September 2021

This month, our podcast features a conversation with Leo Feler and Ioana Marinescu regarding research on how enhanced unemployment insurance affected job search and employment during the COVID-19 pandemic. Below is an edited transcript of their conversation.

Leo Feler: At the beginning of the pandemic, we saw unemployment increase to almost 15% and a loss of about 20 million payroll jobs. Now, 18 months later, we’re hearing anecdotes and seeing data about labor shortages. You wrote a paper with Daphne Skandalis and Daniel Zhao looking at how unemployment insurance affected job search and vacancy creation, and I’d like to ask you some questions about that. What was the effect of the enhanced unemployment insurance at the beginning of the pandemic? Did the enhanced unemployment insurance mean that people searched less for jobs?

Ioana Marinescu: If you look at economic theory, it predicts that when we make unemployment insurance more generous, people are going to search less hard. This was the premise that we started with when doing this study. Then we pulled data from online job searches, from a website called Glassdoor, and looked at job applications. What we saw is that when benefits increased more, this is when people applied less. So, essentially, an increase in the level of benefits – remember that benefits were increased by $600 a week, which was an enormous, never-seen increase – when benefits increased, we did see a decrease in individual job applications.

Leo Feler: How big was this impact? By how much did it reduce job search?

Ioana Marinescu: Our estimate is that if you increase the value of unemployment insurance by 100%, this would decrease applications by 36%. It’s not one-for-one, but a 100% increase in benefits, so that would be a doubling, that would lead to a 36% decrease in job applications.

Leo Feler: You find something interesting for the early months of the pandemic as many states lifted restrictions. Although you find that the enhanced unemployment insurance reduced job search, you find that it didn’t really reduce employment. Why is that?

Ioana Marinescu: We don’t look at employment directly, but other people have looked at that and found no effect. What our paper does is answer the puzzle of why on the one hand you would expect more generous unemployment insurance to reduce peoples’ job search efforts – then shouldn’t you see less employment from this? That’s really the puzzle that we set out to explain here. The way that we reconciled this is by looking at the market level. What you have to understand is that this is not just an issue of individual job seekers deciding to apply less, but rather you have to look at the whole market for a different job. A big thing that happened during the crisis is that the number of jobs tanked big time when COVID hit, and as a result, the big event was that there was a huge decline in the number of jobs. At the same time, you had a huge increase in the number of the unemployed, and each one was applying a little bit less due to unemployment insurance. But, in the end, the decline in the number of jobs was the dominating factor. On average, you would see more applications per job during the pandemic than before. The big factor here is that there were so few jobs, so many unemployed, and that’s really the big factor. And yes, without unemployment insurance we would see even more applications for jobs, but we were already in a glut due to the imbalance of jobs and job-seekers in the labor market.

Leo Feler: How do you differentiate between people not searching for jobs because they’re getting more generous unemployment insurance, and people not searching for jobs because we are in the middle of the pandemic and they don’t want the exposure? Or because their kids are home from school, so they need to be homeschooling, rather than searching for jobs? How do you differentiate the effect of the increased benefits themselves, from everything else that’s going on in the economy?

Ioana Marinescu: That’s always the tricky million-dollar question when we do such research. The way we do it is by comparing people who saw different relative increases in benefits. Even though everybody had a benefit increase of $600, for people who on a regular basis would have been getting $200 in benefits per week, that’s a huge increase. But for people who were already getting $600, it’s a smaller increase. This is what we use to understand the results: comparing people who normally would be getting a low benefit amount versus those who would be getting a high benefit amount. Those who were already getting a lot of money from regular unemployment insurance, the $600 was a smaller increase than for those who were getting very little money from unemployment insurance and the $600 represented a huge percent increase. Therefore, we are seeing that those who were getting low benefit amounts, and therefore a bigger relative increase in benefits – those are the people who cut their applications relative to others who were already getting high amounts of benefits from the regular unemployment benefits, and even though they had $600 extra, that represented a smaller percent increase for them.

Leo Feler: I think it’s also important to think about the time period for this work. What is the time period you’re analyzing?

Ioana Marinescu: It’s important to note that the dates we’re looking at were between April and July 2020. This is right when the pandemic was starting and through the summer of 2020, when we saw the economy reopening. What’s quite interesting is when we look at the effects over time, and we see that even when the economy was reopening and there were more jobs – even then, it turned out that the effect of unemployment insurance was not big enough to make it really hard for employers to hire. The jobs had recovered some, but not completely, even by the summer. Again, remember, there were lots of unemployed people. One big factor you always have to keep in mind is it’s not always about the behavior of every individual job seeker, but also how many job seekers are searching at the same time. When you have a lot of unemployed, it’s going to be easier for employers to fill jobs, because there are more candidates. Even if each one might be sending a couple applications fewer than they normally would, collectively they are still sending a lot of applications.

Leo Feler: Is there a point at which the effect of having fewer job applications actually begins to reduce employment creation? Early in the pandemic, there were so many job seekers for so few jobs, that having fewer job seekers didn’t make that much of an effect on employment. But over time we’ve actually seen that the number of job openings have increased. If we look at the data now, we have about 11 million job openings for 8 million people who are currently unemployed. At what point do we say maybe the unemployment benefits are reducing not just job searches but employment itself?

