One year since the pandemic unemployment insurance (UI) expiration, what were the outcomes, and what can we expect from future reforms?
Fiona Greig, the global head of investment research and policy at the Vanguard Group, and Peter Ganong, economics professor at the Harris School of Public Policy, joined Niskanen’s employment fellow Matt Darling and employment analyst Will Raderman to discuss UI’s impact and the future of UI reform in a recent Twitter Space.
Below are some highlights from the conversation, and the full Twitter Space recording for your perusal.
- Low-income families saw a sustained increase in the amount of money available in their checking accounts balance from the combination of pandemic relief programs, including economic impact payments, enhanced UI, and the expanded Child Tax Credit. As of March 2022, these households had about 70 percent more money in their checking account than pre-pandemic. Since the median household checking account in this category typically had around $1,000, this is equivalent to an increase of $1,700.
- The evidence does not suggest these programs substantially reduced employment rates. For example, when 26 states ended their Pandemic Unemployment Compensation program, those states did not see increased rates of employment relative to those that continued their programs.
- While other studies have subsequently found different results, a unique advantage of Ganong et al. (2022) is that they pre-registered their analysis before the data was available. That means we can be confident that their results are not dependent on flexible analytical methods that allow the researcher to reach a preferred conclusion.
- Despite these positive impacts, unemployment insurance improvements were not incorporated into legislative efforts, including “Build Back Better” proposals and the Inflation Reduction Act. This is partly because the Congressional Budget Office would score any automatic payments as increasing the deficit–even though Congress is likely to do a similar increase through the legislative process.
- Improvements to the “plumbing” of UI are also needed to ensure that the government can effectively implement increases during future downturns, and provide adequate functioning under normal conditions.
Matt Darling: Hey, folks. Glad to see some people who are coming in. This is Matt Darling. This is my first time running Spaces, so I’m going to be sort of learning as I go. My colleague, Will, is joining. Great.
Matt Darling: Let’s see. I know I need to give some invites out.
Matt Darling: Hey. I’m really glad that we’re going to actually get joined by Peter Ganong, and Fiona Greig. I’m not sure I’m pronouncing her name correctly. But they should be joining soon.
Matt Darling: Will, can you test your mic, and see that everything’s working?
Will Raderman: Do you hear me all right?
Matt Darling: I can, yeah. Do I sound all right too?
Will Raderman: You sound great.
Matt Darling: Fantastic.
Will Raderman: A lovely voice.
Matt Darling: I will never beat Jeremiah Johnson at the Neoliberal podcast. He has the most mellifluous voice for podcasting, and it’s always a little disappointing when I hear anyone else talk.
Matt Darling: While we’re waiting for Peter and Fiona, maybe I’ll give a potted history of CARES Act UI, just to make sure we have people coming in. So as people might remember, in March 2020, we had no idea what was going on with the pandemic. And it’s sort of funny looking backwards, there was both more brutal than I think anyone anticipated, but also, to some extent, maybe less. I don’t think I ever would’ve imagined that two and a half years later I would still be masking my daughter as I drop her off at daycare. Or actually not even daycare anymore, she’s in first grade, so that’s how long this has been.
Matt Darling: On the other hand, I don’t think I imagined that life would just go on. In March 2020, I was buying all sorts of canned goods, and things like that, that didn’t end up being really important. I see Peter’s joined.
Will Raderman: I see Fiona as well.
Matt Darling: Okay. They should have gotten invites. Peter and Fiona, you should both have gotten an invite to speak. I’ll start talking, then I’ll introduce you folks.
Matt Darling: Anyway, so March 2020, we had no idea what was in store for us with the pandemic, and neither did Congress, and they sat down … And I love this story that I think Dylan Matthews tells on Vox at one point. At one point, I think Senator Wyden and Steve Mnuchin sat down and were like, “Okay, we need to really give people a lot more money for unemployment insurance, because it just is not adequate given what we’re anticipating happening. That a lot of UI programs only give people 30% replacement of their wages. A lot of the systems don’t even just work.” And they sat down, I think just did a calculator and said, “Okay, how much money do we need to get people to the system?” And they came up with $600 a week.
Matt Darling: And the way that UI works in particular, is that every state runs its own program, they’re horrendously complicated. I try to write these things up, and I can’t even figure … Everyone uses different baselines, and everything like that. It’s super duper complex.
Matt Darling: But they said, “Okay, what we have to do is add just $600 straight up to everybody on a week to week basis.” And I believe they did that through, I think August of that year, 2020. So they were doing it for about four months.
Matt Darling: And then, of course, four months later the pandemic is still going on. That expires, and Trump goes and says, “What we’re going to do is, we’re going to figure this out with executive authority. We’re going to use FEMA money, and we’re going to add $300 a month. And states can sort sign up for that.” Some states did it, some states didn’t.
Matt Darling: Joe Biden becomes president in 2021. He passes the American Rescue Plan, that brings the $300 back officially as legislation. And that expired, a year ago today, on September 6th, 2021. And one of the things I’ve been surprised at, and just how much that UI thing got dropped from the discourse in a lot of ways, that we don’t talk about it as much, we don’t think about it, but it’s something that was just such a huge, huge change in policy making.
Matt Darling: And people were really, really worried, at the time, that this was just going to completely destroy people’s will to work ever again. That people would just stop working, never do it. And it’s fascinating that we didn’t really see that.
