Do you have a fever? Have you had a cough, headache or chills? With hope for a COVID vaccine but no definite time frame, and dayslong lags in COVID testing, many employers and other organizations are relying on self-reported screening to help stop the spread of COVID. Although recent research (see this NEJM study or this CDC report) raises questions about the effectiveness of these screens, they are still being widely used. So, what do we know about their effects on both workplace safety and productivity?

A recent working paper by Krista Ruffini, Aaron Sojourner and Abbie Wozniak provides detailed new evidence on self-reported screening. They use data from this spring’s COVID Impact Survey, which asked 8,500 people about their temperature, symptoms and COVID status of the people they lived with. While they didn’t capture whether or not the person being screened actually had COVID, the researchers used this information to figure out how many workers would be told to stay home under different screening criteria that employers might use.

The main finding is that many workers would screen “positive” on a given day. For instance, temperature checks, a simple but potentially unreliable measure, would flag 4% of respondents with a temperature above 99 degrees. But the key practical finding is that different screening approaches produced wildly different results. When “positive” means reporting two or more symptoms from a three-item list, 2.5% of respondents are positive. When the symptom list includes seven items, though, 26% of respondents are positive! The researchers point out that workers may be more or less honest about their symptoms when their answers would have actual consequences like infecting their coworkers or losing income.

These findings tell us that workplace COVID screening could be fairly disruptive, but some of the most interesting results are on the interaction between screening and workers’ rights. Screens that only use temperature checks do not disproportionately affect legally protected groups (defined by race, ethnicity, age, sex or health status), but they are more likely to miss people who are sick. Screens that use symptom lists are less likely to miss people (and may in fact send home a lot of healthy workers), but they could lead to quite different income losses for women and young workers, who tend to report more symptoms.

The authors speculate that the success and consequences of workplace COVID screening may depend on income support policies that lessen the incentive to hide symptoms and support workers who lose a week’s wages because they were truthful about their symptoms.

Andrew Goodman-Bacon is a senior research economist with the Opportunity and Inclusive Growth Institute at the Federal Reserve Bank of Minneapolis. His research focuses on policy issues related to labor, demography, health and public economics.