Normative Narratives

What is the “True” Unemployment Rate

Weekly Unemployment Insurance Report and August Jobs Report: Update (9/17):

Unsurprisingly, weeks of higher initial claims numbers have manifested themselves as weeks of higher continued claims. The most recent little blue spike in the initial claims numbers corresponds to the recent uptick in the continued claims bar chart.

As a lead indicator it is heartening to see initial claims are down a bit this week. Having said that one week doesn’t make a trend, and it’s hard to see any evidence of a sustained recovery, nor any reason to think there will be one in the near future as the parties abandon stimulus talks and enter the election home stretch.

It is worth noting that the NYT published an article explaining some flaws and misconceptions of this weekly UI data, notably:

“That figure is often treated by economists as an estimate of the number of people receiving unemployment benefits. But that isn’t actually what it measures, at least not directly. It counts applications, not all of which are approved. And rather than counting the number of individuals applying for benefits, it counts the total number of weeks of benefits they apply for.

That distinction doesn’t matter much in normal times, when most people apply for benefits on a weekly basis and are quickly approved. But because benefits are paid retroactively, if there are delays processing applications, people can end up applying for multiple weeks of benefits at once, skewing the continuing-claims number.”

These numbers are still useful. but clearly imperfect. The Current Population Survey’s (“household survey”) small sample size and possible misclassification errors also make it unreliable right now–8.4% unemployment feels a little too good to be true right now, right?

Due to it’s larger sample size, the employment numbers (Current Employment Statistics, or “establishment survey”) are the most reliable gauge of labor market strength right now. That shows us still 11.5 million jobs below the pre-pandemic peak as of the week ending August 15th.

Which brings us to the Jobs Report. Lets first take a look at the official numbers:

Official August figures (not seasonally adjusted; survey reference week August 9 -15). Here is a link to the report, and this month’s accompanying FAQ:

U-3: 8.5

U-6: 14.3

“True” August figures

U-3 (unemployed + OTY change “employed–not at work, other reasons” + OTY change in “not in labor force, want a job now” / labor force + OTY change in “not in labor force, want a job now”); 16.7/162.9 = 10.3%

U-6 (true U-3 numerator + plus part time for economic reasons): 24.2 / 162.9 = 14.9%

Alternate measure of labor force weakness: # claiming UI insurance / labor force + OTY change in not in the labor force, want a job (29.2 / 162.9) = 17.9%

The official U-3 rate remains significantly higher for blacks (13.1%), latinx (10.5) and Asians (10.7) than whites (7.5). On the employment side, the 1.4 million jobs gained was buoyed by almost 240,000 temporary Census jobs. This recovery has been uneven and is clearly losing steam, as the economy remains at least 11.5 million jobs below it’s pre-pandemic peak, and likely much more considering UI numbers.

July Jobs Report:

Moving to the Jobs Report, lets first take a look at some baseline numbers:

Official July figures (not seasonally adjusted; survey reference week July 12 -18). Here is a link to the report, and this month’s accompanying FAQ:

U-3: 10.5

U-6: 16.8

“True” July figures

U-3 (unemployed + OTY change “employed–not at work, other reasons” + OTY change in “not in labor force, want a job now” / labor force + OTY change in “not in labor force, want a job now”); 20.9/164.1 = 12.7%

U-6 (true U-3 numerator + plus part time for economic reasons): 29.5 / 164.1 = 18.0%

Taking a deeper look at the report, the official Black (15.0%) and Hispanic (13.0%) unemployment rates remain much higher than headline number, or the figure for White people (9.4%). These historically last hired, first fired groups will not realize economic or racial justice whenever growth resumes as Trump suggests, we need a real plan to address these issues.

The economy added 1.8 million jobs in July, pointing to a significant slowdown from June when it added 4.8 million. Only 9.3 million jobs have been added of the 22 million lost during the pandemic, so this slowdown is troubling.

This increase also overstates what’s actually happening. According to a Reuters analysis “Government employment increased by 301,000. The model that the government uses to strip out seasonal fluctuations from the data normally anticipates education workers to drop off payrolls in July. This, however, happened earlier because of the pandemic, leading to a big gain in July.”

There is a difference between saying 1.4 million government jobs were lost in March – June, but 300,000 were added in July, and saying 1.1 million government jobs have been lost and more losses are expected. As the official figures stand, one may assume a recovery in government employment, but this is not the case. This is particularly concerning as the fight over the next COVID stimulus in Congress hinges largely on the issue of providing aid to state and local governments.