Matt Darling: I’m going to introduce two of the co-speakers. So Peter Ganong and Fiona Greig, who I think are still trying to figure out how to log on for a bit, they’ve done some great research. I was going to have you guys start talking about some of the research you did that looked at the difference between blue states … or not necessarily blue states and red states, but the states that canceled their UI programs over the last summer, and the effects that you saw there.
Matt Darling: And then we can maybe talk about some of the other papers you guys have done on how we see things with the Marginal Propensity to Consume. And I know, Peter, you’ve done some thinking about automatic transfers. I’m going to get into my DMs, and make sure you guys can join in, and see if I can make sure that works.
Will Raderman: I can hop in for the moment. So some of the additional things that were pretty fascinating with the UI bump early in the pandemic by the government was, not only did they replace $600, or at least added 600 to everyone’s weekly benefits, but they initially tried to … Sorry, one second.
Will Raderman: They initially tried to replace a percentage of everyone’s wages, and then they decided there just was not the administrative capacity to handle that state by state, so they went with that flat sum. And what you ended up seeing, and what I think Peter and Fiona can go into, is that a lot of … I think the majority of unemployment benefit recipients actually were having all or more of their wages replaced, which is a massive departure from how the programs function normally.
Will Raderman:Usually, there’s a huge struggle to actually afford your normal day-to-day basic necessities. And that is assuming you even qualify, it’s very hard to do so in many cases. So to actually reach a point where gig workers were actually able to qualify in a pretty straightforward way, the fact that there was extensions being added so that people were not necessarily worried about getting thrown off when they had not actually found a job yet, it was really a pretty big departure from just the normal function.
Will Raderman:I’m very excited to listen to Peter and Fiona break down the analyses that they’ve done.
Matt Darling: So I think Peter and Fiona might have been logged on via their computers. That’s always how I listen to Spaces, I know, because I’m an alt tab on my work computer person. I think Spaces makes you log on via your phone, so I think they’re going to switch what mechanisms they’re using, and then should be able to get the access.
Matt Darling: I’ll maybe introduce … One of the pieces that I know Peter worked on, and I think Fiona might have worked on it as well, was just looking at the job finding rates over the summer between the states that canceled UI, and the ones that didn’t. It’s so interesting to me, because this was such …
Matt Darling: There’s so many papers that came out on this, and one thing I really appreciate about the one that Peter worked on, was that he preregistered all the analyses, right? Because you can imagine that all these states are going to be different, in lots of different dimensions. It would be a little bit complicated to figure out.
Matt Darling: Also, just the fact that, what it ended up being, which is so unfortunate, is it ended up being this extremely straightforward blue versus red states. If you went and looked at it by what states voted for Biden, and what states voted for Trump, I think almost all of the ones that voted for Biden ended up keeping it through September, and almost all the ones that voted for Trump ended up dropping it. I might be wrong about one or two states there, but it could be the case.
Matt Darling: Okay. It looks like Fiona has figured out. Fiona, can you confirm that you can talk, and it was the phone thing?
Fiona Greig: Hi. Yeah, it was the phone thing. Thank you for that.
Matt Darling: No problem. No problem. It was really funny, it took me a second too when I was logging on, because I was like, “Oh, where his Spaces in Chrome?” And it’s not there, it makes sense.
Fiona Greig: Well, thank you for having me. And hopefully we’ll get Peter on also. A lot of the work that we did, we shared a lot of the concerns and hypotheses that you laid out, which is that, is it going to be the case that with these very high levels of UI payments, especially in the initial CARES Act that were offered, the $600 supplement, that was more than offsetting workers wages that they had lost when they lost their jobs, for most workers, especially low income workers?
Fiona Greig: And so there was a big concern that this would be contributing to the labor shortages that we were experiencing. And we didn’t see that actually. I mean, we saw some of it, but the work disincentive effects we measured were small. They were small according to three yard sticks. The most important of which was the benefits, the spending impacts. The spending impacts were very large. We could see that in the JP Morgan Chase data.
Matt Darling: Fiona, can I pause you there for a second? Because I’d love for you to talk about the JP Morgan Chase data. I’m a person who’s mostly been talking about this and using Fred, but you guys had this really great rich data set to look at, and I wonder if you could add some more details.
Fiona Greig: Sure. Essentially, this is checking account and credit card data, so it’s very high frequency data at the household level. But, importantly, we could observe families receiving their UI payments directly deposited into their checking accounts. And from there, we could see how their income was changing, that it was going up for many families when these generous UI payments started arriving. And also, that their spending went up when these UI payments started arriving. So it’s a high frequency data set that you can see by the day to the day, and on an ongoing basis over time, for families who were receiving UI directly deposited.
Matt Darling: And the other thing I’ll add to that too is, it’s a huge data set. I think it was like 40 million households, is that right?
Fiona Greig: Yeah. I mean the overall universe of checking accounts is on the order of magnitude of what you described. But when we pared down to samples of workers who are receiving UI payments, it’s still a million people or so, so lots of people. And the benefit of that, I think, really relative to some of the federal data sets, is that you have much more confidence in the results, because the confidence intervals are so much smaller on the results than if you’re working with survey data.
Matt Darling: No, I did a randomized trial once where I had the same sort of access to data, but I had it for a thousand people. And it was sort of like, if one person more in the treatment group, or the control group buys a house, that changes everything immediately, because that’s just such a huge payment. But, obviously, with a million plus folks, you have a lot more to go on.
