Four-country comparative analysis of occupational employment change in Britain, Germany, Spain, and Switzerland from 1990 to 2008 using individual-level labor force surveys. Applies the Wright-Dwyer (2003) quintile methodology — rank occupations by median earnings, track quintile shares over time — to test four competing hypotheses: skill-biased technical change (SBTC), routinization, skill supply evolution, and wage-setting institutions. The central result is that massive occupational upgrading occurred everywhere (consistent with SBTC and educational expansion), the middle of the occupational distribution hollowed out everywhere (consistent with routinization), but the bottom quintile expanded only in Britain and Spain — not in Germany or Switzerland. This cross-national divergence is explained by wage-setting institutions: compressed wage floors in Germany and Switzerland priced low-paid interpersonal service jobs out of the market, while Britain's and Spain's more flexible wage structures allowed such jobs to be created in large numbers.
Universal top-end upgrading. In all four countries, >75% of net employment growth concentrated in the top quintile (managers, professionals, computer specialists, teachers). Germany: all net job growth in Q5 only. Absolute numbers: ~2.5M top-quintile jobs created in Germany, 2.8M in Britain, 3.1M in Spain over two decades.
Universal middle collapse. Routine cognitive and manual occupations (clerks, secretaries, production workers, assemblers) declined -29% to -41% across all four countries. This is the strongest cross-national regularity and the key evidence for the routinization hypothesis. Top occupations declining in Britain/Switzerland: other secretarial personnel (−1.67 percentage points [pp]), metal-work fitters (−1.43 pp), office clerks in private services (−2.19 pp).
Bottom-quintile divergence — the main finding. Bottom quintile expanded in Britain (+3.1%) and Spain (+3.8%), contracted in Germany (−2.5%) and Switzerland (−1.8%). This contradicts both SBTC (which predicts bottom decline everywhere) and unmodified routinization (which predicts bottom expansion everywhere). The entire difference is explained by growth or stagnation of interpersonal service jobs — waiters, cleaners, care workers, hairdressers — which expanded strongly in Britain (exclusively Q1) and Spain, but were flat in Germany and Switzerland.
Institutional mediation: relative wages in Q1 predict Q1 job creation. Q1 median wage as % of overall workforce median: Britain 65% → Q1 +3.1%; Germany 78% → Q1 −2.5%; Spain 68% → Q1 +3.8%; Switzerland 73% → Q1 −1.8%. Where compressed wage-setting institutions raise the relative price of low-paid service labor, such jobs are not created. Technology generates the same shift in labor demand everywhere; institutions channel it into more or less polarized outcomes.
Immigration accounts for all of Britain's and Spain's Q1 expansion. Decomposition by nationality-gender group: native men and women in Britain lost employment in Q1 (combined −1.1 pp); immigrant men and women gained (+1.8 pp). In Spain, immigrants added 4.1 pp to Q1 and 3.9 pp to Q2; Spanish men lost 12.6 pp in Q1+Q2 combined. Mechanism: immigrants downgrade upon arrival due to language barriers, credential non-recognition, and discrimination (Dustmann et al. 2008), supplying the low-paid service jobs that flexible wages make profitable. Without immigration, Britain's Q1 would have declined by ~3 pp — similar to Germany and Switzerland.
Upgrading driven by national women; middle decline driven by national men. National women's Q1 share fell sharply while Q5 share rose massively in all four countries. National men's employment share in Q2–Q4 shrank dramatically (Spain: −16 pp; Britain: −9.3 pp; Switzerland: −7.2 pp; Germany: −5.1 pp). National men's only Q5 gains. In Switzerland, the polarization pattern in Q1–Q3 was driven exclusively by national men.
Temporal shift: upgrading → polarized upgrading. 1990–96: clean upgrading in all four countries (recession burdens fall disproportionately on Q1 workers). Post-1996: Q1 begins growing in all four countries alongside continued middle decline. Consistent with Autor et al. (2008): SBTC dominated the 1980s, routinization became more visible in the 1990s.
Skill supply decomposition. Decomposition analysis (24 education-age groups) shows educational expansion closely predicts the observed upgrading. Two residuals: (a) actual Q3 decline exceeds prediction by 1.7–3.2 pp (Britain, Germany, Switzerland) — demand-side routinization at work beyond skill supply; (b) actual Q1 decline is much smaller than predicted (especially Britain: predicted −5.9 pp, observed +0.6 pp) — immigration-driven supply of low-paid labor plugs the gap.
Earnings-employment covariance confirms demand story. Real changes in quintile median wages 1990–2008 co-vary positively with employment changes in Britain (r=0.85), Switzerland (r=0.73), Germany (r=0.57) — consistent with demand-induced shifts. Spain anomalous (r=−0.23): Q1 and Q2 wages fell despite job creation there, possibly reflecting immigration-driven supply shock rather than demand pull.
"Different country institutions may thus channel technological change into a more or less polarized pattern of occupational upgrading."
"While we observe everywhere a collapse of middling jobs in quintiles 2 and 3, growth in quintile 1 only took place in countries where the low-wage sector is relatively unsheltered from market pressure."
This is the most rigorous comparative test of the routinization hypothesis in Europe, and its central finding — that routinization correctly predicts the middle collapse but incorrectly predicts universal bottom expansion — is under-appreciated relative to the impact of Goos and Manning (2007) and Autor et al. (2008). The institutional-mediation argument is compelling but under-identified: four country cases cannot distinguish between wage-setting institutions and other confounders (e.g., public sector size, hours regulation, childcare availability). The immigration finding is mechanically important but raises a conceptual question: if Q1 expansion in Britain is entirely driven by immigrants downgrading, is Britain experiencing "polarization" in a meaningful sense for native workers? Native Britons in Q1 actually declined. The paper treats all workers symmetrically, which is empirically clean but blurs the distinction between polarization-as-demand-shock and polarization-as-immigration-supply-shock.