Economic downturns, skill upgrading, and export-led resilience

Non Technical Summary

Economic downturns exert a profound and multifaceted influence on labour markets, reshaping both the quantity and composition of labour demand. In this paper, we leverage an extensive dataset, 122 million online vacancy postings combined with high-frequency geographic mobility data, to analyse medium-term labour market adjustment in China during the COVID-19 pandemic. Our study is among the first to explore these dynamics in a developing country context, where reliable administrative labour market data is often scarce, and where economic structures and adjustment patterns may differ markedly from those observed in developed economies.

We construct our dataset using vacancy postings from three major job boards in China, which are particularly popular among high-skilled job seekers in the service sector. These postings contain detailed information on wages, education requirements, and work experience. We measure skill requirements, mapping occupations to associated skill demands. We further conduct text analysis of job descriptions to extract within-occupation skill content. To ensure representativeness, we compare the geographic, industry, and occupational distributions in our dataset with census and yearbook data, finding strong alignment with broader labour market patterns.

To capture the severity of the recession at the prefecture level, we employ changes in within-prefecture mobility derived from the Baidu Mobility Map. Baidu, a leading provider of mapping and navigation services in China, collects mobility data through mobile phone location services. This allows us to track human movement both between and within cities. During the pandemic, stringent mobility restrictions, including the national lockdown in early 2020 and the Shanghai lockdown in April 2022, sharply curtailed within-prefecture movement. We find that higher local COVID-19 case numbers and more restrictive government policies are strongly associated with greater reductions in mobility. These mobility changes provide us with a reliable proxy for the intensity of economic disruption across regions.

Our analysis documents clear shifts in labour market conditions from 2020 to 2022. In harder-hit prefectures, we observe fewer vacancy postings, fewer firms recruiting, and fewer job opportunities overall. Job requirements became more stringent, with employers demanding higher educational attainment and longer prior work experience. There is also a notable shift toward non-routine cognitive skills, particularly analytical and interactive tasks, accompanied by a decline in demand for routine cognitive skills. This skill upgrading was driven primarily by service sector firms and was most pronounced in second-tier prefectures, which likely possess greater potential for both technological adoption and skill-related advancement compared to Tier-1 urban centers.

The impacts of mobility shocks were uneven across firm types. We find that new entrant firms were disproportionately affected, with hiring activity significantly suppressed during periods of high mobility restriction. Incumbent firms displayed greater adaptability and resilience, maintaining recruitment more effectively under adverse conditions.

We then examine the role of export growth as a buffer during the pandemic-induced recession. Our findings show that rising export demand mitigated the negative impact of mobility shocks on labour markets. Export-driven prefectures experienced more stable aggregate labour demand, with increases in vacancies, postings, and hiring firms. Incumbent firms in the manufacturing sector were able to leverage expanding international markets to offer competitive wages and attract higher-skilled workers.

Our results diverge from those reported in developed economy contexts that have found recessions often facilitate routine-biased technological change through machinery adoption. In contrast, within China during COVID-19, skill upgrading was concentrated in the service sector, and manufacturing exhibited minimal technological change. These differences underscore the importance of considering structural economic distinctions between developing and developed countries when interpreting labour market responses to recessions.

Overall, our study demonstrates that online vacancy posting data, combined with fine-grained mobility information, provides new opportunities to examine labour market dynamics where traditional administrative sources are limited. Our findings reveal that recessions accelerate shifts in skill demand, while export-led growth plays a critical role in cushioning labour demand and stabilizing wages.