• Over 70% of workers in Accommodation and Food Services, Health Care and Retail Trade work in occupations with high physical proximity to others
  • 10% or fewer workers in the office-based sectors of Finance, Management, and Professional Services work in high-proximity occupations
  • The largest percentage job losses during the pandemic came in economic sectors with large shares of high-proximity workers
  • Within economic sectors, job losses were larger for subsectors with large shares of high-proximity workers, subsectors that are closely connected with high-proximity sectors, and subsectors that provide discretionary goods and services


A defining feature of the recession driven by the coronavirus (COVID-19) is social distancing and its negative impact on goods and services bought or administered in-person. In this report, we use a measure of physical proximity by occupation to calculate how many workers in each economic sector tend to work in close proximity to others. We show that sectors with many high-proximity workers experienced the greatest job losses. A second important factor for job losses is how discretionary a sector’s products are. Job losses were much larger for sectors with services that can easily be postponed or canceled.

To further illustrate these points, we also break out employment and proximity data by subsector. Often, within a sector, some industries have higher shares of high-proximity workers than others or else provide more discretionary services. We show strong relationships between those two factors and greater employment losses.

Economic Sector, Physical Proximity, and Job Losses

What economic sectors have the most workers in close proximity with others?

In Figure 1, the horizontal axis shows the share of workers in each sector (pre-pandemic) who work in high-proximity occupations. Accommodation and Food services, Health Care and Social Assistance, Retail Trade, Arts, Entertainment and Recreation, and Educational Services all have more workers in high-proximity than low-proximity occupations. Office-based sectors like Professional Services, Management, Finance and Insurance and Information all have very low shares of high-proximity occupations.

We define this measure using the publicly available Occupational Information Network (O*Net) program database. The O*Net database, sponsored by the US Department of Labor/Employment and Training Administration, characterizes several hundred occupations by measures of required skills, knowledge, and work environment characteristics. In each area, the program combines survey responses to give each occupation a score. We combine the O*Net physical proximity score with Bureau of Labor Statistics data on employment by occupation and industry as of May 2019 to map the number of workers with high physical proximity to others in each sector and industry. For simplicity, we report the share of workers in each sector with an above-median physical proximity score.

Employment declined 13% between the February and May job reports, with the primary employment losses coming between mid-March and mid-April. The vertical axis of Figure 1 shows the percentage of jobs lost by sector. We use seasonally unadjusted data in this report, so all job losses or gains shown in Figure 1 or the discussion below reflect actual changes. The lone exception is for Construction, which we manually adjust in Figure 1. For the February to May period under consideration, the only area in which using unadjusted data significantly distorts the employment picture is Construction, where employment significantly increases in the spring and summer months.

The biggest percentage job losses came in sectors with a very large share of high-proximity workers: Arts, Entertainment and Recreation and Accommodation and Food Services. These same sectors, along with Construction and Other Services, added the highest share of employment between April and May as businesses in many areas began to reopen. Despite the good news in May, the partial recovery still left these sectors well below pre-pandemic levels of employment.

However, not all high-proximity sectors experienced large job losses. Health Care and Social Assistance has a very large concentration of high-proximity occupations, such as nursing. Retail Trade also has a high share of high-proximity workers, including millions of cashiers, clerks, and salespersons. Perhaps it is not a surprise that Health Care employment would not fall much during a pandemic, but why were Retail Trade’s cumulative percentage job losses of 12% slightly less than overall percentage job losses?

The answer is that the most at-risk jobs during the pandemic were those that require a high degree of physical proximity and are not essential services. At the aggregated level, most sectors—including Health Care and Retail Trade—are combinations of industries where spending can be deferred, and of industries that provide essential goods and services whose consumption cannot be put off.

Key Subsectors, Job Loss, and Essential Services

We can show the relationship between physical proximity, discretionary spending, and job losses by looking at different patterns within some major sectors. There are a few major reasons why sectors’ employment losses vary significantly across subsectors:

We will discuss examples in accommodation and food services, Health Care, Retail Trade, manufacturing, and other services.

