Author: Jeff Rosenthal
These days, there are many articles that offer different insights on the impact of COVID-19 on the economy. One way we can assess the impact is to monitor how online job ads change over time and measure the dynamics in labor demand. Online job ads data come with limitations as they can be biased toward some industries and occupations but can sometimes serve as a timely indication from employers responding to COVID-19. This analysis looks at changes in the frequency of online job ads across North Carolina since early March- right before COVID-19 started to impact the state and its economy1.
During the week of May 18-24, 2020, there were much fewer online jobs posted statewide (-7,353) than 11 weeks before. This represents a decline of about 42%. How did the nature of these job postings change at the MSA level?
- Not surprisingly, the largest MSAs (Charlotte-Concord-Gastonia; Raleigh; Durham-Chapel Hill; Greensboro-High Point) continue to have the most job ads and saw the largest numeric declines. As seen in the chart above, their percentage changes are in line with the statewide figure, with Charlotte-Concord-Gastonia down 48.9%, followed by Raleigh (-44.1%), Durham-Chapel Hill (-51.7%), and Greensboro-High Point (-46.4%).
- While two metro areas, Wilmington and Asheville, had greater percentage declines in online job ads a few weeks ago, they now have similar percentage decline as the state and many MSAs.
- Fayetteville appears to have had less of a decline than other regions with 25.2% fewer online job ads in the week of May 18-24 than March 2-8.
Taken together, we find evidence that employer’s demand for workers has declined significantly since early March, as reflected by both the general news events, and the historic numbers of weekly claims. This decline has been reasonably consistent state-wide with most regions showing declines in online job ads of over 33%.
This snapshot shows the current state of the labor demand seen online and is likely to change as the economy adjusts to the effects of pandemic and possible new norms of social activity as a result of COVID-19. Future research will examine the nature of change in the industries and occupations in these online postings.
About the Data
Online job posting data in this analysis comes from the Conference Board®-Burning Glass® Help Wanted OnLine (BG-HWOL), a private third-party data provider with daily and weekly online job postings available at the detailed industry, occupational, and geographic levels. Note that HWOL, like other data sources derived from online job postings, tends to under-represent industries and occupations that are less likely to recruit through online job boards. In addition, due to different methodologies and web scraping techniques, there can be significant difference in numbers of online job ads reported by individual providers. Finally, HWOL ads consider a variety of factors to improve data quality, and their HWOL Index reflects “a robust time series for measuring change in labor demand over time…by reducing volatility and non-economic noise and improving correlation with local labor market conditions”. Although analysis of this index would be ideal, this information lags for a month with its most current month being April 2020 measuring the period March 14, 2020-April 13, 2020. These changes across the state are substantial with declines in the index over this period ranging from as low a decline as 48% for Winston-Salem to 57% for Raleigh. While online job ads can shed some light on employer demand in real time without the delays of official statistics published by governments, we caution data users to utilize this source carefully and use in conjunction with traditional labor market information.
1This is the first full week before Governor Cooper’s State of Emergency Declaration on March 10. The settings for these analyses tried to capture the new openings each week from the Help Wanted On Line universe of sources for ads that had industry and employer classification, which is the recommended approach for analyzing this data. Reported frequencies are those with both industry and MSA classifications, which have lower numbers than those with only industry classification only.