Even with already-passed relief and recovery measures, job losses from the coronavirus shock could easily exceed 20 million
This blog post is based on a particular GDP forecast from 3/31. If new and substantially different forecasts are released, we will update these numbers.
The effect of the novel coronavirus and the public health measures enacted to slow its spread—particularly “social distancing”—have been profound for economic activity in the United States. We have already seen some of the leading edge of this effect in recent data releases. This post highlights some recent forecasts of the effects of the coronavirus shock on measures of economic activity, and what these contractions in economic activity mean for jobs. Its key findings are:
- Forecasts of the size of the drag on growth imposed by the coronavirus (and associated public health measures) have risen rapidly in recent weeks.
- Currently, forecasts indicate that gross domestic product (GDP) will be roughly 12.8% smaller by the end of June 2020 due to coronavirus effects—an extraordinarily fast economic collapse.
- To return to pre-shock economic health by the middle of 2021, at least $1.4 trillion in additional recovery spending would be needed—with $2 trillion being a more prudent target.
- A contraction this rapid would be consistent with job loss of roughly 19.8 million by the end of June 2020.
- This estimate of 19.8 million jobs lost could be too high or too low, for a number of reasons. For example, employers could reduce hours instead of laying off workers, and policies to keep workers on the payroll may be effective, keeping job losses down. Or coronavirus effects may be concentrated in relatively labor-intensive industries, so employment might actually fall faster than GDP. A plausible range is between 18 and 28 million jobs lost.
On March 15, Goldman Sachs forecast that GDP would contract at a 5% annualized rate in the second quarter of 2020. On March 20, their forecast jumped to 24% annualized contraction. And on March 31, their forecast for second quarter contraction grew to 34%. This pattern of rapid deterioration in projected coronavirus-related contraction has been true across literally every other forecaster. The logic of these forecasts of rapid and historically large collapses in GDP is easy to see. Even a 5% across-the-board contraction in consumer spending over a short period of time (say because a housing price bubble popped) can send the economy into a steep recession. But the coronavirus is causing well over half of all economic activity in some major sectors to stop dead. For example, accommodations and food service by itself accounts for 14% of all consumer spending. If economic activity in just this sector is cut in half, this alone can drive a steep recession. But, of course, other sectors are also affected and contraction in some sectors will be well over 50%.
Because these forecasts are expressed as quarterly changes in GDP expressed at an annualized rate, they can be slightly confusing. You can’t quite just divide the annualized rates of change by four to figure out how much smaller GDP would be in a given quarter, but you get close to the right answer doing that. The figure below shows GDP if it had grown at a trend rate of growth of 2% (roughly what has characterized recent years’ growth) versus what the latest projection from Goldman Sachs shows. The difference between the two is explained in the chart note.
Projected GDP growth: Pre-coronavirus trend and Goldman Sachs forecast
|Projected (post) coronavirus growth
Note: Pre-coronavirus trend assumes 2% growth in real (inflation-adjusted) GDP, a rate consistent with long-run trends in productivity growth (1.5%) and projected growth in the labor force (0.5%).
Source: Author’s analysis using data from Goldman Sachs U.S. Economics Analyst newsletter (not publicly available)
By the end of the second quarter (June), GDP is 12.8% smaller than it would have been absent the coronavirus shock. Crucially, even with fast projected growth at the end of this year, the economy remains 5.1% of GDP beneath its pre-crisis path by the middle of 2021 (and still 4% beneath at the end of 2021).
This forecast accounts for the CARES Act and even assumes a fourth, so-far-unpassed recovery package that contains substantial aid to state and local government (a particularly effective form of fiscal support). This implies that to return the economy to full potential by the middle of 2021, another recovery package that contains “substantial” state and local aid, as well as 5% of GDP on top of that, is needed. If the baseline aid to state and local governments assumed in the Goldman Sachs forecasts was $250 billion, this would imply roughly another $1.4 trillion is needed between the end of the CARES Act and mid-2021 to engineer a full recovery by then.
