Untangling US Employment Changes: Production Technology Versus Trade Structure
Summary
This paper uses structural decomposition analysis (SDA) within a global multiregional input–output (GMRIO) framework to split changes in US employment between 1995 and 2008 into contributions from: production technology (changes in production recipes and employment intensities), trade structure (shifts in import/export shares, including the China channel) and consumption. The authors use the World Input–Output Database (WIOD) and complementary occupational data to quantify effects at industry, skill and occupation levels.
Main quantitative findings: trade-structure changes reduced US employment by about 5.5 million, changes in production technology reduced employment by about 34.8 million, while rising consumption added roughly 58.1 million jobs between 1995 and 2008. Manufacturing experienced larger percentage declines than non-manufacturing. Technological change was strongly biased against medium- and low-skilled and routine jobs; trade-structure effects were similar across skill levels but concentrated more in manufacturing.
Key Points
- Method: SDA in a GMRIO setup isolates seventeen underlying drivers and groups them into production technology, trade structure and consumption.
- Timeframe & data: analysis covers 1995–2008 using WIOD (primary) and robustness checks with OECD ICIO and other releases.
- Aggregate result: production-technology changes (−34.8m) contributed a much larger employment decrease than trade-structure changes (−5.5m); consumption growth offset both (+58.1m).
- Manufacturing was hit harder: trade structure and technology effects shrink manufacturing employment far more (trade −17.2%, tech −43.0%) than non-manufacturing (trade −2.0%, tech −23.2%).
- Skill/occupation bias: production-technology changes reduced low-skilled jobs by ~39.8%, medium-skilled by ~31.3%, and high-skilled by ~6.3%; routine occupations fell more than non-routine.
- China channel: increased US import shares from China account for about a 2.0 million employment decline, but imports from other countries together were larger.
- Robustness: findings persist across alternative data releases and when separating trade-structure versus trade-volume effects.
Why should I read this?
Because this paper actually numbers the debate everyone argues about — is it trade or tech that’s wrecking jobs? Short answer: tech changes win — by a lot. If you care about who loses jobs, why manufacturing suffered, or whether the China story is the whole story, this gives you the hard split, with clear industry and skill detail. It’s the sort of read that saves you digging through a dozen papers.
Author style
Punchy: the authors tackle a polarising policy question with a different toolbox (SDA + GMRIO) and deliver blunt, data-backed results. If you’re into policy or research on labour-market impacts, the detail matters — this paper strengthens the case that technological change, not just trade, has been the dominant driver of employment shifts.
Context and relevance
Where many studies use regressions or regional exposure strategies, this work leverages full global input–output linkages to capture indirect effects across value chains (e.g. upstream suppliers). It is highly relevant to policymakers, labour economists and industry analysts because it disentangles direct and indirect channels (trade versus embedded production changes) and shows how those channels distribute across industries, skills and occupations. The period analysed (1995–2008) encompasses rapid global fragmentation and the strong China import shock, so if trade mattered most anywhere, it would be here — yet embodied technological change dominated.
Source
Source: https://onlinelibrary.wiley.com/doi/10.1111/roie.12805?af=R