Title: Where Low-Income Jobs Are Being Lost to COVID-19? - Monthly updates on the places and industries most at risk

Abstract:

The neighborhoods hardest hit by COVID-19 job losses are home to workers in industries like tourism and transportation, which are bearing the brunt of the economic shutdown. To identify which neighborhoods are most at risk, this project predicts losses of low-income jobs at the census tract and metropolitan area level by industry. The job loss estimates are visualized in interactive web maps and lollipop charts (https://www.urban.org/features/where-low-income-jobs-are-being-lost-covid-19). This feature was first published on April 16, 2020 and we update our estimates every month with BLS national and state employment numbers. These hyperlocal estimates allow users to easily compare estimated job loss across neighborhoods and within industries. We hope that philanthropies, local government agencies, and community residents can use this tool target aid and support where it is most needed.

In order to produce our job loss estimates, we combine data from:

- The US Bureau of Labor Statistics (BLS) Current Employment Statistics data for monthly employment by industry

- The 2014–18 five-year American Community Survey (ACS) IPUMS USA microdata.

- The 2017 Census tract level LODES data

At a high level, the data synthesis process involves applying the BLS data on employment change at all income levels by industry, then adjusting to align to state-level BLS data on employment change and to ACS microdata the Public Use Microdata Area level. Finally we apply the projected job losses by industry and neighborhood to the tract level for low income workers using the Census LODES data. This process allows us to localize employment loss based on where workers live, hopefully giving users a sense of the neighborhoods hardest hit economically by COVID-19.

To correct one-month BLS data lag in most detailed industry observations at the national and state level (for example, in August, the most recent observations available are for the month of June), we project the lagged estimates to the current month by calculating different statistics available for “parent” and “child” industries defined by the North American Industry Classification System (NAICS). We also construct a crosswalk from CES industry classifications to Census American Community Survey industry classifications, using expert judgment.

We hope that by presenting this work, we can partner with other researchers and organizations to improve these estimates over time and use these granular job loss numbers to inform policy solutions.