Nicholas Bloom on Economic Impacts of COVID-19 in the Short-run and Long-run

COVID-19 will cause immediate and lasting economic damage in the form of increased uncertainty, heightened risk aversion, lower productivity growth and more.

Nicholas Bloom is a professor of economics at Stanford University and a leading scholar on management, productivity, innovation and economic uncertainty. Nick is a previous guest of Macro Musings and returns to share his thoughts on COVID-19 and what it means for the US economy, both short-run and in the long-run.

Read the full episode transcript:

Note: While transcripts are lightly edited, they are not rigorously proofed for accuracy. If you notice an error, please reach out to [email protected].

David Beckworth: Nick, welcome back to the show.

Nick Bloom: Thanks for having me.

Beckworth: Well, great to have you back on. I know you're on the other side of the country. You're on the West Coast at Stanford University on the campus there. So, how have you been holding up?

Bloom: For all of us, it's obviously a tough experience, though I think we're very lucky living out here from talking to a lot of my friends and relatives are in London or in Glasgow, and I think being in a suburban setup is a lot easier. So, we have a bit of space. The lockdown's got progressively tighter. Initially it was reasonably relaxed, but now much of the open space, all the parks, sports fields, everything is closed down. I see a lot of my neighbors walking up and down repeatedly on my street. But yeah, it's like everyone else who are trying to get through it.

Bloom: Trying to get food, as well, actually the other issue is, it's increasingly really hard to get any food if you want to deliver it. So, you either go to the supermarket or eat old bags of rice.

Beckworth: So, is it difficult when you go to the grocery store? Do they space you and time you when you go in?

Bloom: We've been too nervous to go out right now.

Beckworth: Okay.

Bloom: Right now old rice is looking more appetizing, but honestly we're running pretty low. It feels like medieval times when you're coming to the end of the winter. Will you make it through to the spring to get more food? I don't know, we've been having this discussion.

Bloom: Actually, around here it seems like in Santa Clara County where we are the infection rates are kind of flatlining, if anything slowing off. So, we may actually in two, three weeks it may be fine to go out.

Beckworth: Okay. Well I hope it continues to get better there. I mean, California's done relatively well, compared to say New York, so you guys have done that really intense locked down early on, so maybe it's paying off. But you have some great work on the crisis. You have a new NBER paper on COVID-19 with your colleagues, ones you've done work with before and we want to talk about that. And we want to frame our conversations today around what you see are the short-run consequences as well as the long-run consequences of this pandemic in the United States and elsewhere as well.

Beckworth: So, you have a paper with your colleagues titled “COVID-Induced Economic Uncertainty” and you've written papers before with these colleagues of yours on economic uncertainty. But it was an interesting paper because you're taking a look at what the implications will be here in the short-run. So, walk us through that paper and some of the big conclusions you find in it.

Short-run Impact: Dealing with Uncertainty

Bloom: Absolutely. So, yes, this is with Steve Davis from Chicago, Stephen Terry and Scott Baker, who were both previous students of mine at Stanford, co-authors now. So, one of the things we started to do is like how would you measure uncertainty in the COVID recession? So, traditionally there's this whole literature on measuring uncertainty. There's a number of different indicators. And until recently, there's a big debate about what you should use for identifying recessions, but suddenly under COVID an immediate factor's been the speed of the change, means you want an indicator that's both timely and forward looking.

Bloom: So, just to put this in context, February, 2020 was the best labor market in probably 60 years based on unemployment numbers. So, they were 3.4%, by April 2020 it's looking like it's going to be the worst since the Great Depression. And so you need an indicator that was available basically with a lag of one, two, three days and is also forward looking. So, that rules out some of the classical measures which look at, say, historic GDP volatility. So, we know that GDP is much more volatile in recessions than in boom times, but of course we don't even have GDP yet for quarter one and we're not like to get it, for a while, for quarter two. And so, we really want something forward looking.

Beckworth: And so, you look at three indicators, stock market volatility, newspaper base, economic uncertainty, and subjective uncertainty in business expectations surveys, is that right?

Bloom: Yes, exactly. So, the three measures you look at is basically, firstly there's a set of financial market indicators, so things like realized vol, so how much the S&P moves up and down within the day, or a day by day. Very related to that will be for example, VIX, so that's the implied volatility in the S&P 100 Index. You know the standard measure is one month ahead. But in fact with Ian Dew-Becker and Stefano Giglio, we've been looking up to two to four years ahead. So, one is a series of financial measures. You could look at things like bond price volatility or exchange rate volatility. A second set looks at newspapers and, in fact, in particular looking at all daily U.S. newspapers, there's around 2,000 of them available on a data source called NewsBank and how much their articles discuss economic uncertainty. And then thirdly is looking at surveys.

Bloom: So, with the Atlanta Fed and with the Bank of England, I've been involved in two monthly online surveys and they get mailed out on the first Friday of the month. And then firms have two weeks to respond and then it closes. And all three of them have the big upside that the data's available almost real-time. The surveys take two weeks, but the other two are basically that day or the next day. And also, of course they're forward looking, so you notice, for example, throughout March that the Economic Policy Uncertainty Index starts to really climb very rapidly.

Bloom: So, at the beginning of March, COVID wasn't, it was an issue ... Remembering, I really date it from the 24th of February, which is the first major stock market drop in the U.S. based on COVID. And so, by early March you're still really a week in, but by the end of March fears are for a great recession and it's just totally dominating all measures.

Beckworth: Now, if I understand your paper correctly, you take this measure and you plug it into a model, and you try to pull out or extract what it means for GDP. And you have this pretty staggering conclusion that GDP will contract 11% in 2020, quarter four, and there's a wide confidence interval around that, but is that number right?

