Did Remote Work Drive the Global Surge in Job Openings?

Like inflation, the historic shifting out of the Beveridge Curve was experienced across countries. One speculative reason might be the rapid expansion of remote work.

The 2025 Economic Report of the President (ERP), the last one under President Biden, came out right before the end of his administration. It’s a fun document.¹ At the National Economic Council I got to make a small contribution, giving some comments and edits to try and get my grubby politico fingerprints all over it help advise, monitor, coordinate, and establish consistency in the policymaking process.

There’s a chapter on remote work that is particularly useful at this moment. It reminds me of an interesting question: was the global rise in job vacancies and the shift in the Beveridge Curve driven by remote work?

Let’s step back. You may have seen this graphic of inflation and growth among advanced economies following the pandemic. It shows that while the United States’ inflation was experienced across peer countries, economic growth was not.

The election is over, so I can tell you the truth. As someone who has really sweated the details of this graphic (reproduced it, written the talking points anticipating the incoming we’d get on it), the fact is: it’s correct. Inflation really was similar across peer countries. The United States had the inflation wave first (we were also the first to vaccinate), but within two quarters the trend was everywhere. Even countries not in Europe as well as countries not particularly exposed to the energy crisis following Russia’s invasion of Ukraine, like Australia and Canada, saw the same inflation.

Beveridge Global Space

What I think is less understood is that the same thing happened to the most interesting dynamic in the post-pandemic labor market: the historic rise in job openings. Or, more technically, the shifting out of the Beveridge Curve, which is job openings given a level of unemployment. This is a controversial tool labor economists use to analyze the strength and efficiency of the labor market.

Here is the Beveridge Curve for the United States, a chart of job openings/vacancies versus the unemployment rate.

In case you haven’t followed, that the blue line extended so far - that there were a lot more job openings than you’d expect given a level of unemployment - was a central puzzle of the recovery. Most assumed there was something wrong with the United States in particular. Too much demand from the American Rescue Plan, a “Great Resignation” deriving from US lifestyles and workplace cultures, and so on.

But below are numerous Beveridge Curves from other countries, copy and pasted from multiple sources:

Sources: EU/Euro, Canada, Australia, UK

As you can see, in Australia, Canada, the EU/Euro area, and U.K., all of their Beveridge Curves had the same shift out and then (most of these graphics are from mid-2024) steep drop down. Just like the United States.

I’ve been following the inflation debates since 2021 closely, but I wouldn’t have caught this early if it wasn’t for EconTwitter. One day the Canadian economist Brendon Bernard (good follow) was posting about new data showing how far the Beveridge Curve had shifted and all the debates about why, and I thought “but it’s not a JOLTS data day.” It wasn’t—he was having the exact same debate we were having in the United States!

It’s actually quite odd. The US debates over the Beveridge Curve have been extensive but they don’t generally mention other countries. And, on a quick skim of summaries from other countries, they may mention the US but don’t cite e.g. (Figura and Waller 2022) versus (Blanchard, Domash, and Summers 2022) about the “steepness” of the Beveridge Curve either.

Also - and this may be wrong for some specifics - googling around the labor market story looks similar across peer countries. Real wages and the labor share fall during the peak of inflation. But there’s higher wages at the bottom, leading to income compression. Real wages have picked up faster in the United States as growth has as well.

One Common Thing - Remote Work

What could cause all these curves to move in tandem? The most obvious answer is COVID. COVID shook up labor markets with shifting spending patterns - so lots of job switching - and the risks of catching COVID as a job risk to be compensated.

But I also wonder how much the adoption of remote work played a role. This was also a global phenomenon:

Source: (Hansen et al 2023)

And that brings us to the remote work chapter of the ERP. Below is my replication of their Figure 2-13, which itself is replicated from Job Amenity Shocks and Labor Reallocation by Sadhika Bagga, Lukas Mann, Aysegül Sahin, and Giovanni L. Violante (Bagga et al 2024), a great paper which provides a more formal and in-depth theory of this argument.²

Let’s take this slowly. Openings, quits, and hiring all increased during the immediate recovery. We want some measure of relative gains, so here we use hiring over openings. We take the change in that ratio from Jan 2020 to March 2022 (when job openings peak). A negative number means that openings increased faster than hiring. We then compare that to the share of people who telework by industry.³

Here we see a statistically significant relationship - industries where there was less remote work saw openings increase faster than hiring. As the ERP finds, after 2022 this relationship is flat. There was one giant reshuffling of the labor market that has now ended. I could do this since I know the US data well; I hope researchers from other countries can give this a try and share.

Takeaway

It’s reasonable that this would have a massive labor market impact. As the ERP notes, 20 percent of the American workforce working partially or fully remotely is “double that of workers represented by unions and about the same share of the workforce with an occupational license, two groups deservedly receiving considerable research focus.”

Inequality did decrease in the past recovery. Strong labor market demand played an important role as (Autor, Dube, and McGree 2023) find. Remote work is more likely to be done by white collar and college educated workers, and their wages did not increase as fast as those at the bottom. There’s an open question over whether or not their real wages falling/slowing was instead compensated in part by being able to remote work. People, in surveys, say that they are willing to take pay cuts to work remotely. Though I hear from researchers that remote work tends to get swamped by education terms in microdata level regressions, it’s still an open and interesting question.

Note that the argument that remote work is unfair as it disproportionately benefits better off workers may be already compensated and priced in by the labor market search process. Reworking an important part of compensation for 20 percent of the labor market is going to shake up our economic statistics. But yanking it out - as many CEOs and the Trump administration want to do - would upend a lot of the adjustment the labor market has already made.

The early, preliminary results on productivity (via a helpful Nominal News summary) look neutral to good - indeed productivity is almost where pre-pandemic CBO forecasts had predicted, a wild result given the chaos of the past years and the major shift in remote work. Also, something I want to visit more, there is the possibility that it boosted birth rates during the reopening, with a preliminary case made by Patrick T. Brown here. What’s not to like?

1. One thing I like about the ERP is that it’s a great historical document. If you wanted to know the best arguments that administrations made for themselves in real time, as well as the things they wanted to flag as important and the justifications they made, old ERPs are worth visiting.

2. One of the researchers is Sadhika Bagga, who had a fellowship at Columbia that worked with the Roosevelt Institute’s Macro Team I had led. It’s great to see this paper out - I think it should generate a lot of interest. Other researchers on that fellowship include Joana Duran-Franch, one of the first to catch how quickly wages were growing at the bottom of the income distribution, and Justin Bloesch, whose helped call the soft landing and whose latest paper on the labor market post-COVID can be seen at PIIE here. A great crew all around!

3. Starting in October 2022 the CPS now asks “At any time LAST WEEK, did you telework or work at home for pay?” and aggregates the data in a non-seasonally adjusted way here. I use the 2023 average since I can yank that from the BLS flatfiles; CEA and Bagga et al each use other measures for remote work.