Read on Twitter
The Pandemic Unemployment Assistance program has helped millions of Americans who are left out of the normal unemployment system, and has helped keep the economy from suffering even worse damage.
It has also been plagued by fraud and data problems. https://www.nytimes.com/2020/09/11/business/economy/pandemic-unemployment-assistance-fraud.html
It has also been plagued by fraud and data problems. https://www.nytimes.com/2020/09/11/business/economy/pandemic-unemployment-assistance-fraud.html
Our story (with the Dream Team of @ConorDougherty @NelsonSchwartz @PatcohenNYT) delves into why and how the PUA data is such a mess, including widespread fraud and double-counting. https://www.nytimes.com/2020/09/11/business/economy/pandemic-unemployment-assistance-fraud.html
But in this thread I want to delve deep into a related question: How many people are actually getting PUA right now? Strap in, folks. This could take a while...
OK, so first of all, a quick refresher: The latest weekly claims release from DOL says that there were 14.6 million continuing claims for PUA filed the week of 8/22. That often gets reported as being equivalent to 14.6 million people receiving benefits. But it's not!
Two reasons:
1. DOL reports *claims,* not payments. Claims can be rejected for all sorts of reasons. This is a bigger issue for initial claims than continuing claims, but it's relevant to both. (I'm not going to get into it much more here, but it's worth remembering.)
1. DOL reports *claims,* not payments. Claims can be rejected for all sorts of reasons. This is a bigger issue for initial claims than continuing claims, but it's relevant to both. (I'm not going to get into it much more here, but it's worth remembering.)
2. DOL reports *weeks,* not individuals. In normal times, those are the same thing -- you file for benefits each week, so 1 person = 1 week = 1 continuing claim.
But these are not normal times!
But these are not normal times!
As has been widely reported, there were huge backlogs, especially for PUA. By the time people got through, many of them were owed multiple weeks. And in at least some states, those multiple weeks all seem to have gotten counted as separate continuing claims in the same week.
(There are other issues with the data too, but this thread is already going to be massive. So let's skip ahead.)
SO: We know that the 14.6m figure is almost certainly an overstatement. What's the correct number?
SO: We know that the 14.6m figure is almost certainly an overstatement. What's the correct number?
One approach would be to call the states and ask. Unfortunately, many states can't or won't say exactly how many individuals are receiving benefits. (Also: Calling all 50 states is very cumbersome. Trust me.)
But some states do have at least approximate counts. One of them is California, which says ~1.8 million people are receiving PUA there, even as their continuing PUA claims figure has soared to nearly 7 million. https://twitter.com/bencasselman/status/1304142540317380615
California seems to be an outlier. Other states seem to estimate the number of individuals receiving PUA at ~80% of the continuing claims total. (This is based on a sample of only a handful of states, so confidence in that number is LOW.)
If we take California's 1.8m number, and discount every other state's continuing claims by 20%, that leaves us with a total number of people on PUA of about 8 million.
So that's one approach. Can we come up with another?
So that's one approach. Can we come up with another?
Here's one approach via @ernietedeschi, which I've refined a bit further with @ENPancotti. It attempts to exploit the end of the $600 "FPUC" bonus payment to come up with an estimate.
The idea is this: We know that through July, essentially everyone on any kind of UI got a $600 bonus. So if we knew how much we were spending on that program, we can figure out total UI recipiency. And if we trust the numbers for non-PUA programs, we can back out the PUA total.
Alas, we don't have a direct measure of FPUC spending for the relevant period. But we do know *total* UI spending on a daily basis from Treasury:
https://fsapps.fiscal.treasury.gov/dts/issues/2020/4
https://fsapps.fiscal.treasury.gov/dts/issues/2020/4
For the week of July 20 (the last full week of the $600), Treasury says the U.S. paid out $25.5 billion in total UI. For Aug. 15, that total was just $11.1b. (I'm skipping the two weeks in the middle because they could reflect some leftover FPUC.)
Regular state claims fell by about 1.7m over the same period. If we factor in that drop (at avg UI of $300/wk), and assume ALL the rest of the drop in $$$ was the FPUC expiration, we get about 23 million total people on UI the week of Aug. 15.
That implies about 7.5m on PUA.
That implies about 7.5m on PUA.
We can take that a step further and assume the ratio of PUA recipients to continuing PUA claims is basically constant over time. Applying that to the Aug. 22 total, we'd estimate there were about 8.3m people on PUA that week. (That's make no allowance for the weirdness in CA.)
One last, much simpler approach: The latest Household Pulse Survey from @uscensusbureau says 21.5 million people report living in households that used UI benefits "to meet spending needs" in the past 7 days. https://www.census.gov/data/tables/2020/demo/hhp/hhp13.html
The data is at the household level, so if there are two people on UI in one household (not at all unusual!), it will effectively only count one of them. It also asks about spending, not actual UI recipiency -- someone spending saved-up UI might still tick that box.
With those caveats: If we assume 21.5 million total UI recipients, minus approx. 15 million regular state UI recipients, that's ~ 6.5 million PUA recipients. (Data is for second half of August.)
To be clear: All of these methods rely on some pretty heroic assumptions. The fact that they reach reasonably similar totals could easily be coincidence. So use all of these estimates with *extreme* caution.
As @keds_economist points out: We will get the "real" answer eventually, via IRS data. But that will take several years. (Please, please, please tell me we won't be talking about weekly UI claims by the time that data comes out!)
Lastly, it's important to note that none of this changes the big picture: Unemployment is extremely high. Millions of people are relying on UI in general, and PUA in particular. And those numbers are falling only slowly.
Thanks to @ernietedeschi @jc_econ @keds_economist @DanielBZhao @mpbitler @ENPancotti & others for helping me think through this. Errors, absurd gaps in logic, etc. are mine alone. <fin>
And of course, if you have critiques of these estimates, or better yet an estimation approach I haven't thought of, I'd love to hear 'em!