Bloom: Right, a VAR based on another paper I've been working on with Stephen Terry and Scott Baker, which looks at GDP growth over a panel of 40 countries, going back 40 years. But the idea of that paper originally was to look at the causal effect of uncertainty and to get at that we wanted to look at what we call disaster shock. So, kind of exogenous unpredicted events like terrorist attacks or natural disasters, coups or revolutions, and pass them into the first moment effect. So, the impact of those that literally just lowers demand and supply. And then the second moment effect, the effect that's really all about uncertainty. And use that to try and think about how much these events, and basically first and second moment effects, drive GDP growth.

Bloom: And we take the parameters from that model that we estimate on the big panel and then feed in the COVID news from the U.S. So, we feed in the stock market level and volatility, and that predicts a drop in GDP this year bottoming out at around minus 11 or 12% later in this year.

Beckworth: Yeah. Which is a very large number. I mean, the Great Recession, what was it, 3%, 2%? I mean, the contraction we had back then and ... Yeah go ahead.

Bloom: Yes. I was going to say, I mean, it's huge, but on the other hand I think, I'm aware that I've been very pessimistic. So, at the end of last week I had to give a reasonably big talk here in Stanford about it. And my numbers are ... I've been looking a lot at the consensus forecast. So, the consensus forecasts for the U.S. are about minus 4% in quarter two. So, that's a quarterly growth rate. If you times it by four you get the annual growth rates of minus 16%, which you hear in the news.

Bloom: Just to give you a benchmark, Goldman Sachs gave that as minus 6%, which is where their annualized growth rates of minus 24 is. But the consensus forecast actually has growth rebounding in quarter three, 2020. So, to give you for example, just to put the numbers on the table for Goldman Sachs, the latest figures, quarter by quarter they predict the U.S. economy is going to be down almost 10% in quarter two but we'll bounce back at about four to 5% in quarter three, and about a couple of percent, quarter four, and quarter one beginning of next year.

Bloom: So, on their numbers by quarter one next year, so by about a year from now, we're back to trend. My forecasts, I have to say, are much more pessimistic than that. I have as a much bigger drop next quarter, and quarter three, I don't think is going to see a large rebound. So, that even by mid next year I think we'll still be quite below trend from where we are now.

Beckworth: Yeah. Just to put this in historical perspective, so the Great Depression over three years or so, from '29 to 1933 saw contraction of the real economy, about 20%. Nominal GDP fell about 50%, so another 30% in the price level. But a 20% contraction over three years, still a pretty devastating experience. But you're suggesting something half that big within a year. And if you look at your confidence intervals, you actually do take it all the way down to 20% possibly. So, it could be as severe as the Great Depression but compressed within a year.

Bloom: Yes. So far, so there's two kind of parts of the debate. One is the short-run effect, which I say, what's going to happen in the next three months? And I don't know, this podcast probably goes out next week. So, is right now, April the 13th, we're not going to know for another two, three months, but pretty soon we're going to know what the, at least, measured GDP figures look like for 2020, quarter two. And we can come back and discuss what the actual GDP, because GDP measurement has a number of issues, which is problematic right now. So, one debate is how big the drop will be immediately. I'm more pessimistic than others and I can come back to some other data on that. I think the much harder to pin down issue, and what there's far more uncertainty over, is what the longer-run rebound's going to look like.

I think the much harder to pin down issue, and what there's far more uncertainty over, is what the longer-run rebound's going to look like.

Bloom: And back in early March, people were talking about I-shaped recession, so an instant drop and rebound. And then, mid-March it was more the debate about the V-shapes recession. So, I don't know if you remember that old saying by Milton Friedman, he compared recessions to like plucking a guitar string. So, the harder you pull it down, the faster it snaps back. So, that's the V-shape and now I think the debate has moved on to rather we're going to have a U-shape, so a longer drop and recovery, and that's my own personal view. I don't think we're going to quickly bounce back throughout 2020 and that's also consistent with the forecast out of our VAR model.

Beckworth: Yeah. So, you have a very depressing outlook for this year, to put it mildly. And something that's really hard for, I think, anyone in our generation to fathom, go back 1930s, '40s, people might understand this better than us, but even they, again, they had three years to kind of embrace this shock that caused the economy to contract this much. Now, one other thing that you bring out in this exercise is that most of that contraction will come from the uncertainty. I just wanted to think through some of the channels through which this virus, this pandemic shock is affecting the economy. So, one obvious one would be simply the negative supply shock. We've shut down a portion of the economy. We were talking about this early on when China was shut down, people were talking about global supply chains are shutting down. So, that's clearly a supply shock and now people are saying, well there's demand shock involved as well.

Beckworth: So, you can think of like the stock market contracting, there's wealth effects, there's maybe risk aversion effects, but there's many channels you could tell the story through. But what you find is the most important channel is the uncertainty channel. So, walk us through why is uncertainty so high in this particular crisis as compared to say a financial crisis?

Bloom: Sure. So, I mean our numbers are like 60/40 but I would say there's big confidence intervals around them. I'd say, broadly speaking, uncertainty is roughly equal magnitude to the first moment component. And then the question is, why? Well, firstly, uncertainty is incredibly high right now. So, the best two metrics for real time analysis, I think, are stock market volatility and the VIX is up about 400% on this baseline.

Beckworth: Wow.

Bloom: It's fallen back down, but it went over 80, where it was averaging about 15 beforehand. And then the other measure is our Economic Policy Uncertainty Index, again, has gone up about four X, it's long-run average around a hundred, and it's right now averaging around 400. I mean, actually interesting, other indicators that are useful are more anecdotal ones. Like last week, General Electric pulled all its forecasts for profits, saying it was impossible to make any predictions in the current scenario.