Furthermore, the reference period for the July report (July 12-18) is before the $600 / week enhanced UI benefits expired. US GDP is about 67% consumption based, and about 28 million Americans are collecting benefits right now. Low income household spending, buoyed by enhanced UI benefits, has kept the economy afloat these past few months (we still lost 8.2% of GDP in the 2nd quarter even with it). Without this prop, aggregate demand will fall precipitously, taking lots of jobs with it and having a deflationary effect on the overall economy. This could be the beginning of that W shaped recovery or double dip recession you may have heard of.

Macroeconomics aside poorer people need this money for necessities–without it poverty will rise.


June Jobs Report Update:

I have recently discovered more information at the end of the Department of Labor’s weekly UI claims report. Apparently the headline initial claims number only covers state UI claims. In the chart below, weekly state UI coverage is in blue, whereas federal coverage is in red. (The other chart shows initial state UI claims per week; I will update both of these charts every Thursday from hereon out).

Starting around mid-April, states began reporting data on the Pandemic Unemployment Assistance (PUA) program to the DOL. This is the program that covers gig workers and others normally not covered by UI. As you can see, in recent weeks this program has made up a significant portion of total covered people. The most recent week there is data for, the week ending June 13th, has the following breakdown:

Total State UI coverage: 17,654,303

Total UI coverage (state and federal): 31,491,627

Focusing on just the state numbers understates the depth of the problems facing the labor market, as well as the amount of relief the U.S. government is providing workers. In a moment when I compare the overall number of people covered by UI with the broadest measure of labor underutilization–the “true U-6”–we will get a more meaningful idea of how well the government has aided the American worker (the looming expiration of the $600/week Pandemic Unemployment Compensation supplement at the end of July notwithstanding).

The week ending June 13th is also the reference period for the June Jobs report. See below for updated “true” unemployment rates:

Official June figures (not seasonally adjusted; survey reference week June 7-13). Here is a link to the report, and this month’s accompanying FAQ:

U-3: 11.2

U-6: 18.3

“True” June Figures

U-3 (unemployed + OTY change “employed–not at work, other reasons” + OTY change in “not in labor force, want a job now” / labor force + OTY change in “not in labor force, want a job now”); 22.9/163.8 = 14.0%

U-6 (true U-3 numerator + plus part time for economic reasons): 32.2 / 163.8 = 19.7%

Notably, these rates are higher for certain racial groups. African Americans and Latinx workers have the highest rates by race, followed by Asians. This is a surprising reversal of fortune, as Asian Americans have traditionally had the lowest unemployment rate by race. Perhaps there has been some anti-Asian backlash related to the origins of COVID-19 playing out in the labor market.

Other figures:

Employment – Population ratio: 54.9%

% unemployed covered by UI (Total UI coverage, state and federal / true U-6 number of labor underutilization): 31,491,627 / 32,200,000 = 97.8% !

Aside from the undocumented workers not covered by UI benefits, America has done a very good job of keeping it’s unemployed and underemployed covered. This reflects both the expansion of UI coverage under the CARES act and that state UI offices have now had time to work through their claims backlogs.

But the underlying conditions needed for a true economic recovery haven’t changed; we need to get this virus under control, or else this aid will become [even more] unsustainably expensive. Unfortunately we have done very poorly on that front.

I have decided to continue to update this post with the release of every month’s jobs report for the forseeable future. The updates will be very brief, just a few numbers. For methodology and an explanation of my thinking, see the original post.

There is one caveat–a figure I included in the “true U-6” last month likely should have also been included in the “true U-3”–the over the year change in persons “not in the labor force, want a job now”.

It seems unlikely to me that these are people who were not working and not looking for work, but now in the middle of a pandemic decided they want to start looking for a job. Instead, I think they are another group of misclassified unemployed.

Again, all figures are not seasonally adjusted

Official May figures (survey reference week May 10-16):

U-3: 13.3

U-6: 20.7

“True” May Figures

U-3 (unemployed + OTY change “employed–not at work, other reasons” + OTY change in “not in labor force, want a job now” / labor force + OTY change in “not in labor force, want a job): 18.1

U-6 (true U-3, plus part time for economic reasons in numerator): 25.2

Other figures:

Employment – Population ratio: 52.9%

% unemployed covered by UI: 65%


Disclaimer: I work for the Bureau of Labor Statistics. The views here are my own and do not reflect the views of the Bureau of Labor Statistics or U.S. Department of Labor. Furthermore, they should not be interpreted as a disagreement with what the Bureau reported. Indeed, it is only because of the transparency of the Bureau’s reporting that I was able to put together this analysis (all numbers in the analysis below come from the link above or the Employment Situation news release, unless otherwise noted).