Matt Darling: I think Peter’s on now. Peter, can you maybe say something?
Peter Ganong: Can you hear me okay?
Matt Darling: Yep. I can hear you. Great. Great. I was wondering, could you maybe talk … I was mentioning that free registration process, and I was wondering if you could maybe talk about how you guys approached that? Because I think that was just such a big strength of your paper.
Peter Ganong: Sure. Well, thanks for … I’ve never done Spaces before, so thank you for having me.
Matt Darling: It’s a learning experience for all of us.
Peter Ganong: I’m sure I’m going to do some things wrong, but I appreciate the opportunity regardless.
Peter Ganong: In many co-authorships, like with Fiona and I, it’s often hard to figure out where one person started the sentence and the other person ended. But in this case, it’s very easy to do clear co-author attribution. So we had pre-written code, because we wanted to hit go on code, like the second the BLS posted the state employment data. And at 9:00 PM the night before, Joe Vavra, one of the co-authors, posted on Slack like, “Oh, we should definitely preregister this.”
Peter Ganong: And it was a great idea, because we’d already written the code. It was just a description of what we had done. But the specific reason that we wanted to do it, we wanted to pre-register, was we knew that there was going to be cherry picking on both sides. And indeed, there ultimately was cherry picking on both sides.
Peter Ganong: And so, as you said, I think it gave our estimates a measure of credibility, because we had said, before we saw the data, what we were going to do. And so you had people like cherry picking California versus Texas. I don’t even remember if that was someone who was pro-termination or against termination, but there’s a lot of cherry picking.
Peter Ganong: Usually, you see pre-registration for RCTs. And so the first thing we did actually was we tried to pre-register with the AEA, and the AEA rejected our pre-registration, because they said, “Look, this is only for RCTs. You’re looking at observational data.” And then the Open Science Framework accepted our pre-registration, and so that’s where it continues to live.
Peter Ganong: And then the results, it’s interesting about the results-
Matt Darling: That’s so funny too-
Peter Ganong: I want to avoid being too much inside baseball, there was no statistically significant difference in growth between states that terminated benefits early, and states that terminated benefits late. And no matter how much you pushed and pulled the data with different assumptions, you couldn’t really get yourself further than termination created 0.15 jobs for every person who lost benefits, or termination destroyed 0.15 jobs for every person who lost benefits. So it’s a big fat zero when they were … Some people said they were expecting either disaster, or a big hiring boom, and neither of those things materialized, as viewed through the lens of average state level statistics.
Matt Darling: Yeah, I’m not going to call out any papers, but it is sort of funny, because I think you got … There’s a bunch of papers that came out right out of the hat, and I think yours was the first one that I saw, which makes sense, if you had the code ready, and you just had to press the button, and then had all your graphs set to put into the world the next day.
Matt Darling: But it is sort of funny, because I feel like there is a thing that’s, every month or so there’s a new reanalysis that comes out and it’s, “We only look at these states, and we square these three variables, and we use this wacky instrument,” and it goes pseudo-viral in different communities. And I’m like, “Well, are these …” I would never have sat down to think about those. You can always make a storyline that makes anything work like that, but it’s always so hard.
Matt Darling: The other thing that’s so fascinating to me is, I’m coming from that RCT world, where preregistration is getting very, very common. And it is always so funny, because here on Twitter, and also just the policy discussion, obviously, in general, we end up making huge inferences out of these incredibly noisy day-to-day data sets. And, of course, no one’s preregistering anything, right?
Matt Darling: Maybe people will predict stuff. They’ll say, “I expect there to be 400,000 jobs next week.” But there isn’t that sort of like, “Aha, I’ve written out the code, I’m going to be fully accountable,” sort of thing.
Matt Darling: It’s very interesting coming from that RCT world, where you’re playing for smaller stakes too, where you’re talking about, “Okay, we’re going to see if this program works. Going to then maybe scale it up if it works.” And then policy is like, “Well, this data set that comes out once a month was good or was bad.” And so policy has to react, and the Federal Reserve has to change interest rates, or …
Matt Darling: The example I always think about is from Spring 2021, which basically caused a lot of the states to stop doing their pandemic program. I think April 2021 had this really low jobless rate, I think 226,000. Obviously, March and May were fine, but a whole bunch of people … governors were like, “Well, this is clearly because of UI, and that’s what’s holding it back, and cancel the program.” Which, not great for all the people who, of course, were relying on it, even if it gave some plausible way of you guys being able to look at that.
Matt Darling: I maybe want to turn to one of the other findings that you had. I’ll go back to Fiona for this, although of course both of you can answer. As you said, there’s a division of labor here.
Matt Darling: One thing that I thought was really interesting, from that macro perspective, was one of your papers was looking at saying, “Hey, we’re giving all the people who normally have very little liquid cash, just no cash on hand, all of a sudden they’re getting these big UI checks, or the EIPs, and all of a sudden have a lot of cash on hand, but still acted like people without too much.” I was wondering if both you could … Or, I’ll turn to Fiona first. If you guys could speak to that?
Fiona Greig: Well, certainly. And it wasn’t just the UI payments, it was also the three rounds of stimulus, which were much more concentrated in terms of how they impacted people’s liquidity picture. And you’re absolutely right that UI recipients, really their cash buffers, how much they had in their checking accounts, were considerably boosted by both the stimulus payments and the expanded unemployment insurance payments.