Bloom: Starbucks, FedEx, various other companies have literally pulled forecast, which I've never seen before. So I think the uncertainty is incredibly high. The way in which it's damaging growth is from a couple of dimensions. One on the business's side, you can imagine that no businesses right now are going to be investing in tangible, and particularly intangible capital, so R&D, training, et cetera. You're going to be preserving cash. They're going to probably almost completely frozen hiring. Even Amazon is not dramatically increasing hiring, nowhere near the amount that other firms are dropping.

Starbucks, FedEx, various other companies have literally pulled forecast, which I've never seen before. So I think the uncertainty is incredibly high.

Bloom: Then if you flip it to the consumer side, of course, consumers are likewise not buying durables, becoming much more risk averse, saving. So it's hard to tell how much of this is pure uncertainty versus an increase in risk aversion. But I think both consumers and firms are going to be dramatically reducing expenditure, and that's driving a lot of what we see as the drop in demand. Now of course, supply is also dramatically hit directly, so it's hard to parse the two of them out. But just feeding in the stock market drop and increase in goal based on prior numbers gives us a roughly 50/50 split between just first moment versus the uncertainty effects.

Beckworth: The uncertainty stories sure seems intuitive to me. I mean, every day there's a different story for how long this will take or have we bent the curve, yes, no, what's the approach going to be, are we going to open up the economy, yes, no? It goes back and forth. There just seems to be so much uncertainty in general. This has to be an important challenge based on anecdotal readings of news stories and such, and you provide systematic evidence that that is the case.

Bloom: Yeah, again, it's very hard to be precise with numbers, but just as measurements, the only uncertainty measures I think we have that are available going back are the long time series and looking forwards, which is stock vol, these media measures are up four fold. The survey data I've been getting is also horrendous, so the firms forecasts and their reduction in sales. You know the numbers in March, at the beginning of March these surveys are only open for two weeks, but even the first people responded in early March, they're giving minus five, 6% numbers by late March is like minus 15, 20%. Some more recent work I've been doing with a large FinTech company here, they survey thousands of small businesses, they're predicting sales slowdowns of about 40%.

Bloom: If you look at, there was just an NBL working paper that came out today by, I'm going to forget all the people, I feel terrible, but Zoe Cullen, Ed Glaser, Marianne Bertrand, there's like five or six people. They had sales drops that were predicted from their respondees of minus 40, 50%. Another way to look at it that I actually really liked, the analysis is by Moody's, so what Moody's have done is they've done a bottom up looking at state by state, sector by sector, looking at the amount of activity that's been prevented by the shutdown. And they estimate that across the U.S., we are now per day losing roughly 30% of GDP. So once you do those numbers, you can figure out that quarter two 2020.

Bloom: Imagine the best case scenario, the lockdown runs until the end of April and then from May onwards, a return to normal. In April alone you've lost about 30%, so on average for the three months you must be down 10% but then on top of that I can't imagine we're going to instantly bounce back in May or June. Which is why it's hard to see GDP dropping by five, 6%. I think GDP in quarter two is going to drop, at least if it's properly measured, by 10, 15, maybe even 20%. The caveat that is on the measurement because a number of sectors in the U.S. it's hard to measure GDP and they typically measure on input.

Bloom: So for example, the public sector, universities, they measure it based on expenditure. For example, salaries and how much these sectors are spending. And of course they're still spending the same amount because people are still employed. But I don't know what other faculty think, but I think my output in terms of teaching research, annual government officials output is substantially lower than it normally is. But of course we're apparently as productive as we ever were by the official statistics.

Bloom: And I think how this is going to show off is the rebound is going to be really slow because there's a lot of stuff that's being shelved right now. A good example would be R and D, so all of the labs across a Stanford is shut down. Unless you're working on COVID, your lab has been closed. It's the same across MIT and other elite universities of people I've talked to. The same would be true in firms. The local NASA, there's a big NASA facility nearby here, and of course if all of that shut down, there's a huge drop in intangible investment that in the U.S. is not now taking place and it's going to really hamper the recovery.

Beckworth: Yeah, we're going to come back to this later in terms of long run effects, but I imagine just starting up the lab again is a very tedious process because you got to get funding, you've got to get the right staff, you've got to get approval. We got boards sign off on projects so you can easily turn the switch to off, but you can't easily turn it to on.

Bloom: Yeah, exactly. You're exactly right. A friend of mine, she was telling me about some research involving animals that she was terrified before the shutdown about how would this continue, how would it restart? And there's years of work. If you think of things like the OncoMouse, some of this work, genetic work with animals, is very hard to restart. There's a number of long run projects in the material sciences that go on for long periods of time. I think the short run is extremely bleak. A figure I saw today was actually fascinating from Deutsche Bank pointed out that the entire gain in employment, the roughly 20 million jobs that are being, or 22 million jobs, I think it was that were being gained between July, 2009 and February, 2020 and the long expansion, the longest expansion, I think since records have begun will be entirely have been lost within four weeks. So by the end of next week, forecasts are we lost 22 million jobs.

Beckworth: How tragic. I mean we were just at the cusp of a truly full employed economy, a hot economy. We're so close to the finish line and then suddenly -

Bloom: I know, I know. 11 years of amazing job creation, through thick and thin, through everything, through the Euro crisis, through the debt, through the fiscal cliff, all of it, through the change in administration and suddenly in four weeks it's gone. And this is why, for what it's worth, this is not about financial advice, but I'm actually much more pessimistic than the stock market. The drop, I think, in quarter two is going to be pretty bad and I just don't see us so rapidly bouncing out. And the uncertainty measures are very high in part reflecting it. People are skeptical slash uncertain about what's going to happen and that itself feeds back into the process and makes it hard to rapidly recover. You need firms. In order to rapidly recover what you need is massive hiring by firms in quarter three 2020 and massive investment. And of course they're not likely to do that while uncertainty is very high, so uncertainty also retards the recovery.

