BIS国际清算银行-Identifying-regions-at-risk-with-Google-Trends_-the-impact-of-Covid-19-on-US-labour-markets_8页_918kb

BIS Bulletin No 8 Summary: Identifying Regions at Risk with Google Trends

Core Content

This BIS Bulletin explores the use of Google Trends data to identify regions in the United States most vulnerable to the economic impact of the Covid-19 pandemic. The study combines pre-pandemic local industry-level employment data with real-time Google search trends to construct a measure of regional employment exposure to the virus.

Key Takeaways

• Covid-19 exposure measure (CV19 exposure) is based on the share of local employment in sectors most affected by the pandemic, such as transportation, leisure and hospitality, and mining/oil and gas.
• Regional exposure varies significantly, with some areas having as low as 2% of employment at risk and others as high as 98%.
• Google search trends for terms like "corona" and "unemployment benefits" are used to validate the CV19 exposure measure.
• There is a strong correlation between CV19 exposure and the volume of Google searches, indicating that the measure can effectively predict regional economic vulnerability.

Main Points

1. Regional Exposure to Covid-19

• CV19 exposure is calculated for 209 Designated Marketing Areas (DMAs), which are geographic regions comprising multiple counties with similar media consumption patterns.
• The most affected industries include transportation, employment services, leisure and hospitality, and travel agencies, which are directly impacted by containment measures.
• Mining/oil and gas is also a high-risk sector due to the pandemic's effect on global demand and oil prices.
• On average, 34% of jobs in DMAs are at risk, with exposure ranging from 19% to 54%.
• Highly exposed areas include Las Vegas (NV), Atlantic City (NJ), and Midland (TX).

2. Validation with Google Trends

• Google search data is used to validate the accuracy of the CV19 exposure measure.
• The search term "corona" is used to gauge public interest in the virus, while "unemployment benefits" and "EDD" reflect the actual impact on local labor markets.
• "Corona" searches increased sharply from mid-February 2020, aligning with the rise in active cases.
• "Unemployment benefits" searches surged over eightfold in relative terms by mid-March 2020, coinciding with reports of job losses.
• DMA-level regressions show a strong positive relationship between CV19 exposure and search volumes for "corona" and "unemployment benefits".

3. County-Level Exposure

• The study extends the analysis to county-level exposure, revealing that county exposure ranges from 2% to 98%.
• Texas counties are particularly exposed due to the mining and oil industry.
• Florida and Hawaii counties also show high exposure due to the tourism sector.

Conclusion

• The employment-based CV19 exposure measure is a valid and effective tool for identifying regions most affected by the pandemic.
• This method provides granular insights into regional economic impacts, which is crucial for targeted policy responses.
• The integration of official statistics and real-time data from non-traditional sources like Google Trends can enhance policymakers' understanding of the heterogeneous effects of the pandemic and improve their response strategies.

References

• Autor, D, D Dorn and G Hanson (2013): "The China syndrome: local labor market effects of import competition in the United States"
• Buckman, S, A Shapiro, M Sudhof and D Wilson (2020): "News sentiment in the time of COVID-19"
• Doerr, S (2019): "Unintended side effects: stress tests, entrepreneurship and innovation"
• Jun, S, H Yooab and S Choi (2018): "Ten years of research change using Google Trends"
• Muro, M, R Maxim and J Whiton (2020): "The places a COVID-19 recession will likely hit hardest"
• Zandi, M (2020): "COVID-19: a fiscal stimulus plan"

Previous Issues