Fiona Greig: And so it’s interesting, because that was at a time when people had lost their jobs. It was at a time when actually the spending of unemployed workers had actually increased. So because the benefits from UI were generally larger than their pre-job loss earnings, their spending increased, rather than decreased, at a time when the spending of everybody else, those of us who were still working, had generally decreased. We saw big spending declines at start of the pandemic among employed workers, to the tune of 20-30%, because of stay at home orders, et cetera.
Fiona Greig: And so, if we looked at the unemployed workers, their spending had increased, and their liquidity had increased. So it was a very interesting financial picture to watch over time. And those elevated cash levels actually have sustained for quite a while. We continue to watch that by income level, and by jobless workers. And even as of the end of March, they were still elevated to the tune of sort of 40-50%. So these were considerable benefit increases, that really helped families weather what was a very financially volatile time for them.
Matt Darling: Can you say more about that? When you say 40-50% higher, you mean if you look at what someone’s normal bank balance is, it’s still 40-50% higher, is that right?
Fiona Greig: Yeah. So the baseline here would’ve been pre-pandemic, the 2019 levels. So over time we’ve been comparing cash balances in people’s checking accounts to those 2019 levels. Now, over time that starts to feel a little bit dated as we get into 2022, with inflation impacts and things like that. That’s a work in progress. But certainly during the time period that you’re talking about, that was the right baseline. I mean, cash balances were elevated to the tune of 70% for low income families.
Matt Darling: And when you say low income families, because this is low income families within your data set, so what would be the cash amount? I’m trying to contextualize this.
Fiona Greig: Well, obviously in terms of absolute levels, this doesn’t amount to a whole lot of cash. So low income families had on the order of $900 to $1,000 in their accounts, and then that went up, to above that level, as a result of all of these different sources of support, government supports. So when we’re talking about elevated cash balances, we’re not talking, in dollar terms, about a huge amount of money for families. It’s not necessarily something that might sustain them for months and months and months, and on and on. But in proportional terms, they were larger proportional boosts for low income families, who generally have much lower balances than say high income families.
Matt Darling: Yeah. That’s so interesting. I haven’t been tracking that closely, but it’s interesting, because I feel like we keep having this myth that people are about to run out of money and then rejoin the labor force. I know Mark Zandi was talking about that December of last year, where he was like, “Hey, we’re …” He’s got a slightly different data set, and he was saying, “Okay, we’re anticipating that the pandemic payments will have basically gone away in January, and then we’ll see a bunch of labor inflows.”
Matt Darling: And I think there’s a Wall Street Journal article, or oped, even just last week, that is sort of the same thing, saying, “Hey, we’re still trying to do this, and the job market still is seeing some sort of slacking because of that.” But at the same time, obviously, we’ve recently re-hit, at least with the primary age population, near the pre-pandemic high of around 80.3% of people between 25 and 54 have jobs, so that’s been exciting.
Matt Darling: It’s just so funny to me that it’s this permanent increase to people’s liquidity … Or, not permanent, but a really long lasting time, and people just aren’t spending it down. Do you, or Peter, do you want to speculate about that? Either why that’s happening, or will we ever see this go away?
Fiona Greig: I thought, Peter, you were going to jump in. He went off mute.
Fiona Greig: Well, I’ll just say, I have no idea. It’s something that the JP Morgan Chase Institute continues to track and watch with interest.
Fiona Greig: I will say, as context, that even before the pandemic, or just in general, we observed high levels of income volatility, high levels of spending volatility, even in good times. And so, from the perspective of somebody who cares about financial health of Americans, it’s actually a good thing that these cash buffers would go up, because we noted that most families didn’t have enough of a cash buffer to weather a 25% dip in income, and a 25% increase in spending that might occur at the same time.
Fiona Greig: So in a world of razor thin cash buffers, which have been documented through various different forms, that notorious survey that the Fed does, the SHED, which asked, how would you cover a $400 expense? And an extraordinary number of people say, “I’d have to beg, borrow and steal. I couldn’t cover it.” So if there is a new normal that has been set, a new waterline, where families are used to seeing $1,300 in their checking account, instead of $1,000, that would be a good thing.
Matt Darling: Yeah, absolutely, absolutely. It is always sort of funny, I always am quoting that SHED one, because people would always be like … And you’re like, “Actually, this is at an all time high.” I can never remember what the exact numbers are, but that thing where it’s like, “How many people could cover that $400 surprise expense?” People would be very upset that it’s, I think, 40%, something around that. But you’re like, “40% is actually as good as it’s ever been. Normally, it’s 30% or 25%.”
Matt Darling: Obviously, we want that to get much, much higher. We want people to be able to deal with these sorts of, well, I’d say normal shocks of life. I think the comparison I always think is, that’s how much a minor repair to your car costs. It’s not necessarily the car doesn’t work ever again, but it is something where it’s like, okay, you need to get it fixed so you don’t get into an accident later. So getting those things is really important.
Matt Darling: But, also, I know for me trying to figure this out is like, I say, “Okay, I’m definitely more risk averse post-pandemic than I …” Not post, during still, of course. But I definitely think about risk in a very, very different way, and sort of saying, “Oh, what’s the next bad thing that’s going to happen, that’s going to just completely turn over my life?” And especially as a person who’s spent most of his life in the boring era of the ’90s, and everything like that, where it was like, “Oh yeah, things are really bad, but not in a way that necessarily affects you in the day-to-day,” it’s a big change thinking about that.