In order to rapidly recover what you need is massive hiring by firms in quarter three 2020 and massive investment. And of course they're not likely to do that while uncertainty is very high.

Implications for the Economics Profession

Beckworth: So there was another paper that compliments or really underscores the points you're making here. It's the New York Fed Paper and James Stock along with Daniel Lewis and Karel Mertens are the authors. I think the latter two are New York Fed economists. But it's a paper and they've updated it, but it's titled “U.S. Economic Activity During the Early Weeks of the SARS-Cov-2 Outbreak” and they have a weekly economic indicator and they have a number of indicators. I'll just read them real quickly here. I'm looking at them. They have red book research, retail sales, they have Rasmussen Consumer Index, unemployment insurance claims, insured unemployment, American Staffing Association staffing index, federal withholding tax collections, raw steel production, fuel sales to end users, railroad traffic, electric utility output. And they some kind of, on a primary component or some kind of factor model, but it's a weekly series and they parameterize it so that it comes out on the annualized growth rate on an annualized basis. So it kind of reflects what GDP, so you can think in similar terms.

Beckworth: And as of the week ending April 4, they have almost a 9% decline on annualized basis. And that's April 4. If you look their series actually goes back to before the Great Recession, before 2008, 2007. And during that crisis, they show about 3% of a contraction at the bottom, at the worst point. And so we're well past that according to their data. And presumably it's going to get worse, because it's only through April 4th. So it kind of confirms your story, but it's another interesting use of data. And one silver lining, if there is one, and again I don't want to make light of all the suffering that's going on out there, but one at least silver lining for economists and for folks like us, it's great to see all this big data being used and that people like you and others are tapping all these data sources. And again, maybe one silver lining is we'll rely more on those types of data than some of the data we use for GDP surveys that take a long time to record. What are your thoughts on that?

Bloom: One is, I think, on the first point you're exactly right. It was interesting for unemployment. People have switched, I'm still discussing this for a while with Steve Davis. People have switched from the non-farm payroll, let's say labor Friday, looking at the unemployment figures from that to actually now looking at the weekly unemployment insurance claims and Steve was saying, well administrative data is just very quick to come out. It has all its flaws and issues, but it comes out extremely fast and so all the attention's focused to that. In a similar way when I've been measuring uncertainty, I wouldn't have traditionally focused so heavily on stock market volatility or newspapers because there's a range of other indicators, but things are moving so quickly you need to focus on what you have that's available real time.

Bloom: The other thought I have that I think is going to be important for the profession and why, actually, I think these forecasts are overly optimistic. And I've been kind of looking through Blue Chip forecasts, they have about 30 different forecasters, is this is really kind of a play on the Lucas critique. So if you remember the old Lucas critique the turning point is when these non-structural models get things wrong. It feels like if ever there was a turning point, it was March, 2020, where you've had 11 years of expansion and suddenly completely abruptly, everything's changed. And I think a lot of the traditional forecasting models, the VAR types, the extrapolations or implicitly complicated trend fitting models, but they really don't work in periods like this. So if I'm forecasting in March 2020 I basically throw away all the prior data. Trends up to February 2020 are really uninformative.

Bloom: And at this point it's alternative data sources or looking at similar episodes like the Spanish flu of 1919, 1920, Hurricane Katrina ... A fascinating thing is Louisiana's labor market, the increase in unemployment is almost identical to the increase in unemployment in a couple of weeks after Hurricane Katrina. So it's more like a natural disaster than a regular recession. I think it's a challenge or call to arms for economists to find good, alternative, real time data sources and alternative forecasting methods, actually.

I think it's a challenge or call to arms for economists to find good, alternative, real time data sources and alternative forecasting methods.

Beckworth: Yeah. So this is going to be a very fruitful time for economist, but again, it comes at the expense of mass human suffering unfortunately. Well let's move to the long run effects. We've touched on the short run effects and just to summarize, bleak, not very optimistic. This is going to be a rough year. So much for the hopes of 2020 being a great year, maybe in an election year, maybe ... You know, one thing I was looking forward to this year was the Fed's big review coming out in June. Will we moved to average inflation targeting. Of course, I'm in the back row raising my hand, how about nominal GDP targeting?

Beckworth: But all that seems so quaint, so far removed from the present challenges we face and then moving forward, long run, there's going to be even bigger ones now that we're going through this. So we have a sharp contraction this year and we're going to have some long lasting effects. And we've talked about this offline before and I want to run through a list. So you made a list. I'm going to run through and I've added a few and I want you just to respond to them as I go through them. So the first one I want to bring up is just people and businesses, households and businesses may have permanently higher risk aversion because of this crisis.

Long-run Impacts: Increased Risk Aversion

Bloom: Yes. I noticed even just today, in fact, there was a piece in the Wall Street Journal discussing a paper by Stefan Nagel and Ulrike Malmendier, I think it's called Depression Babies, but they have a number of papers and other people around this arguing that individuals that live through major shocks are just more risk averse running forwards. And of course we've all lived through this shock. We will have our implicit distribution or function of outcomes based on what we've seen. And suddenly we've all been hit by left tail outcome. Another paper that's related to this, it's interesting, I teach is, Antoinette Schoar and Marianne Bertrand's paper called “Managing With Style”, back in the QJE in 2003. And that looks at the behavior of CEO's, and it argues and finds evidence that CEO's in large firms that lived through the Great Depression, so the oldest CEO's, tended to be much more risk averse, have lower debt levels. I think there's plenty of evidence that this is going to increase the risk aversion parameter for all of us for a while. It does mean, by the way, that running forward, for the next few years, the equity risk premium should be higher. So oddly enough, the expected return from the stock market for the next few years, once things settled down, should actually be higher, relative to interest rates versus normally, because of course the risk aversion parameter is higher and risk, at least, is probably at least as high as it was before.

