This morning’s Bureau of Labor Statistics (BLS) data show that total payroll employment contracted by a whopping 701,000 jobs in March, including a loss of 713,000 private-sector jobs. As the coronavirus pandemic ripples through the economy, the monthly employment numbers released today show just the leading edge of the deep recession we are certainly now in (this is due to the fact that these numbers do not capture the entire month of March but only refer to the payroll period containing March 12, before shutdowns accelerated). Nevertheless, this is the first month in nearly ten years—since September 2010—in which the U.S. economy lost jobs. We did see an expected uptick in hiring in the federal government as temporary hiring for the 2020 Census ramped up (+ 17,000).

Despite already being genuinely out-of-date, today’s data provide some important information about the early days of the labor market crisis we are now in, including which sectors were the very first-affected, most vulnerable sectors. In particular, leisure and hospitality lost 459,000 jobs, mostly in food services and drinking places, down 417,000 in March. Further, both average weekly hours and aggregate weekly hours are down for March, as employers began to reduce hours worked and lay off their staff. Average weekly hours has hovered around 34.4 hours per week over the last year and fell to 34.2 hours per week in March, reflecting the first signs of weakness. Notably, average weekly hours fell in leisure and hospitality, from 25.8 in February to 24.4 in March.

Aggregate weekly hours are the product of estimates of average weekly hours and employment. BLS creates an index of aggregate weekly hours by dividing the current month’s estimates of aggregate hours by 2007 annual average aggregate hours. Therefore, when the index is above 100 it means that average aggregate hours are greater than in 2007 and when the index is below 100 it means that average aggregate hours are lower than in 2007. In March, the index of aggregate weekly hours fell 1.1%, reflecting losses in both weekly hours and overall payroll employment.

Nominal wages grew a solid 3.1% over the year. At turning points in the economy, compositional effects tend to swamp any changes in wages within sectors. Since many of the workers who had already lost their jobs by mid-March are in lower wage sectors (a pattern that has likely continued since the reference period), stronger wage growth reflects the dropping of lower wage jobs from the total, which results in higher average wages for the remaining jobs, and what appears to be faster overall growth. This phenomenon occurs at turning points generally, as can be seen by stronger nominal wage growth in 2008 and early 2009 before it fell off sharply by mid-2009.

The household survey shows initial signs of weakness as well. The unemployment rate rose from 3.5% in February to 4.4% in March. This is the highest the unemployment we’ve seen in nearly three years. The last time the unemployment rate hit 4.4% was in August 2017. People working part-time for economic reasons also rose as employers cut work hours; 1.4 million more workers were working part time because their hours had been reduced or they were unable to find full-time jobs.

Labor force participation and the employment-to-population ratio fell in March, dropping 0.7 and 1.1 percentage points, respectively. If participation hadn’t fallen, the unemployment rate would be even higher. During the COVID-19 pandemic and this period of social distancing to flatten the curve, I expect the unemployment rate to understate weakness as would-be workers follow the advice of health experts and stay home instead of actively seeking other employment (which results in a lower unemployment rate because if an out-of-work individual is not actively seeking work, they are not counted as unemployed). That’s why, when policymakers determine when to turn off any relief or stimulus, they should use measures aside from the unemployment rate, such as the employment-to-population ratio as the policy trigger.

We won’t see the devastating job losses or large reductions in hours from COVID-19 in the employment data until the next report on May 8. Then we will see what’s happened across the economy in terms of not only total employment, but also important labor market measures such an unemployment, underemployment, and labor force participation across and within demographic groups such as age, gender, race and ethnicity, and education.

Today’s data, combined with yesterday’s unprecedented unemployment insurance claims, show that Congress must act quickly to mitigate as much of the economic harm from the coronavirus as possible. EPI estimates that nearly 20 million jobs could be lost by July, and many more could be lost if further action isn’t taken. Congress should now turn to crafting another relief package that includes a large and much-needed infusion of state and local aid, continues cash assistance to households beyond a single payment, and extends the unemployment insurance provisions in the earlier bill with explicit triggers-off tied to economic conditions. It should also provide additional funds to states that create or strengthen programs that incentivize employers to keep workers on payroll—rather than laying them off—through better “short time compensation” capability, so as few families as possible face the near- and long-term consequences of job loss and so that workers are ready to jump back in to work as soon as the lockdown is over.