Editor’s Note: Based on new data released April 2 showing that 6.6 million people filed for unemployment claims in the week ending March 28, The Century Foundation has revised upwards our projections for the true March unemployment rate nationally to 18.3 percent. Updated state projections are available upon request, as well. Please contact [email protected]org with questions or for the updated estimates.


The United States is in the thralls of a devastating jobs crisis. On Thursday, the Bureau of Labor Statistics announced that an historic 3.3 million Americans filed for unemployment insurance (UI) in one week alone—an increase of more than 1,000 percent over the previous week, and nearly five times the country’s previous one-week record set in 1982. Worse, these data tell only part of the problem, as issues with overwhelmed state UI systems and eligibility restrictions that exclude many gig workers and independent contractors from receiving benefits suggest that the true number of job losses is substantially higher. As more businesses are forced to shutter, and with an increasing number of Americans now living under stay-at-home orders, we could potentially see new unemployment claims reach into the millions in the weeks to come. Indeed, the chance of this happening increased substantially on Sunday, when President Trump extended the “physical” isolation requirements through April.

The current jobs crisis is unlike anything we’ve seen before. Past recessions and economic downturns have brought a snowstorm of job losses—a steady fall, spread over many months, if not years, whose impacts we could address in real-time, with existing tools, to mitigate the damage and keep the accumulation manageable. Today, however, we’re facing an avalanche—an unexpected force of nature that comes all at once, one which we are unable to stop and must simply brace for and try to survive. And, as with all natural disasters, this jobs crisis will hit the most vulnerable Americans the hardest—low-income and working families, immigrants and frontline workers, and communities of color.

We would expect this Friday’s jobs numbers to paint the clearest picture to date of just how bad the crisis is and how many workers are already being affected. However, because the March jobs survey closed on March 14, or the week preceding the historic three million-plus increase in UI claims, the data that will be released Friday will understate the true magnitude of the current unemployment crisis. To present a portrait of what the March unemployment rate really looks like, William M. Rodgers III, a TCF fellow and one of the nation’s leading labor economists, modeled potential job losses due to COVID-19 to estimate the nationwide unemployment rate, as well as the rates in each of the fifty states, and broken down by demographic group.

Below are our projections for the true COVID-19 related unemployment rates for March 2020. These projections also provide a forecast of what the jobless rates will be in April. We anticipate the largest one-month spike in the unemployment rate in history, which underscores the critical importance of extending support to working families and those who have recently lost their jobs. The CARES Act passed by Congress last week, which broadened UI eligibility and increased weekly unemployment benefits by $600, along with providing direct cash payments to Americans, was a good and needed first step. But given the scale and scope of the current crisis, much more is needed to give relief to struggling families.

MAP 1. Projected True March 2020 Unemployment Rates, by State

The true national unemployment rate for March could increase to between 15 and 20 percent this week.

Our forecast projects a true March unemployment rate of 17 percent. This will have exceeded the peak of 9.5 percent during the Great Recession, and will be significantly greater than the 4.0-to-4.5 percent rate that we project will be reported Friday.

FIGURE 1

Men, teens, and minorities will bear the brunt of COVID-19 job losses.

We project a true March unemployment rate of at least 25 percent for teens, 19 percent for African Americans, 17 percent for Latinxs, and 16 percent for men. As Rodgers’s previous research has documented, younger workers and African-American workers, in particular, are especially sensitive to labor market shocks, and we expect the same pattern to occur here.

FIGURE 2

The change in state unemployment rates will vary widely state-by-state.

A few states, such as Utah, can expect to see only modest increases in their unemployment rates, whereas the majority of states are projected to see unemployment rates increase by double-digit percentage points.