I think there's plenty of evidence that this is going to increase the risk aversion parameter for all of us for a while.

Beckworth: The implication of this is that even if we get through this pandemic and there's a vaccine, is that households and businesses may be very leery to start new endeavors. So households may be cautious, they may save more than they normally did. Businesses will be slow to invest and do more capital formation. So even if things on paper should move back online, they may not very easily.

Bloom: Yes. Knowing the net impact of that, I totally agree. Deciphering the net impact of that is tricky because on the one side there should be a global increase in the savings rate. And in the U.S. there was after the Great Depression. So Americans, Europeans, Asians should all be increasing their level of savings and you'd think that would increase the level of investment, but as you point out on the other side, businesses themselves may be more risk adverse. And which effect dominates it is hard to tell. But certainly you may think another twist in this may be longer run, more risky investments. For example, if you think of startups or in R&D, maybe reduced in favor of shorter run, less risky investments. And in fact, another stylized fact that has just come out is if you look at the number of business startups in the U.S., it's fallen by 40%. So you can record that by the number of EINs, employment identification numbers that the IRS has requested to handout to new businesses, and that's down versus the same month last year by 40%. So startup rates have absolutely flatlined, which is A, not surprising but B, is a strong reflection that a lot of people are very risk averse and don't want to start new businesses.

Long-run Impacts: Decline in Trend Productivity Growth 

Beckworth: Second potential long run effect is a decline in trend productivity growth. Total factor productivity, you argue may decline. And a lot of this has to do with the lockdown. So we are now away from our places of employment. If you're at Stanford and you're doing research, you're not at your lab anymore. For you it may not be interacting with your colleagues, maybe in a different environment. Now what's interesting is you did research earlier in your career where you showed there were some productivity gains from working at home, but not this kind. So why don't you talk about your earlier research and why it's different now.

Bloom: Sure. So it actually relates to two very different papers. I never thought that they'd intersect. But one is I have something, it coincidentally is just out in the AER last month in March, 2020 with Chad Jones, Mike Webb and John Van Reenen, just pointing out the productivity growth in the U.S. and Europe has actually been declining in a pretty secular level since the '50s. So it's called “Are Ideas Getting Harder to Find?” And already we were on a gradual downward trend in productivity growth that's been going on. It's basically harder and harder and it requires more and more R&D expenditure to get the same increase in productivity as it did historically. So that's a longer run trend. What's thrown on top of that that's tricky is right now we're all working from home and I'm skeptical about the productivity impact of mass working from home.

Already we were on a gradual downward trend in productivity growth that's been going on. It's basically harder and harder and it requires more and more R&D expenditure to get the same increase in productivity as it did historically.

Bloom: So you asked me about an earlier paper I did. So I ran a somewhat unusual, a randomized controlled trial, I'm working from home, in China in about 2012-13. The backdrop to this was a grad student of mine, a Chinese PhD student turned out to be, strangely enough, this founding CEO and current chairman of the large Chinese startup called Ctrip. So Ctrip is like China's version of Expedia. It's worth about $10 billion on the NASDAQ. It's a really large company at about 15,000 employees. And their headquarters was in Shanghai and they were interested in having more employees work from home to save on office space. So we managed to persuade them to set it up as an experiment. And they took a group of 500 volunteers that asked to work from home. They randomized them by even an odd birthdays. You're one if you're odds. If you're born on the second, fourth, sixth, eighth, 10th of the month you got to work from home for four days a week for nine months. And then they just evaluated them.

Bloom: And they found that productivity of people working from home is up 13%, which is huge. That's more or less an extra day a week. The reason being the home workers said it was quieter and they saved time. They weren't late from commuting or having to take long lunch breaks, et cetera. And then the other factor was quit rates dropped by about 50%. So that study seems like I'd be really positive on working from home, but the COVID crisis is really different. There's a number of problems. One is we have our kids. So my four-year-old regularly bursts into my office. I actually had to lock my office to prevent her bursting in. But she bursts in looking to play. From everyone I'm speaking to, this is such a common problem about having kids home.

Bloom: We're five days a week. So Ctrip only did it four days a week because it's actually really hard to be creative and remain motivated if you're permanently away from the office. Many of us have totally inappropriate working spaces. We're sharing offices or we're working from... I just had a video call with one of my students that she took it from her closet since she shares a bedroom with her boyfriend and she has no other space. And then finally there's a lot of an issue about choice. So in the Ctrip experiment, we deliberately chose jobs that people could do at home. They're answering calls. And gave them the choice, and only half of them actually opted to work from home.

Bloom: In the U.S. there's a number of studies and one I saw a recently I thought was really great, by Jonathan Dingel and Brent Neiman at Chicago. They looked at what share of U.S. jobs currently out there could actually function well at home and discovered that came to around one-third of employment and roughly 50% of the wage bill. So basically higher paid people tend to more effectively be able to work from home. That tells you there's a lot of economic activity, much of retail, a lot of services, being a dentist, being a pilot are activities you really can't do effectively at home. Or as you mentioned earlier, R&D lab scientists, you just can't access your lab equipment. So I think right now there's a massive reduction in productivity of even those of us that are able to theoretically work at home, our actual true productivity, I'm sure, is way below what it would otherwise be.

Higher paid people tend to more effectively be able to work from home. That tells you there's a lot of economic activity, much of retail, a lot of services, being a dentist, being a pilot are activities you really can't do effectively at home.