Matt Darling: Peter, can you still log on, or are you-
Peter Ganong: Can you hear me?
Matt Darling: Okay. Yeah, I can hear you.
Peter Ganong: Okay. I thought I was talking before, but I guess it didn’t work, because of Spaces or something.
Matt Darling: No problem. I wanted to ask you, so I know … I think maybe we’re at a half hour, so I maybe wanted to answer some questions from the whole group for a little bit. And I know Fiona might have to leave.
Matt Darling: I know you’ve thought a little bit about automatic triggers, and how that could work and I was wondering if you can maybe say something about that? Especially how they tie into your other research.
Peter Ganong: Testing. Can you hear me okay?
Matt Darling: Yeah, I can hear you.
Peter Ganong: Great. So the thing that policy wonks always want to do is make UI more automatic. And just to be clear, I self-identify as a policy wonk, and therefore wrote this paper with Gabriel Chodorow-Reich and John Gruber, with two other economists, thinking about making UI more automatic. And there were a few conclusions that were potentially surprising to me, and we wrote this paper because we thought that there was going to be a UI reform process happening in, I guess it was 2021. And as happens with many reform processes, it didn’t materialize. However, the way that any reform process happens is political will emerges out of nowhere, and then you just run to the shelf and grab the most recent studies from the last time when there was a process that didn’t work out. And so I am totally happy to put this on the shelf for whenever there’s interest, which I’m sure will come up again soon. So I’ll talk a little bit now about UI-
Matt Darling: Let me add something to that, because I’ve been so fascinated by that, because I’m sort of new to the think tank world, and always been an econ wonk nerd person since I was reading these things, but it was fascinating. I know we do a lot of work in this cannon on the child tax credit, and one of my colleagues, Josh McCabe, has this big 800 page book that … I think he was on earlier, but he’s logged off.
Matt Darling: I can abstract it a little bit, but to some extent it’s just sort of like, “Oh yeah, here’s the evolution of child’s allowances in the United States versus United Kingdom.” And it’s literally what was on FDR’s desk at the time where he said, “Hey, we need an extra benefits program.” And what was on Winston Churchill’s desk when he said the same thing.
Matt Darling: It’s just so funny to me that it is literally what’s on the desk when they walk into the office. And you can see that process historical change a lot, the more you look for it. So sorry for interrupting, you can go back to-
Peter Ganong: No, that’s really cool. I did not know that. That’s really, really neat.
Peter Ganong: We’re doing work for the shelf, and then it’ll get picked up when someone decides that they want it to be picked up. So policy nerds and economists always want to make UI more automatic. The idea here, let me describe first the motivation, and then a little bit about what we learned from the policy.
So the idea is, there’s always a lot of headlines about Congress fighting about whether or not to extend UI benefits. Until 2020, the fights were always about how many weeks should there be. I think next time around there will also be a fight about how big should the supplement be, as there was this time. And then there will also be a fight about something to expand eligibility, because UI right now is Swiss cheese. Meaning, lots of people who lose their jobs are not eligible for UI.
But there’s always a thing of, well, why is congress fighting? Why don’t we just make up a rule that will make things automatic? And you can think about, the analogy you can have in the back of your head is, for monetary policy, you’ll often hear economists advocate for the Taylor rule. Which means, why do we need the Fed? Why bother with the Fed, let’s just have an automatic rule that sets interest rates as a function of inflation. And, similarly, why do we need Congress to decide UI? Why don’t we just have it be automatic that UI extensions happen?
And so what we did was, we asked basically two types of questions. There’s a trade-off anytime you design any set of triggers, which is that there’s going to be false positives and false negatives. So by false positives, I mean you trigger on, but a recession doesn’t happen, or you trigger on, but you trigger on earlier than you need to. And then false negatives is, you don’t trigger on, and people are really in bad shape, because you didn’t trigger on when you were supposed to.
What we learned basically is that, of all the policy rules that we, that’s the three of us, were able to come up with, they all generated results in terms of that false positive, false negative trade-off. That was similar to what Congress actually did in the 2001 recession, and in the Great Recession. So that is to say that the typical argument for automatic triggers is framed as, if the automatic triggers will do better, why would you let Congress do this, when you could have a bunch of technocrats do it?
And that argument, it turns out, doesn’t hold water. Congress, ex-post, did just as well as the technocrats. Who, by the way, we have the benefit of hindsight, so maybe Congress even did a little bit better at this trade-off between false positives and false negatives.
Now, the flip side … Sorry, did you want to jump in, Matt?
Matt Darling: Oh, no, I was just going to comment that was so surprising, especially coming from the technocratic perspective, where you’re like, “Oh yeah, I assume we’re going to be right.” That’s great.
Peter Ganong: Right. You hear the pointy heads say … Actually, the thing that all the think tanks, and I, would’ve thought going in, was wrong. And that’s why you do research, is you find out that it turns out Congress did just as well, at least as well as the three of us, Gabe, John and I could do on automatic triggers.
However, there’s a flip side to this. I feel like if you’re in these Twitter spaces, you’re probably here for some of the inside baseball, and so we’re now going to do inside baseball together for a few minutes. So the true reason, or one of the true reasons, why comprehensive UI reform did not show up in Build Back Better, and then the Inflation Reduction Act, is that it had a big fat CBO score. Meaning, the CBO says, “If you put in automatic triggers, it’s going to cost a lot of money.” And my understanding, and this could be wrong, is that the votes were there if the CBO score had been zero, but not for a big fat CBO score on the order of the hundreds of billions of dollars that we spend on UI in the recessions.