This additional $1.4 trillion in aid needed gets substantially larger, however, if we do not actually see the extraordinarily fast 15% growth in the last half of 2020 that Goldman Sachs is currently projecting. Prudence would argue for a recovery aid package more on the order of $2 trillion. What we’ve learned from the recovery from the Great Recession is that demand growth needs substantial help from policymakers—there are long-run trends in the economy suppressing this growth, so we should always aim to overshoot on recovery spending.
In regard to what this means for employment, if the economy is producing 12.8% less in terms of goods and services, it should be using 12.8% less labor, all else equal. Applying this 12.8% decline to the 2019 annual employment level in the Current Population Survey (CPS) and the Current Employment Survey (CES) of 154.2 million gives the 19.8 million job loss figure.
This figure on job losses might be too high—often in recessions it is average hours rather than payroll headcount that are reduced as firms contract (and hours not headcount that rebound when growth starts again). To neutralize the effect of changing hours, one can look at the change in full-time-equivalent employees (FTEs) associated with a 12.8% smaller GDP by the end of June. This translates into roughly 18.1 million FTEs shed due to the GDP decline.
However, much of this margin of adjustment absorbed by hours in recessions is less a function of firms’ response and more a function of which sectors are actually affected. For example, manufacturing jobs are disproportionately lost in most recessions. These jobs have high average hours of work and as they are lost, average hours worked economywide falls. But the coronavirus recession is landing first in service-sector jobs that are low hours on average, so as these jobs are lost, average hours economywide may actually rise, forcing more adjustment to show up in headcount.
Further, in the coronavirus recession, employment may fall more than one-for-one with GDP, as the sectors that are being affected first are extremely labor-intensive. A contrast with manufacturing might again be useful. When manufacturing bears a disproportionate share of job loss in recessions, because manufacturing is extremely capital-intensive, each job lost brings with it a lot of lost GDP. But for each service-sector job lost, much less GDP is brought down. So, any given decline in GDP will lead to more job loss in the service-sector industries that seem to be bearing the first brunt of the coronavirus shock.
Earlier this week we examined initial unemployment insurance (UI) claims by industry from the state of Washington—the epicenter of the coronavirus outbreak in the United States. These claims—which should correlate tightly with layoffs and be a good sign of which sectors will suffer earliest from the coronavirus—skew heavily toward labor-intensive sectors with low ratios of value-added to hours worked. In fact, if we calculate the average labor intensity of the sectors weighted by their share of new UI claims, they are fully 30% more labor-intensive than the economywide average, indicating that job losses associated with a 12.8% decline of GDP by June might be substantially larger than 19.8 million—closer to 28 million.
More hopefully, the historical relationships between GDP and employment loss could be dampened by smart policy, particularly policy that encouraged employers to engage in “labor hoarding” during the economic shutdown—keeping workers on payroll even at zero hours until the economy restarts. In recent decades, U.S. firms have been extremely disinterested in hoarding labor over recessions, instead jettisoning workers at the first sign of distress. This has been one manifestation of typical workers’ lack of labor market leverage in this time. But the recently passed CARES Act includes provisions that could encourage this labor hoarding. For one, firms can reduce hours—sometimes all the way to zero (furlough) and workers can have their wages recouped through UI benefits. Additionally, the $350 billion set aside for small business aid provides a powerful incentive for firms to maintain payroll throughout the crisis—if they do, generous loan terms become even more generous and are completely forgiven. The more firms hoard labor during the economic shutdown, the easier it will be to organize a rapid ramp-up of production once the social distancing measures ease. We should certainly hope these measures work and job losses are less than what the GDP forecasts might imply.
The Goldman Sachs forecasts are consistent with substantial labor hoarding. They see unemployment reaching nearly 15% soon after June—a level consistent with job losses of at least 18 million workers, all else equal. Yet they forecast payroll employment to fall by only 5 million workers in the second quarter of this year. The only way these numbers really hang together is if many workers are on temporary layoff or furlough (and hence are counted as unemployed in household surveys where workers are asked their status) yet still appear on the payroll of firms (which are the surveyed unit for calculating total employment in the United States).
Enjoyed this post?
Sign up for EPI's newsletter so you never miss our research and insights on ways to make the economy work better for everyone.