Beckworth: So that means going forward, less capital formation and lower economic growth.

Bloom: I was going to say, the interesting thing, it may not actually show up in GDP this quarter. So both of us are working in the not-for-profit sector and we're typically measured as expenditure for GDP. So apparently our GDP is unchanged. But the way it was sharp is, of course, we and large chunks of the economy are not going to be doing as much training and we're not going to be creating as many products and innovations. For example, I bet patenting in 2021 and 2022 will be well down because labs are all closed down. And so this is really going to... And investment, physical investment would be down. So this is really going to hold back the rebound that we're looking for.

Beckworth: So lower productivity growth. Now do you see this as a decline in trend growth or maybe a one-time blip?

Bloom: It's a good question. Productivity now in 2020, so if you take a classical solo model, output equals K to the alpha, times A, L to the one minus alpha, it's hard to know how we'll eventually decompose the impact, but a lot of it must be from A. It's really inefficient working from home. It's really inefficient doing stuff by Zoom. But as you imply it, I think quite a chunk of that will be persistent. So three years from now most of it will have blown through, but for example, all the lab closing. There's a long debate about what's driven the American success story. How did America go from basically similar levels of GDP or below Europe 120 years ago to far above it? And one of the big stories is just being very focused on innovation. And I think that innovation engine is almost completely stalled right now and we're going to see lower growth and lower productivity growth, in fact, in the next two or three years from now.

Beckworth: And the point you raised earlier is a good one that this decline in any other time would be less consequential because now it takes more effort and more ideas for that next dollar of economic growth. So this is the worst time we could have had an experience like this when it's harder to generate new ideas and new economic growth. Something else this raises though in my mind is both this point with there'll be lower productivity growth, but also the first point, permanently higher risk aversion. It's going to mean that we're going to be stuck at zero interest rates for a long time. Both of those factors are going to weigh down on rates so it's going to complicate life for the Fed, even beyond the many complications that now faces. Monetary policy around the world is going to be one that operates with interest rates at zero or slightly negative numbers going forward. So I think there's all kinds of effects here we haven't begun to process. We've just looked at two of them so far.

Bloom: The Fed, as you know, I religiously listen to all of your podcasts.

Beckworth: Well, thank you.

Bloom: There's a frequent discussion of the Fed and I think as previous guests have mentioned it, it's an incredible time for the Fed and its normal tools, the interest rates, they're jammed a zero. There is the huge risk of politicization of the Fed as it becomes more and more involved. Another related concern I have for longer run growth is the expanded share of government in the economy. Another factor that's really driven U.S. success versus Europe is Americans don't like government. So Americans, traditionally, have done everything they can to cut back the role of government. It goes back to the spirit of the founding fathers, to be honest, to escape the clutches of government. And we've seen a huge expansion through expenditure and through controls, and I think that's going to actually hold back growth. The Fed is doing, right now, and I don't think I would have done anything different, but the problem is a very strong medicine, it comes with very strong side effects and I think this is... The very strong medicine of the Fed of massive QE and slashing of rates and extension of loan programs has all kinds of programmatic side effects.

Another related concern I have for longer run growth is the expanded share of government in the economy.

Long-run Impacts: Decline of Globalization 

Beckworth: So in terms of long term effects, we've discussed potentially higher risk aversion, lower productivity growth, and another point you make is we may see a sustained drop in globalization. So drop in trade and immigration. So walk us through that argument.

Bloom: Again, since World War Two really, if you look at what's a big driver of growth, it's been an increase in trade. You notice going back, it now seems like ancient history, but the market was bouncing up and down, driven by every tweet war that Trump had over trading with China. So every time it looked like trade with China is about to slow down, the market dropped. And the reason is trade is important for growth. It's actually interesting and particularly important for some of the largest publicly listed American firms that are particularly international. So I don't see how we're going to rapidly get back to the levels of globalization we had before the crisis. It's easy to put in trade barriers, it's much harder to remove them.

I don't see how we're going to rapidly get back to the levels of globalization we had before the crisis. It's easy to put in trade barriers, it's much harder to remove them.

Bloom: And then on the other side, immigration, again, the U.S. and particularly Silicon Valley and some of the science and engineering parts of America that drive a lot of overall growth are very immigrant dependent. So I think the figure is within Silicon Valley more than half of the PhDs are immigrants. I would be an example. I'm not a scientist, but I'm an immigrant. And the COVID crisis has probably almost completely stalled immigration, which is in some odd sense is, to the gain of the rest of the world that was leaking this massive brain drain to the U.S., but it's going to be a huge cost to America. That it's just not going to have the flow of top end immigrants. Bill Kerr has numbers looking, as you go higher and higher within science and academia and research, you start to see a higher and higher share of immigrants. So I think in the U.S. something like half of U.S. Nobel prize winners, American located are also immigrants. So that end is going to be really hit.

Beckworth: So this reinforces your earlier point about the decline in productivity. You'd take away the future brain power from immigrants, total factor productivity will be going down even more. So just another strike against U.S. growth in the future and, again, reinforces my point about zero lower bound. This is not looking nice at all, Nick.

Bloom: I hate being a... I was talking to some journalists. They said now I know why you economists, it's nicknamed the dismal science. It's like all I can do is suggest sell your shares. I wouldn't even... putting them in bonds is... I don't know what to do. One way it could manifest itself would be we have some form of snapback but to a lower trend for the next 10 or 20 years. Productivity has been declining. You could see that trend rate of productivity growth drops and remains lower for much longer than were predicted. A lot of these factors are slower burn, so reducing immigration, reducing innovation, stopping trade. I mean another one we should come to discuss is tax rates, higher tax rates. All of these things tend not to be so immediate, but if you accumulate them over five, 10 years, they do hold back growth rates. And if you look at the things that America was lauded from as driving a success, innovation, trade, immigration, low taxes, light government, almost all of these seem to be reversed in the current COVID crisis and the response to it.