Accommodation and Food Services. Restaurant and hotel employment have both plummeted during the pandemic. At food services and drinking places, 88% of workers are in high-proximity occupations, the highest of any subsector. Interestingly, fewer than half of Accommodation workers are in high-proximity occupations. Maids and housekeeping cleaners is the largest occupation by employment in Accommodation. While there are certainly a number of public-facing jobs as well, Accommodation’s huge job losses come more from the halt in travel than the essential nature of their business.

Thus, while Accommodation is part of the Leisure and Hospitality sector, its employment path may be more like Air Transportation than like restaurants or retail. A restaurant or shopping mall depends on consumers who are allowed and willing to leave the house. But the Accommodation subsector relies on travel. Because of a lower share of high-proximity occupations, hotels may need to make fewer adjustments to job tasks or operating procedures than restaurants and retail stores do, and a smaller share of their workers may be at direct risk of infection. However, they will recover more slowly if traveling and tourism recovers more slowly than local shopping.

Health Care. In Health Care, all subsectors have very large shares of high-proximity occupations, between 70% and 81%. But April job losses were small in Hospitals and Nursing and Residential Care Facilities and large for Ambulatory Health Care Services and Social Assistance. While most health care jobs involve face-to-face contact, there is a big difference between routine and preventative care that can be deferred and more serious or emergency care that may not. In May, Ambulatory Health Care Services gained back around 400,000 of the 1.2 million jobs lost in April, mostly in dentists’ and doctors’ offices. People are beginning to make the routine visits that they were not making in late March and April.

Retail Trade. All retail subsectors employ a lot of high-proximity occupations: cashiers, clerks, and salespersons. But furniture, clothing, and hobby stores closed down, losing half their combined employment between February and April, whereas grocery stores and general merchandise stores’ employment held steady. As with Health Care, essential businesses kept workers whereas discretionary consumption stopped. These retail subsectors’ employment started to recover in May, growing 16% from the April trough, but is still barely half of what it was in February.

Manufacturing. The lack of demand for certain types of goods played out in Manufacturing as well. Manufacturing has a low share of high-proximity workers across the board, although several large frontline assembly and fabrication occupations have only a slightly below-average score.

Still, working conditions did not play as large a role in Manufacturing employment losses as in many public-facing sectors. Most Manufacturing subsectors experienced smaller job losses than the national average. Food Manufacturing has one of the highest shares of high-proximity workers among manufacturers, 37%, but suffered only a 5% loss in employment. Computer and Electronic Equipment Manufacturing lost only 2% of jobs.

The most notable Manufacturing employment losses came in the subsectors that support markets that were put on hold by the pandemic: Apparel, Furniture, and Transportation Equipment. As discussed above, employment plummeted in clothing and furniture stores due to temporary closings and lack of demand. Transportation equipment employment is mostly auto manufacturing, and new car sales fell by nearly half between February and April.

Other Services. Other Services, as a catch-all sector of jobs that do not neatly fit elsewhere, is the most notable example in which subsectors vary widely by share of high-proximity occupations and job losses. Repair and Maintenance has a relatively low share of high-proximity occupations, and experienced moderate job losses—down 11% from February. Personal and Laundry Services, whose two largest occupations are hairstylists and manicurists, lost half its employment as salons were shut down. This subsector had a 27% bounce-back in employment in May, larger even than the month-over-month growth for Food Services, but employment is still down 42% since February.


Sectors that had large shares of high physical proximity workers and larger shares of discretionary services lost more jobs during the coronavirus recession. In Health Care and Retail Trade, the driving factor was the essential or discretionary nature of services, as almost all their subsectors have a lot of high-proximity workers. In other services, proximity was the key factor, as the high-proximity industries such as hair and nail salons lost the most jobs. In a few industries, changing demand mattered more than the nature of the work. Manufacturing subsectors that produce clothes and cars lost jobs, while those that produce food or computer equipment did not. Hotels employ many low-proximity workers, whereas transportation employs many high-proximity workers, but they both took large hits because of a lack of travelers.

Many of the sectors and subsectors that experienced large losses had partial recoveries in May as businesses began to re-open, but as detailed in the companion section of this QEB, these recoveries were typically less than one-fifth the size of February-to-April job losses.