However, that CBO score is defined relative to a do nothing baseline. And that’s the wrong baseline, because UI benefits do in fact get extended in every recession. Basically, saying that the automatic triggers put us in a similar place to where Congress has put us, is to say that automatic triggers have no cost relative to what Congress already does.
And that means that basically, if you are listening to this podcast, and you’re thinking, not, how is CBO going to score this, but how should I as a citizen, or as a voter, or as a policy analyst, think about the cost of these triggers? You should think that their cost is pretty close to zero, because they’re codifying into law more or less what Congress has done anyway. So then why would you bother to codify them at all? Can’t we just rely on Congress?
Well, just like mutual funds, past performance is no guarantee of future performance. And so, if you like what Congress did in 2001, and in the Great Recession, you should be in favor of automatic triggers. If you don’t like what Congress did in 2001, and the Great Recession, then you should not be in favor of automatic triggers. I feel like I’ve talked a lot now, so I’m going to cut myself off, but would welcome anyone to jump in on the question.
Matt Darling: No, that’s fascinating. I think that sort of inside baseball is always fascinating to hear.
Yeah, I do want to open it up for questions. I know Jody Chen has asked on Twitter, he doesn’t want to speak, but he asked, how can we … operational and familiarity improvements with UI from the pandemic.
I’m not sure exactly how to think about that, but I think one of the things I know, for me, is the pandemic obviously revealed a lot of just issues with UI, and I think were clear to some of the policy wonks and nerds, to some extent, like you were saying earlier, but then became really salient to people.
My colleague, Will … I know Will … Yeah, you got your hand up, so do you want to jump in? Because you actually did this, you’re now a UI wonk, but used to be a UI applicant.
Will Raderman: Yeah, so I would love to hear Peter and Fiona’s thoughts on what policy is the way going forward. Peter, your breakdown of automatic triggers is very much my fear for future reform efforts, where seeing how extended benefits … The additional weeks of UI that are activated when the unemployment rate is high, most states’ programs did not function well. And so one of the lines of thought is, we need to make those triggers better, like you said, and ideally have federal standards, or federal programs that are doing that.
But my fear is that there’s going to be a lot of focus that goes to having benefits, benefit enhancements, that could activate, and then those activation standards are not effective. Whereas, if there was more focus put to reforms that help everyone at all times, you know that those kinds of changes are going to be worthwhile for everyone, in essentially every economic environment. Because we know that unemployment’s affecting people, whether the economy is good or bad, there’s a constant flow.
But there’s potentially going to be that necessity to decide what gets prioritized. Do we prioritize those recessionary policies, or are we prioritizing some kind of permanent improvement that you’re seeing everyone experience? But I would be very curious to hear your thoughts on those.
Peter Ganong: Great. I’ll just jump in briefly, and then Fiona you should please jump in if you have something that you want to add. So there’s this old book from roughly 1960, that UI wonks really like, by Haber and Murray, and it’s like History of the First 30 Years of UI. And there’s this great line in Haber and Murray, where they basically say, “When a recession is going on, everyone is supportive of more generous UI, and they think the UI system is great. When there’s a boom, no one’s interested at all in UI, and everyone thinks that anyone who’s receiving UI is lazy.” I think that 1960 description of the UI system is a very good description of how the political dynamics of UI also are in 2020.
Will Raderman: Agree.
Peter Ganong: Having said that, I don’t see the trade-off that you’re describing. And I’m not saying this as, we can have our cake and eat it too. So let me be slightly more specific about why I don’t see the trade-off you’re describing.
First, Senator Wyden released a discussion draft, so it was not quite legislative language, but edging towards legislative language, in 2021. And because he is, and was, Chair of the Senate Finance Committee, he was going to have the jurisdiction, if there had been legislation that occurred on this. And the discussion draft both had permanent fixes around eligibility and benefit adequacy, and it also had triggers.
Will Raderman: Yeah, I liked that draft a lot.
Peter Ganong: Most times, when there’s a proposal to fix it, people will both want to fix the longterm issues and the cyclical issues. That’s just how it gets written up. And then the question is then, when do we find the will to fix it? Because in the heat of the moment, you’re focused on firefighting. And then the fire is out, and then everyone says, “Oh well, I guess we have no need for firefighters anymore, let’s just defund the firefighting department again.” Now, I’m getting really cynical, but I’ll stop there.
Fiona Greig: Just one other thought on the longterm, and sort of evergreen policy, versus the firefighting policy, is that what matters for policy is not just the economics of it, but also the plumbing. And what we learned, I think the hard way, during the pandemic, was that we were building whole new pipes that didn’t exist.
Fiona Greig: We were trying to calculate individual level benefit levels, or replacement levels. That didn’t work. That’s why we had to do a flat supplement. We were trying to expand eligibility, and had to throw verification out the window, because we had no means of doing that. Extending benefit durations operates in the same pipes as we already had.
Some of the key innovations that we explored during the pandemic, where we were increasing benefit levels, and expanding eligibility, those were entire new pipes that had to be built. And so in some sense, if that’s going to be part of the future recession response, or even an evergreen response, if we think on a permanent basis we should be expanding eligibility, it’s important so that when the storm hits, when the recession hits, we’re at the ready. We are not trying to learn how to build new pipes at a time when we really need to be pushing money and benefits through those pipes.