As measured by percentage point change, we project that the following ten states will have had the largest increases in true unemployment rates from January to March:

  1. Louisiana: 44.9 percent March unemployment rate, up from 5.3 percent in January.
  2. New Hampshire: 44.6 percent March unemployment rate, up from 2.6 percent in January.
  3. Maine: 44.3 percent in March unemployment rate, up from 3.1 percent in January.
  4. Rhode Island: 42.7 percent in March unemployment rate, 3.4 percent in January.
  5. Minnesota: 38.8 percent in March unemployment rate, up from 3.2 percent in January.
  6. Ohio: 36.7 percent in March unemployment rate, up from 4.1 percent in January.
  7. North Carolina: 36.0 percent in March unemployment rate, up from 3.6 percent in January.
  8. Pennsylvania: 34.6 percent in March unemployment rate, up from 4.7 percent in January.
  9. Michigan: 33.3 percent in March unemployment rate, up from 3.8 percent in January.
  10. Indiana: 32.0 percent in March unemployment rate, up from 3.1 percent in January.

We project that the ten states with the lowest true unemployment rates in March will be:

  1. Utah: 2.5 percent March unemployment, unchanged from January.
  2. Georgia: 4.6 percent March unemployment, up from 3.1 percent in January.
  3. California: 6.7 percent March unemployment, up from 3.9 percent in January.
  4. Alabama: 8.1 percent March unemployment rate, up from 2.7 percent in January.
  5. Wyoming: 8.2 percent March unemployment rate, up from 3.7 percent in January.
  6. Vermont: 8.2 percent March unemployment rate, up from 2.4 percent in January.
  7. Hawaii: 8.5 percent March unemployment rate, up from 2.7 percent in January.
  8. West Virginia: 8.8 percent March unemployment rate, up from 5.0 percent in January.
  9. Oregon: 8.8 percent March unemployment rate, up from 3.3 percent in January.
  10. New York: 9.7 percent March unemployment rate, up from 3.8 percent in January.

Methodology

Overall and State Forecasts

To construct the national and state estimates, I created a quarterly state panel data set that follows each state from 1976 to 2019. I then regress a state’s unemployment rate in a particular year and quarter on the natural logarithm of its initial unemployment insurance claims for that quarter and year. Along with the initial UI claim measure, the model also includes year, quarterly and state fixed effects. The regression yields an estimated coefficient of 1.27 which means that a ten-percent increase in a state’s UI claims is associated with an average increase of 0.13 percentage points in the unemployment rate. The regression is weighted by a state’s covered employment.

To construct the predicted March 2020 unemployment rate based on the growth in initial claims for the week ending March 14, I use the Department of Labor’s initial claims filed during the period from March 7 to March 14. This corresponds to when the unemployment data is collected in the Current Population Survey. I multiply the 1.27, the estimated coefficient by the growth in claims. For example, during the week of the CPS survey, seasonally adjusted claims increased by 71,000, for a 34 percent increase. The predicted increase in the unemployment rate equals 0.40 points, the product of 1.27 and 34 percent. This predicted increase is then added to the 3.5 percent current jobless rate.

To construct the predicted March 2020 unemployment rate based on the growth in initial claims for the week ending March 21st, I use the Department of Labor’s initial claims filed during the period from March 14th to March 21st. I multiply the 1.27 by the growth in claims. This predicted increase is then added to the current jobless rate.

Demographic Group Forecasts

To construct the national estimates for each demographic group, I created an annual state panel data set that follows each state from 1999 to 2019. I then regress a state’s unemployment rate in a particular year on the natural logarithm of its initial unemployment insurance claims for that year. Along with the initial UI claim measure, the model includes year and state fixed effects. The regression is weighted by a state’s civilian population.

To construct the predicted March 2020 unemployment rate for each group based on the national growth in initial claims for the week ending March 21st, I use the Department of Labor’s initial claims filed during the period from March 14th to March 21st. I multiply each demographic group’s estimated relationship between the unemployment rate and growth in claims. This predicted increase is then added to the current jobless rate.

Caveats

The estimates do not account for individuals entering or exiting the labor force. If labor force exit exceeds entry, our estimates will be biased upward.

These estimates do not include who shifts from employed full-time to working part-time for economic reasons. To include them, we would need to use the U-6 series that BLS publishes. Unfortunately, it is not available at the state level.

Sources

The quarterly UI initial claims, covered employment, and unemployment rates come from the U.S. Department of Labor’s Office of Unemployment Insurance, https://oui.doleta.gov/press/2020/032620.pdf.

The annual unemployment rates and civilian population data for the various demographic groups come from the Bureau of Labor Statistics, https://www.bls.gov/lau/ex14tables.htm.