Beckworth: Yeah, absolutely. So maybe the best way to paint all these bleak points as we're creating a baseline, so that anything better than this will be amazing.

Bloom: Yes yes, the only way is up.

Beckworth: Exactly. Set your expectations appropriately, then you'll be pleasantly surprised when things are better. All right, so you mentioned taxes. Taxes are going to have to go up in the future. There's going to be some form to accommodate the bigger footprint of the US government. Although I would mention now we still don't see any increase in the treasury bond rates. So maybe the issuance of debt has just been proportionately met by an increase in risk aversion, but we don't know that will hold after everything calms down.

Bloom: Yeah, I mean if you look at total government debt is about 25 trillion right now, of which about 22 is federal, and about 1.5 is state and 1.5 is local, municipal. So you've got about $25 trillion worth of debt in the US, that's roughly the same as GDP. So roughly speaking we're about debt to GDP, of government debt of around 100%. Forecasts are that's going to go to something like 100 to 130 to 140%. That's going to overtake the level we saw at the end of World War II, which is about 110%. But then after World War II, if you look back at tax rates, tax rates surged from the '50s onwards. So the top marginal tax rate, it was 90% throughout much of the 50s. So there're two ways you can repay back debt is high taxes or inflation. And assuming ... If you think inflation is the less likely outcome, then higher taxes seems the more obvious course. And I presume it will be income taxes will be a chunk of it. Some of it would be corporate income taxes.

Bloom: So I see, given the trends in US politics with ... They didn't get elected, but obviously Sanders and Warren were popular in the democratic side of things. I wouldn't be surprised if there's an increase in taxes and there's pushes towards the top end of the income distribution.

Beckworth: So Nick, do you think I'm being too optimistic that the increased risk aversion will bear some of that additional debt that we issued, so it won't be as bad as it otherwise would be?

Bloom: Yeah no I think, I mean, interest rates are not going up for, as you say, the market is just absorbing, seems to be able to absorb an unlimited amount of-

Beckworth: For now.

Bloom: ... fiscal debt. For now. At some point that stuff matures. So even if we think the market absorbs it, we have to pay it back. I mean, I guess you could roll it over indefinitely. The problem with the rolling it over indefinitely strategy is what's happening to Italy and southern Europe right now. So the US could say, "Let's just keep debt. The GDP ratio. Government debt to GDP 150% indefinitely." And if interest rates remain at 1%, that's 1.5 percent of GDP goes in debt servicing, that's not too terrible. The problem would be if we get hit by another crisis. So what's happening right now in southern Europe is they'd never really paid off the debt from the financial crisis, the European crisis. So that now, whereas the US is entering this thing with debt levels that are around 100% of GDP. For example, Italy right now has debt, public debt to GDP of 140%, so it has very little wiggle room. So the US either just has no spare cash left, or it has to pay it down. And personally I'd probably rather pay it down but that of course implies higher tax for it.

Beckworth: Yep. Okay. Let's move on to the next item you talk about. You mentioned the impact of the pandemic will not be evenly distributed. So talk about that.

Who Bears the Brunt of these Impacts?

Bloom: Yeah, I mean another unpleasant factor, the pandemic, I mean as part of a broader theme honestly going back for 30 years is, it appears to be increasing inequality. So the pandemic, maybe it's just kind of more obvious and apparent, but the people who've lost in the short run are much more likely to be hourly pay contractors or self-employed. So large numbers of those people who've lost jobs more proportionately than we see in the data from people who are salaried. And if you look historically at salaried employees, if you look at prior recessions, you can do something called estimate a wage beta, which is how sensitive your wages are to the economic cycle. Again, we've done that. I've done this in the UK and Matthew, for example has done this in the US using administrative data. And you see that typically people, younger people, in fact younger men working in small firms turn out to be the most sensitive historically to the economic cycle.

The people who've lost in the short run are much more likely to be hourly pay contractors or self-employed.

Bloom: There's a bunch of factors why that, and it's not so clear this recession would be so actually anti-male, but it's certainly likely to hit, amongst salaried employees, younger men and much more strongly. So those with part time or hourly pay. And they tend to also be the lowest paid. So, in terms of the loss of earnings and the GDP cost, I think it would disproportionately fall on lower income people in society.

Beckworth: Yes, it's not easy being poor in America, that's for sure. The system-

Bloom: No, and it's harder now. It just ... I mean on the other side it's even ... If you look at some of the articles they've used cell phone data to track who's self-isolating, you notice that people at the wealthy end, so the top 10% by income have self-isolated much more effectively than the bottom 10%, in part because they can. In part because if you're a salaried employee that can work from home, you can do that. So this I think is important politically because it means ... I mean there's a welfare side of it obviously about increases in equality. I think it's politically important for their recovery, because it suggests that in terms of A, politics becomes very unpredictable. If you have a set of people that have lost very heavily, you're not surprised they're going to be extremely unhappy with where things are, and you have more radical politicians potentially coming to the front.

Bloom: But the other issue is just in terms of where taxes are likely to fall, I think it suggests they're going to be falling on wealthier parts of society, which would be probably corporate income taxes and top-end income taxes, and possibly even wealth taxes. But Sanders and Warren were pushing for wealth taxes. So I don't see that as impossible as a way to pay down some of the COVID depression debt.

