Unemployment Bonus: An Update

Last week a story made the news about how the termination of the $300/week unemployment bonus has not had any positive effects on employment. Written by a group of journalists with the Associated Press, the story claims:

Earlier this year, an insistent cry arose from business leaders and Republican governors: Cut off a $300-a-week federal supplement for unemployed Americans. Many people, they argued, would then come off the sidelines and take the millions of jobs that employers were desperate to fill. Yet three months after half the states began ending that federal payment, there’s been no significant influx of job seekers.


In states that cut off the $300 check, the workforce ā€” the number of people who either have a job or are looking for one ā€” has risen no more than it has in the states that maintained the payment. That federal aid, along with two jobless aid programs that served gig workers and the long-term unemployed, ended nationally Sept. 6. Yet America’s overall workforce actually shrank that month.

The implication is that states should have continued to pay out the bonuses.

Unfortunately, the actual numbers and trends on the labor market do not accord with this story. To start with the last point they make, namely that the workforce shrank in September: the fact of the matter is that the workforce always shrinks in September. The reason is simple: people who work seasonal summer jobs leave them and, e.g., go back to school. Therefore, to use this statistical regularity as an argument against terminating the unemployment bonus is a bit like suggesting that you are tired today because the sun rose this morning.

Not only do the AP journalists fail to control for regular variations in employment and workforce participation, but they also fail to check how the labor market in general has responded to the bonus terminations. As an example of how you actually have to read this type of data, back in August I reported that for the month of June,

there was a larger increase in unemployment in withdrawal states: they saw the number of unemployed rise by 16.2 percent, compared to 12.7 percent in bonus states. … The list is topped by bonus-remaining states Vermont and Kentucky, which are followed by withdrawal states New Hampshire, Alabama, North Dakota, Ohio, Mississippi, South Carolina, Montana and Tennessee. … the question comes into mind why unemployment would rise more in withdrawal states than in those that remained in the bonus program. Reasonably, this is explained by the fact that in order to get a job, one first has to start looking for a job; a person actively seeking employment is categorized as involuntarily unemployed and therefore registered as such in labor-market statistics. In other words, it is quite plausible that unemployment rises because more people in withdrawal states consider the loss of the bonus a good enough reason to leave the ranks of the non-employed.

I also explained, with reference to the month of July (emphasis added):

the withdrawal states saw private-sector payrolls expand by 0.8 percent while bonus states averaged a 0.9-percent growth. Workforce participation increased by, respectively, 0.1 and 0.5 percent, with the latter group’s numbers being pulled up by an unusually high 3.2-percent increase in Connecticut. A total of 26 states saw workforce participation increase between 0 and 1 percent, with this group being equally divided between withdrawal and bonus states. The big difference shows up in unemployment numbers. In July, withdrawal states experienced a decline of the number of unemployed individuals by an average of 13 percent, while the bonus states saw a drop of 5.2 percent. Of the 25 states with the largest decreases in the ranks of the unemployed, 19 are withdrawal states.

In other words, in June people started to actively look for work in states that withdrew from the bonus; by July they had either left the states where they were unemployed, or actually found jobs.

Another possibility is that many people who had left the workforce altogether decided to register as unemployed in June in order to cash in on the benefits before they disappeared. Depending on how unemployment-benefit agencies track the unemployed, the termination of the bonuses led to either of these two changes in behavior:

  • Either people who had not been looking for work were able to cash unemployment benefits and bonuses anyway and therefore decided to go look for work over the summer;
  • Or the ranks of the unemployed filled up in response to the pending termination of the benefit.

There is a third possibility that straddles these two scenarios: employed individuals register as unemployed in a neighboring state. This is a speculative suggestion, but it has some merit. Ever since the unemployment bonuses were introduced, we have seen some major statistical anomalies in the data over benefits payouts and the number of unemployed. For example, in the months of May, June and July 2020, the total amount of unemployment bonuses paid out was so high that when divided by the number of officially unemployed, the average individual received more than $5,000 – per month. In January this year that amount was almost $4,400.

It is unreasonable to imagine that the average unemployed person collected these amounts: even with the $600 weekly bonus, the numbers do not add up. A more likely explanation is that there were systematic irregularities in the benefits system, encouraged by the size of the bonus and the rapid rise in unemployment registrations. In fact, the Reason Foundation reported back in February of “a mind-boggling amount of unemployment fraud” as a result of the bonuses.

As the bonuses taper off, it is only logical that excessive unemployment registrations – for fraudulent purposes – also taper off. Unfortunately, these instances of fraud can easily make their way into official statistics, giving us an inaccurate picture of what is actually happening in the economy.

The AP reporters do not even touch upon this aspect of the unemployment numbers. This is unfortunate, especially since a drop in fraudulent unemployment filings would be consistent with one of the accurate observations they make: that there is no difference in employment trends between bonus states and withdrawal states. If a large number of those registered as unemployed actually were not unemployed, the usual negative correlation between employment and unemployment breaks down.

Once the fraud has been weeded out of the system, we can once again get a more accurate picture of the facts on the ground. It is fair to assume that by the month of September, the bonus-motivated fraud was largely gone and numbers back to “normal”. This makes the following numbers particularly interesting:

  • In the states that held on to the unemployment bonuses for as long as possible, by September the average unemployment rate was 4.7 percent; workforce participation stood at 59.2 percent;
  • In the states that ended unemployment bonuses during the summer, unemployment in September averaged 3.2 percent, while workforce participation was 60.2 percent.

The states that did end the bonuses early have an unemployment rate that is 1.4 percentage points lower than the states that kept the bonus as long as they could. The fact that workforce participation is a full percentage point higher in states that pulled out of the bonus, is not insignificant. At the national level, a one-percent increase in the workforce represents the addition of roughly 1.5 million workers available to take a job.

When higher workforce participation is combined with lower unemployment, it means that people not only go look for work, but also find it. The aforementioned AP story leaves these facts out entirely, raising questions about the integrity of the journalism that went into it.