Ioana Marinescu: What we know is that the impact of the benefits on the employment is going to be more negative as we have more and more job openings relative to the number of job seekers. The number of job seekers hasn’t dramatically increased since last summer. You have the number of vacancies going up dramatically, and the number of job seekers somewhat constant or even declining. That means that what we call “tightness,” the number of vacancies relative to the number of job seekers is going up, and generally that makes it harder for employers to hire. Therefore, it’s more and more likely that the generosity of unemployment benefits can contribute to a lower level of job creation. But it’s hard to say for sure how big that effect is in the grand scheme of things given that there are many other drivers of behavior we can talk about. People have looked at when states cut benefits, and we can talk about those results. You would expect that unemployment benefits are going to have a more negative impact on unemployment in the recent months – during the spring and summer of 2021, than during the spring and summer of 2020, but it’s hard to pin down that number.

Leo Feler: You talk about labor market tightness in your paper, and I’m glad you brought up the term. And, you talk about an efficient or inefficient level of tightness. What does that mean? What does it mean if we have so many job seekers for so few jobs, or very few job seekers for a lot of job openings? What does that mean for the efficiency of being able to find good labor market matches?

Ioana Marinescu: In the labor market, you want to balance the incentives for each side of the market to search and form matches. Workers must be motivated enough to find a match, but firms must also be motivated enough to hire. There is an optimal level of tightness that is a middle ground between too high and too low. It’s not good to have a lot of job seekers chasing very few vacancies, or at the opposite extreme, having a ton of vacancies and there’s nobody to hire. That will reduce employment in the economy and be a bad thing. But the exact level that’s optimal is going to depend on specific modeling assumptions, a bunch of things we have to assume that cannot come directly from the data. You have to have a theory in the background plus the data, and then you can attempt to give some idea of the optimal level of tightness, but it’s not something that will come directly from data analysis itself.

Leo Feler: How should we think about the balance between the incentive effects – in this case job search – and the stimulus effects on the economy of the enhanced unemployment insurance, or the welfare effects for those who received unemployment insurance?

Ioana Marinescu: This is important to bring into the conversation. Yes, the benefits provide some disincentive for people to search for jobs, but at the same time they allow people to consume. That means that there is demand for companies’ products, and that ultimately ends up creating jobs. So, you have these two sides of the equation – that on the one hand, people searching less could lead to less employment (although we discussed that it really depends on how many vacancies there are). But on the other hand, people getting more money from the benefits could lead to higher employment through higher aggregate demand (the fact that people are spending, and therefore businesses have more customers, and want to hire more workers). What research has shown from prior recessions is that this aggregate demand effect is quite strong, especially during a recession. Meaning that, generally, if you give people money during a recession, they tend to use it. Firms use it, people use it, and that tends to generate more jobs. Through that other channel, the unemployment benefits are a positive in the economy. That again brings to the fore this idea of balancing. How do you balance multiple considerations to get to the right solution? I think part of it is going to be judgment and priorities from the point of view of policymakers. You take the best evidence out there, you understand the different sides of the question, and you make an informed judgment about how this is going to balance out, and also what are your priorities? Would you rather err on helping workers a little bit more, even though maybe there’s some chance that employment will be slowed down a bit? Is that the more important consideration, or are you more intensely worried about filling jobs quickly?

Leo Feler: So, we give more generous unemployment benefits, that stimulates the economy. That means that we have jobs being created and a lot of demand for workers, which means that we have more vacancies, and it’s harder to fill these vacancies because people are receiving unemployment insurance benefits. And maybe there’s a little bit of a disincentive effect. But we wouldn’t have had the higher vacancies from the job creation to begin with unless we’d had those more generous unemployment insurance benefits.

Ioana Marinescu: Exactly! That’s always why I tell business owners who say, “Oh this is so bad, the unemployment benefits,” and I say, “Yes, they create some disincentives, but without them, the economy would not be where it’s at today.” That’s really important to keep in mind.

Leo Feler: It’s very hard for policymakers to decide, what is the optimal level of benefits? When should the benefits be reduced? How long should the benefits go on? I wanted to get your thoughts about what we take away from this experience. We set this $600 enhanced unemployment insurance benefit early in the pandemic. There were moments where maybe it would get extended, maybe it would not get extended. You talked in your paper about how this might lead to some changes in job search behavior if people don’t know how long the benefits are going to last. A lot of economists have talked about automatic stabilizers, something that will trigger benefits to go up, or something that will trigger the benefits to start coming down. What are your thoughts on whether or not this would make it easier for policymakers to accomplish this delicate balance?

Ioana Marinescu: First, let me note that we already have some degree of automatic stabilizers with unemployment insurance because at the state level there’s a program called extended benefits that gives people additional weeks of unemployment benefits when the state unemployment rate exceeds certain thresholds. It’s already there to some extent for states. Historically, the federal government has tended to top-up and make unemployment insurance even more generous during recessions. The debate is: should we make this extra layer at the federal level more permanent and have it vary with economic conditions? I think our paper would support this, in the sense that we’re showing that being more generous when the economy is doing badly is not costly in terms of employment, because as we were saying before, even though there are some disincentives, in a bad crisis, there are just not enough jobs. So it’s okay if the many job seekers search a little bit less. I also think there are some political considerations. From the point of view of policymakers, they like to say, “There’s a crisis and we’re doing something.” If you make it too automatic, you’re removing from policymakers a tool that allows them to look good during a crisis. So I think, yes, in principle it would be nice to have more automatic aspects in the unemployment benefits system, but I think that may be against the incentives of policymakers who want to be able to have a bit more discretion and be able to claim that they’ve done something when there’s a crisis.

Leo Feler: Ioana, thank you so much for taking the time to talk to us about your research. The paper is called “Job Search, Job Posting and Unemployment Insurance During the COVID-19 Crisis.” Ioana, thank you. We really appreciate you talking to us on this topic.