Beckworth: I live in Nashville, and technically I'm on the outskirts of Nashville, so I have one foot in the world that both of us exist in. Economists, research, the urban setting. And I have one foot in the more rural setting, and it's striking to see the difference. It's been there, as you mentioned, for many years, but in the present situation I feel so blessed, so fortunate to have the job that I do have, as you mentioned, I've got a salaried job. It's been, if anything, an opportunity for me. I hate to say this, but this crisis is great for people like us who write papers and discuss ideas in how to solve problems. But for many of my friends and my community here, it's been a horrible time, horrible experience. And one practical side is I really drilled this idea into my kid's head, that this education's important, and you can see the difference. We're just very blessed, very fortunate to be on this side, but many are not. And moving forward, this divide's probably only going to grow as the year continues to deteriorate.

Bloom: Yeah, I think so. One of the economic questions, I was actually talking to pretty senior policy official about one of the challenges about trying to figure out when to end the lockdown or resume to work. But if you think about who's taking the choices right now, it tends to be older salaried people. So everyone that's running companies, that's running the government, all the politicians, are salaried, so their losses are typically lower. And they're older so therefore their health risks typically are higher. On the other hand a number of the people that are worse affected tend to be younger. So I think there's a political tension between the people calling the shots and what ... The most likely outcome is I think is we go back and we have rolling waves of clampdown and recovery and clampdown and recovery.

Bloom: I should say, on another note on productivity, I don't know if you can hear in the background, but one of my kids is ... My wife is a Scot, and one of our kids is learning to play the bagpipes. And so working from home, there's an enormous amount of noise now pouring through the wall.

Beckworth: No I don't hear anything, you're good. But that's great to know.

Bloom: Okay. Tells you everything about working from home with your kids around, it's hard to be that fully productive.

Beckworth: Well, we like to keep it real on Macro Musings, so that's a great story to add to the show. Perfect. Okay. One last point I thought I would bring up, and this is something you didn't raise in your presentation, but we've discussed it before, and that is urban living. What will this mean for urban living? Will places like New York City be less attractive now that we know one, you can ... at least some people can work from home, do meetings via Zoom, but also maybe there's again more risk aversion. Maybe you don't want to live in a dense urban area. So do you think this effect will be big or just minor, or will there be meaningful changes going forward?

Long-run Impact on Urban Living

Bloom: Yeah, I think this is the end of a long trend. So I was at the Ely Lectures and the AEA a couple of years ago, and David Autor was talking about income and population density. And he was pointing out that basically from the '60s to the '80s, the center of cities did really badly. So they did increasingly badly. So what happened in the '60s onwards is, or really the '50s, there was a rollout of the freeways and roads and cars became a lot cheaper, and a lot of middle class people moved out to the suburbs. And by the early '80s the center of cities become incredibly low income, with the crack epidemic and really some of the worst blighted parts of the country with incredibly low property prices. And the '80s was really the turning point, until now, so the last almost 50 years, or 40 years. We've seen increasing valuation of the center of cities as more and more young and high school people have moved in.

Bloom: My guess is 2020 maybe another turning point, in A, the move towards working from home, not full time but probably part-time. And B, honestly, fears of social contact may lead people to start to move out again from the center of cities. And so we could see the '80 to 2020 trend of increasing value. So now if you, by the way, if you look at charts, so he showed two charts. One is the value of property versus density. And in 1980 the cheapest property was in the most dense areas, city centers, which were absolutely blighted. They were like gang land areas you'd never want to step foot in. And then by 2020 they were the most expensive real estate in the US. That premium, I suspect, is going to rapidly start shrinking back down again.

Beckworth: Yeah. It will be interesting to see what happens to the areas like San Francisco close to you, New York City, where it's been very difficult for people to move in there because of the high real estate prices.

Bloom: Yeah, I mean that's a very ... I was talking to someone recently, a policymaker locally. To end on a positive note, that's actually a much more positive take on it. So at least around us there has been a big fear of affordability. And Silicon Valley, there in particular, has this crisis of no one in a normal job can basically afford to rent an apartment around here or to live here. And that at least is going to be one of the benefits of this. I think there'll be a lot of medium and higher income people, take folks like us that may basically end up commuting into work two, three days a week. And if you're doing that, of course you half your commute frequency, you may think about adding an extra 50% distance. And so I wouldn't be surprised if the center of cities empty out a bit, and therefore it reduces the cost of living there.

I wouldn't be surprised if the center of cities empty out a bit, and therefore it reduces the cost of living there.

Beckworth: Which is interesting because I have colleagues, you have colleagues too that have been working really hard on the fight against nimbyism, and allowing more urban living, building higher, building more dense living. And that, I don't know, that struggle might not be as important now. But on the other hand, again, the developments we just talked about may lead to deal with the same outcome that they were fighting for.

Bloom: Yeah. I mean so much stuff has been changed. The environmental movement. A another short run at least, a silver lining from the COVID crisis has been of course the reduction in pollution. The affordability ... I don't know how much ... I assume what will come out will be ... There's no ... So a lot of young people want to live in areas with good bars, restaurants, amenities. Rebecca Diamond's stuff on the fact that the center of cities is this self-fulfilling outcome. Young people move in, therefore bars and restaurants come, therefore it becomes more appealing and more people move in. You may find that instead concentrates on certain suburbs. So if you go to cities like LA, in LA there are suburbs that are just known to have good bars and restaurants, but they're not all in the center of town. So I wouldn't be surprised if LA or Houston is more like the template for much of America whereby you have localized centers, but nowhere gets too expensive.

Beckworth: Okay. Well on that note, our time is up. Our guest today has been Nicholas Bloom. Nick, thanks so much for coming back on the show.

Photo by Pixabay

About Macro Musings

Hosted by Senior Research Fellow David Beckworth, the Macro Musings podcast pulls back the curtain on the important macroeconomic issues of the past, present, and future.