Matt Darling: Great. I’m going to give the mic to Jody, who had a question, I think I had garbled it earlier.
Jody, you should have gotten the approval. You need to be on your phone, as we discovered earlier. I don’t know if that is working. Okay, it looks like you got it. Jody, you’re still muted, but it does show you as a speaker now.
Jody: You got it basically right, and there was a lot embedded in it, and you guys covered a lot of good stuff. One thing on the positive side, I’d like to just throw in is, we just threw so much through the system, I just have this feeling that, both in terms of policy shaping, but also in terms of daily familiarity of the process, just a number of different people have gone through this, in terms of on the household level, in these administrative offices, in the policy, that we must have learned so much, and we must have tilted opinions in different ways.
Jody: I’d be curious more to know what people think about what got fixed. And you were getting at some of these, Fiona. What things do we go through, and either little fix … You would think we’ve improved the system overall as a result of the stress test. Certainly, I see some of the stuff around test grounds for new stuff, it being a good experience for that, but also people generally only thinking about unemployment insurance when it’s something they go through. But we’ve had so many people go through it, I wonder what that means for the policy environment too. I mean just the sheer numbers of individual households who have done it.
So my big question is, what are the positives we’re walking away here, in terms of, we went through all this, what are the good stuff we might see, just working from the ground up? If that made any sense.
Fiona Greig: Well, I think in terms of the economic responses, that we’ve learned that UI, of all the recession tools that were used during the pandemic, I think it’s fair to say that Peter and I think that UI was the most effective, in terms of boosting spending, and offsetting consumption losses. And that’s actually a huge lesson, and that’s a comparison exercise comparing UI, thinking about it in the context of the rest of the CARES Act, stimulus, PPP, et cetera. So I agree with you, I think there was a lot of good news, in terms of how effective UI was, and all of its various … the three huge expansions, the increase in eligibility, the increase in levels, the increase in duration. So from a top line, I wouldn’t want us to leave this conversation without naming that top line.
Matt Darling: Well, one thing I want to add to Jody’s question is, and this goes to Peter’s point earlier, we really thought there was going to be a big UI reform bill, and it couldn’t get past CBO as the main challenge I guess, but there’s so much evidence that going through the experience of seeing how bad a lot of these social insurance programs operate can often help build that sort of political will to change them. And so it really felt like we should have had something, because we never had 40 million people be using these systems before, and then we did. Obviously, that didn’t have the effects that we might have expected in the short term, but I do think that a lot of people went through this, and it’s a lot easier to go and explained like, “Hey, we want the UI system to work a lot better, because everyone will … if they hadn’t gone through it themselves, they’ll have a cousin, or a brother, or a friend who has.”
Fiona Greig: Yeah. One other success I would make, in addition to worrying about the work disincentive effects, there was a lot of concern around overpayments and fraud. And, of course, given the sheer number of people and dollar amounts that were being distributed … Sorry, the number of people benefiting from UI, and dollar distribution sizes, the price tags on those fraud numbers are high, and they are jaw dropping. But, if we think about them in terms of overpayment rates, and this is I’m sure still jury out in terms of these claims getting assessed, and adjudicated, and things like that, but the overpayment rate during the pandemic was 18% for fiscal year 2021. That compares to about 10% in normal times.
A different way of saying it is that 82% of the time the correct person was getting the benefits. So despite the fact that we relaxed eligibility requirements dramatically, I think it’s important to underscore a glass half full picture here, in terms of the share of people who were the right people getting it.
Jody: And maybe I’m just feeling a little too optimistic today, but it feels to me like your top line takeaway of this being something that’s recognizable now a little bit more of an importance, but also it just feels to me like all this data on folks who have gone through the system, all this fraud, it’s got to lead some benefits going forward to cut it down a little bit in the future. In all these individual cases that people went through, administrative offices, I just have to believe, you have so many different anecdotal examples of different things that can go wrong, either innocently or not, I don’t know, maybe I’m too optimistic at thinking it leads to a better place, but … I’m trying to be optimistic today.
Matt Darling: No, Jody, I totally agree with those points, and I think there is this sort of root in … I think the hope is that we’re able to get a coalition. This is all we’re trying to do with this cannon, of course, is say, “Okay, the people who want benefits to be better, and more generous, and really help people stabilize income as they look for new jobs, there’s also the folks that really are concerned about fraud and overpayments and everything like that.” And I think there’s a lot of good faith reforms that solve both of those things.
Matt Darling: There’s some political problems that are caused because there’s some coalition that benefits from the problem, and that’s not the case here. There isn’t the coalition of COBOL programmers who are demanding that UI be in COBOL forever. And hopefully what we can do is figure out ways of getting some of these improvements in the administration of the programs, at a sort of bare minimum.
I do want to pause. We can continue responding to Jody’s questions, but I did want to see if anyone else had a question. Please just raise your hand if you have something. Jody, if you have another question, of course you can do the same and jump back in.
Will Raderman: I see Andy Stettner is here. Andy, if you have any thoughts you’d like to add, that’d be great. Oh, there we go.
Matt Darling: Okay. I think I approved Andy.
