the productivity side if you want, but that's a complex thicket of the productivity effects. But it does have a whole range of effects on productivity.

It affects how companies organize their productive activity and how managers operate. So managers who have hybrid or fully remote workers have had to learn new management skills and styles. Organizations have to adapt their HR practices and compensation and evaluation policies if they're going to have some workers who are remote.

So big changes in organizational practices management, it's had an effect, in my view, on the structure of wages. So wages for people in often highly compensated professionals, but say in the finance sector, and the business services sector, have seen rather slow compensation growth since 2021. Even though these are the industries and occupations that tended to see rather rapid compensation growth relative to most other sectors in the previous decades.

I think, and we have plenty of evidence to back this up, some of that slow compensation growth is because people in these sectors have decided to take some of their compensation in the form of the benefits that come along with working remotely. So could go on in this vein, we can drill in there more if you'd like.

But you also asked me, where do I think it's going? So I think the main changes, the main rapid changes are behind us. I base that partly because not really much has changed in the last 18 months. Even though you hear all these news accounts about calling workers back to the office.

And there continue to be, in some organizations, struggles between management and staff as to exactly how much remote work there will be and what the parameters are around that. And there may be changes. In some organizations, the overall data doesn't show much change, both our data and other data sources.

In the past twelve to 18 months, we have, through the Survey of Business Uncertainty at the Atlanta Fed, asked business executives forward-looking questions. And what we do is we ask senior business executives about the outlook at their own firm, and we've asked them to project five years ahead.

So we say first we ask them, well, what fraction of your workforce hours are currently remote? And so on. But then we ask them, well, projecting forward, what do you think is gonna happen to the extent of your hybrid work, your workforce, and your fully remote workforce? And we break down those two.

And if you average across all of the employer executives that we surveyed, and you weigh them by the size of the employer, and here it doesn't really matter whether you weigh them or not. What you find out is executives, when you ask them about their own firm, they foresee over the next five years modest further increases in the extent of work from home, both on the hybrid side and the fully remote side.

Now, I'm a little less confident in that projection for the fully remote workers because I think the workers who are fully remote are also the ones who are potentially vulnerable to domestic offshoring. And it may be harder for business executives to factor into their own thinking the changes that might happen in that regard.

But I take away from all this evidence and other evidence that we could talk about is kind of in a new normal. The best guess is probably over the next five to ten years, modest further increases from where we are now in the extent of remote work. And by that, I don't mean fully remote work.

I mean, the fraction of workdays that are performed remotely.

Jon Hartley: That's fascinating. I, too wonder to what degree, say, remote jobs might be more susceptible to, say, generative AI. There's speaking of offshore jobs or remote jobs being more likely to be put offshore. My understanding of generative AI jobs and tasks that are being somewhat more quickly automated away, at least in terms of what sort of data and evidence we have so far know jobs like those that say call centers that are largely offshore are more susceptible to displacement from generative AI. I wonder if there's maybe some sort of intersection there.

Steven J Davis: I think so. Call centers would be the first one that comes to my mind too. There's already been a lot of offshoring of call centers, as you know, especially to places like India and the Philippines, where there's a large domestic population that speaks English and is reasonably well-educated, and is willing to work for wages that are below American wages.

And so they're often handling maybe the more routine type customer service queries or customer relationship type queries. Those do seem amenable to AI, generative AI, more so than most jobs, at least in the next few years. And so I do think there is the intersection exactly that you described, those jobs that are currently fully remote in the United States are relatively susceptible to displacement through generative AI, but they're also closer to the margin of being offshored.

Jon Hartley: Well, it's fascinating and really a fascinating discussion and a real honor to have you on, Steven, here about your amazing career and ideas. Really want to thank you so much for joining us today.

Steven J Davis: Thanks for having me, John. And good luck as you continue your podcast.

Jon Hartley: This is the Capitalism and Freedom, the 21st Century Podcast, an official podcast of the Hoover Economic Policy Working Group, where we talk about economics, markets, and public policy.

I'm Jon Hartley, your host. Thanks so much for joining us.