The Bureau has good reasons for not correcting likely misclassified survey responses. As a private citizen, I simply have more leeway in my analysis than the famously impartial and consistent BLS does.

Note: All data used in this analysis is not seasonally adjusted


For those who follow such things, the April Employment Situation news release (“Jobs Report”) was perhaps the most anticipated in history. Never has such a dramatic over-the-month change in the U.S. economy been recorded. The Great Depression predates the Bureau’s employment numbers, and even if it didn’t it was more of a sustained contraction over time than an abrupt shutdown by design.

The official unemployment rate (U-3) came in at 14.4% as of the week ending April 18th (the data for this month’s release were based on a survey conducted April 12th-18th). This is calculated by dividing the number of unemployed people (22.5 million) by the number of people in the labor force (the unemployed and the employed, 155.8 million). Typically to be considered “unemployed”, one must be out of work and have actively looked for work in the past four weeks. This requirement has been relaxed somewhat to account for the realities of COVID-19, most notably to include people who are temporarily laid off but expect to be recalled, and therefore not actively looking for work. In fact this group represents the vast majority of those currently considered unemployed (17.9 of the 22.5 million), offering a ray of hope that some of them may indeed be recalled.

That figure, however, likely should’ve been even larger. According to the Bureau, it is likely that 7.5 million people should’ve been included in this group, but were instead considered “employed–not at work, other reason” (based on the fact that there are 7.5 million more people in this group now than there were a year ago). If you add this group to the unemployed, you get an unemployment rate (call it the “true U-3”) of 19.3% (22.5 + 7.5 / 155.8).

But many people believe the official unemployment rate doesn’t really capture the state of the labor market. These people often turn to an alternate measure, the U-6, to determine a “true rate” of labor underutlization. We do not call call the U-6 an “unemployment rate”, because as you will see in a moment it includes some part time workers.

The U-6 is the number of unemployed (22.5 million) plus the number of people working “part time for economic reasons” (would rather work full time but can’t find a full time job – 10.7 million) plus those “marginally attached to the labor force” (want a job, available to take a job, have looked for a job in the last 12 months but not in the last four weeks (2.2 million)), divided by the labor force (156.5 million) plus the marginally attached (2.2 million). In April, the official U-6 rate was 22.4% (22.5 + 10.7 + 2.2) / (155.8 + 2.2).

If we add those 7.5 million people that likely should’ve been coded as unemployed, we get a U-6 of 27.2%. But there was also a large increase in the number of people who are not in the labor force who currently want a job, but do not meet the criteria to be considered “marginally attached”; there were 4.8 million more in April 2020 than there were a year earlier. While the Bureau did not note this as a potential survey error, I think it is fair to say most of these people should have been included in that “marginally attached” figure. Lets replace the “marginally attached” with this broader figure of people who are not in the labor force but want a job. Using these figures, we end up with a “true U-6” of 28.3% (22.5 + 7.5 + 10.7 + 4.8) / (155.8 + 4.8). This 28.3% is a good estimate of labor underutilization (or at least was as of the week ending 4/18).

No matter how you slice it the economy is in bad shape right now. Even taking that conservative “true U-3” number of unemployed (22.5 million, and the 7.5 million misclassified as employed), we have 30 million people. At the same exact point in time (the week ending April 18th) the Department of Labor reported that 17,776,006 people were covered by unemployment insurance (UI). In other words, only 59% of unemployed people were receiving benefits. In fact, since some states are offering partial UI for people with reduced hours (who would not be counted in those 30 million unemployed), the true percentage of unemployed people receiving benefits could be significantly lower.

Things have only continued to deteriorate since the jobs report survey was taken. From that point to the week ending May 2nd (the most recent data), 6,338,351 new people have filed for UI. While weekly initial claims are on a downward trend, they are still at very high levels and could unfortunately plateau there if reports of unfiled UI claims and processing backlogs are true.

If all of these different unemployment rates (official and otherwise) are confusing, here is a number that cuts through the fog: only 51.3% of America’s working age population (16+) is employed, the lowest rate since the Bureau began tracking that figure in 1948. 

The U.S. economy is almost 70% consumption based. Our Federal and State governments must do a better job processing UI claims, keeping people employed, and figuring out some sort of debt forgiveness system while shutdowns are in effect (as Senators Warren and Brown have proposed). If they fail to do these things not only will people suffer, but our economic recovery will be anemic regardless of how soon things open back up.