Andy: [inaudible 00:53:08] the space. I guess it’s a comment and a question. Maybe a form of a question for Fiona and Peter. What do we know about these workers whose benefits were exhausted? What have we learned? Have they been able to get back to work? And who’s not there? That’s the question.
Then a comment. I think today I’m more just remembering how we, at least in my theory, really held people whole, and we didn’t see the scarring that has been life changing in so many other recessions, and so feeling a lot of gratitude for how many people did get these benefits, and the difference that made in people’s lives.
Peter Ganong: I’m eagerly awaiting to read that paper, and I have not seen it. I think it would be really important when it is written, in terms of follow up for people who, A, got a lot of benefits, but then, B, saw benefits cut off potentially earlier than planned.
Matt Darling: I’ll add to that. One thing that I’m always fascinated about this, and I don’t know the international data as well, to really look into it, but if you look at the experience of what the labor systems look like in the United States versus other countries. UK, Canada, obviously the most easy to make examples, they saw faster employment recoveries. And I think largely because they had more PPP like systems, which said, “Okay, we’re going to keep people in your old jobs. We’re going to pay your payroll through the federal government, everything’s going to go normally.” And so they saw much smaller decrease in employment, they went down 5%, or unemployment up 5%, as opposed to unemployment up 30-40% in the United States.
Matt Darling: But also, and this is a little bit hard to see, it doesn’t look as much as they’ve been having this sort of huge sectoral shifts that we’ve been seeing, in terms of just people quitting their old jobs, finding new jobs that pay more, have better benefits. That narrative is hard enough to disentangle in the United States, in terms of being like, “Okay, did wages go up? Once you account for inflation, did they go up? On what percentage of previous wages did they go up?” Once you account for the increased risk of just getting COVID, or having someone yell at you because you’re enforcing mask mandates, it’s so hard to tell. But I definitely think there’s indications that the United States employment response has been different from other countries. And I think, like Peter was saying, that paper hasn’t come out yet, but I’m hoping it does, maybe next year, or this year.
Will Raderman: Yeah, I guess just to add one fact, so Sam Hammond, in his paper Faster Fairer Growth, he talks about the labor market spending differences between the US and other countries. And to match what the OECD average is, as a proportion of GDP, the US would need to spend a hundred billion dollars more per year to reach that average. That’s how much less spending we are actually putting forward to make sure that just normal transitions between jobs are made to be swifter, and more efficient, in making sure that people are finding positions that are well suited for their skillset, and making sure that they’re getting the wages that they should be paid. And that is very much something where there was less attention to that before the pandemic, but seeing the overall response, and the shortcomings of needing to put together programs right in the moment, it’s definitely a place to look for more investment going forward.
Matt Darling: Great. Great. We’re almost at an hour. I do want to end at 6:00. Some people can have dinner, or if you’re in a central time like Peter is, you can end your workday. Is there maybe one more question, or I can leave with some concluding thoughts?
Will Raderman: I thought we were going to just do a 24 hour?
Matt Darling: 24 hour live Twitter Spaces in protest until they announce a new UI reform aspect. Well, great. It was so great having people. Peter and Fiona, thank you so much for joining. I know I search Peter Ganong UI on Twitter, to get some of your graphs, I think every week it feels like, so I really appreciate getting the chance to talk to you, and hear some of that inside baseball. Same thing, Fiona, and especially all the wonderful stuff that you’re able to get from the JP Morgan site.
I think this is so important, especially right now where we’re at, where we’re like, “Is the Federal Reserve going to pull off the soft landing?” I hope they do, of course. But it’s also just all these problems with UI that were cataloged still exist, and still need improvements. But at the same time, I think what we saw in the pandemic UI system, and what your research really contributed to is, it worked absurdly well. I don’t think that the people who were worried about UI labor scarring, or anything like that, were coming at from bad faith, but it definitely shows that there’s just a lot more flexibility, and a lot more freedom to design really great programs otherwise.
My plan is to … I think Twitter is going to record this. Peter and Fiona, I’ll share it with you guys before we post anything publicly, but I think this would be great to be able to share to the public. And then I know my colleague Tara’s on, she wants to make sure that I plug the Niskanen Social Policy Center Newsletter. I will link it on my Twitter feed right after this. I don’t think I can say it out loud, it’s a MailChimp thing. Peter and Fiona, I’ll let you guys have the last words.
Fiona Greig: Thanks for having me.
Peter Ganong: I’m going to leave you with a number, because clearly if you stayed this long, you’re here for the inside baseball. So Fiona was talking before about the ways in which UI is more effective than other types of stimulus. And so, in a recent paper, we did a horse race. We said, “Would it be better to give out untargeted stimulus, like what has been done in every recession since 2001? Or to give out additional payments to unemployed households, larger than the normal amount of money that they get through the regular UI system?” And let’s make the following absurd assumption, that we don’t intrinsically care at all about helping unemployed workers, we just care about managing aggregate demand. I’m the Federal Reserve of 1990 or 2000, I don’t have any of the distributional frameworks, I just care about managing aggregate demand.
Peter Ganong: What we found is that, before paying out even a single dollar of untargeted stimulus, it would, in a normal recession, not a pandemic recession, make sense to pay about $2,000 in severance payments to unemployed households. And the idea is that, these are households that really need the money, and that benefits those households, but it also has benefits to the broader economy. So the number I want to leave you with is a $2,000 severance payment. Thanks so much.
Matt Darling: Thank you. And thanks everyone for participating. And I’m going to end it now. Have a good night.