admin | October 23, 2020
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
Changes to unemployment benefits during the pandemic along with an unprecedented flood of applicants have created chaos in initial unemployment claims and rendered them far less useful than usual for reading the state of the labor market. Many forecasters over the past few months have cited elevated initial and continuing claims as evidence that the labor market has stalled, but the numbers come with a lot of noise. Payrolls continued to recover, and although the pace is slower than most would like, the rebound in the labor market is faster than most predicted back in the spring.
Rush for benefits
As part of the CARES Act, Congress vastly expanded and sweetened unemployment benefits. The legislation added $600 a week to benefits, which meant that unemployed workers in the lower half of the wage scale would be making more from jobless benefits than they earned at their previous posts, in some cases dramatically more. By comparison, in 2009, the bonus addition to benefits was only $25 per week.
In addition, the CARES Act created a new program, Pandemic Unemployment Assistance, to offer benefits to those who are usually not eligible to collect benefits, including independent contractors, self-employed workers and others.
Further, as has been a regular feature of recessions, the federal government offered extended benefits beyond the 26 weeks offered in the regular programs, the Pandemic Emergency Unemployment Compensation Program and Extended Benefits.
When much of the U.S. economy was forced to shut down during the lockdowns in March and April, state labor offices were overwhelmed with filings for benefits. In a rush to help workers quickly, states streamlined normal vetting protocols or, in some cases, dispensed with them altogether. Even with these shortcuts, many states still took extended periods getting filers properly enrolled and receiving checks.
From the beginning, there were stories of flaws with the unemployment claims reporting. People were calling automated systems to file and facing hours-long waiting times and, in many cases, getting kicked out of the system. As a result, there were reports of people calling several times and never knowing whether they had successfully completed their applications.
There were also stories, from near the beginning of the pandemic, of widespread fraud. Hackers were using every stolen identity that they could get their hands on from the dark web to file for unemployment benefits. When states acknowledged that they would pay out benefits first and circle back to confirm eligibility later, scammers pounced. Moreover, it was far more difficult to vet applications for the PUA program, as there was not necessarily an employer to confirm the claim. One would imagine that fraud would mainly have been an issue early on and would have faded as states worked through the initial backlogs and were able to process claims more normally. However, this is apparently not the case. California was forced to pause the processing of claims for several weeks this month, as it was still facing a significant backlog and had uncovered rampant fraud in the program that was still ongoing.
Another source of inaccuracy is the double-counting of initial claims. Early on, this mainly took the form of duplicate filings, as those people who were struggling with automated phone or online systems may have inadvertently created multiple claims. Double counting remains an issue, as state labor offices are required to recertify that claimants remain eligible for federal benefits (perhaps once a month), and those recertifications have apparently repeatedly found their way into the initial claims.
Market participants, the media, and even most economists follow the headline initial and continuing claims but fail to dig into the details. Looking at the state-by-state data can often lay bare some of the anomalies detailed above. For example, a week ago, when initial claims were reported to have jumped to nearly 900,000, far higher than expected, several states posted implausibly large increases. Indiana reported a tripling in claims in a single week, from 10,000 to 30,000. However, after the release, state officials went back and double-checked the data and realized that they had counted a massive number of recertifications as initial claims. This week, that 30,000 reading was revised down to 18,000, and the latest reading fell further to 16,000. Several other states made similar revisions, while others did not revise the prior number but posted sharp declines this week, suggesting that they are making the same mistake but still do not realize it.
In the early days of the pandemic, it made sense to track the state-by-state initial claims relative to the February, pre-pandemic payroll levels to gauge how long the flood of filing might last. It quickly became evident that some states were posting implausibly large figures. The most egregious example has been Georgia. By the end of May, initial claims in the state surpassed half of the pre-pandemic payroll level, an unrealistically high proportion. Yet, the flow of initial claims in the state has continued to run quite high, approaching 50,000 a week even now. Since March, something close to 4 million initial claims have been recorded in Georgia compared to a pre-pandemic employment level of 4.6 million, and that does not include the other jobless benefit programs. Clearly, the state is engaging in some sort of erroneous double counting, or the state and federal governments are getting taken to the cleaners by fraudsters.
The California episode this month is Exhibit A in the case for massive fraud. The state, which has by far the most initial and continuing claims, halted its processing efforts several weeks ago because it had such a large backlog and because it had received credible reports of rampant fraud. After not reporting new data for the better part of a month, California was able to get back to normal reporting with the latest week’s release as well as offering revised readings for the prior two weeks. It was not surprised to see that the levels were far lower than the pre-hiatus estimate. Initial claims dropped from 226,000 in the last pre-hiatus week to 159,000, a 30% plunge, while continuing claims in the regular program were lowered by nearly a million from 2.8 million to 1.87 million, a 33% drop.
If the other 49 states did a similar intensive scrub of their program, it is likely the national numbers would drop even further. In any case, it is easy, especially when you do not have a detailed knowledge of the underlying data, to simply take whatever headline figure prints at face value. However, in this instance, common sense can take you far. Is it really plausible that the pace of new layoffs could be running at the highest level ever seen prior to the pandemic seven months after the lockdowns? Is it even possible? Sure, the level of unemployment remains high, but the notion that additional layoffs are hitting the economy at a pace of 800,000 a week even now, as the economy is slowly but surely returning toward pre-pandemic levels of activity, does not make sense.
Many of those forecasters who have been taking these data at face value have argued virtually constantly since about the end of June that the labor market is stalling out and that the next payroll figure would potentially be negative. As we know, that argument has fallen flat every month so far, and it is likely to do so again in October.
The continuing claims figures may be less distorted, but they are more difficult to parse, especially lately. The levels have been implausibly high for months. For example, at the time of the May employment survey, the combined count of people collecting benefits in the regular and PUA programs was about 32 million, while the Bureau of Labor Statistics reported 21 million unemployed. In July, the figures were 30 million and 16 million, respectively. Note how the continuing claims tally was falling much more slowly, which might be explained by a rising number of people collecting benefits who should not have been.
Mid-September marked the six-month mark since the first wave of lockdown-related layoffs. Thus, an increasing number of beneficiaries are running out of benefits through the regular and PUA programs and, if they are still out of work, being shifted to the PEUC and Extended Benefits programs. To get an accurate picture now, we have to add in the number of continuing claims for all four programs. At the time of the September employment survey reference period, the combined total was 26.7 million, but the BLS reported only 12.6 million unemployed for last month.
Not only have the continuing claims figures consistently overestimated the level of unemployment, but the gap has been widening since the peak in joblessness was reached. Even so, there is good news, as the combined tally of beneficiaries for the four programs combined moved down by 3.3 million in the first three weeks of the five-week span from the September to the October employment survey reference period.
Even these data do not portray a labor market that has stagnated. Various survey and anecdotal evidence is also consistent with ongoing gradual recovery in employment, including the Beige Book released a few days ago. Some of the drop in continuing claims may reflect the fact that California, and perhaps other states, have begun to sift out the false and fraudulent claims, so it is still hard to take the data at face value. However, they are consistent with what most other sources of information suggest, which is that the labor market recovery, while not what we would hope, continues at a pace that is faster than most predicted back in the spring.