Routine-Biased Technological Change (RBTC) is the hypothesis that computerization and automation disproportionately substitute for "routine" tasks — those that can be specified as a sequence of codifiable rules — while complementing "non-routine abstract" tasks and failing to replace "non-routine manual" tasks. The framework was formalized by Autor, Levy, and Murnane (2003) and is the leading theoretical account of occupational polarization in industrialized economies.
Autor, Levy, and Murnane (2003) classify tasks along two dimensions — routine vs. non-routine, and cognitive vs. manual — yielding four quadrants:
| Task type | Computerizable? | Direction |
|---|---|---|
| Routine cognitive (clerical, data entry, bookkeeping) | Yes — explicit rule-based | Declining |
| Routine manual (assembly, production, operative) | Largely yes — precise repetitive motion | Declining |
| Non-routine cognitive / abstract (analysis, management, design) | No — complement to capital | Rising |
| Non-routine manual (service, personal care, janitorial) | No — context-dependent dexterity and interpersonal | Stable/rising |
The key insight is that "routine" does not mean simple or unskilled. Production work and back-office clerical work are highly skilled but rule-bound; software can substitute for them. Conversely, cleaning hotel rooms or serving food involves complex physical navigation and social adaptation that resists automation even at low skill levels.
Middle-skill occupations — clerical, administrative support, sales, production, and operative jobs — are intensive in routine tasks and therefore most exposed to computerization. High-skill professional/technical/managerial work is intensive in non-routine abstract tasks and is complemented by technology. Low-skill service work is intensive in non-routine manual tasks and is neither substituted nor complemented. The result is a barbell: employment gains at top and bottom, losses in the middle. See Occupational Polarization.
All 16 European Union (EU) countries in the EU-15 + Norway sample show the same employment polarization pattern – (Autor 2010; Goos, Manning, and Salomons 2014). Germany's task-content shifts mirror the U.S. with a short lag (Spitz-Oener 2006). The international scope is strong evidence that the mechanism is technological (a global input-price shock) rather than a U.S.-specific institutional or policy story.
The earlier "skill-biased technological change" hypothesis (Katz and Murphy 1992; Juhn, Murphy, and Pierce 1993) posited a monotonic shift in demand from low-skill to high-skill workers, driven by capital-skill complementarity. SBTC predicts monotonically rising wage inequality across the full distribution. RBTC predicts polarization — wages grow at both tails, fall in the middle — which matches the data better for post-1980 trends. Card and DiNardo (2002) showed that SBTC cannot explain the widening at the bottom of the wage distribution; RBTC via Autor, Katz, and Kearney (2006/2008) resolved this by separating the upper and lower tails.
RBTC is a technological demand shift story. It explains why middle-skill employment fell, but the equilibrium wage outcome depends on labor supply responses. Crucially, RBTC overpredicts bottom-end employment expansion in high-wage economies: Oesch and Rodriguez Menes (2010), comparing Britain, Germany, Spain, and Switzerland 1990–2008, find that the middle collapse is universal (consistent with RBTC) but bottom-quintile expansion occurred only in Britain and Spain — not Germany and Switzerland. The difference is explained by wage-setting institutions: compressed wage floors price low-paid interpersonal service jobs out of the market. A single technological shock thus produces polarization in flexible-wage economies (Britain) and closer to pure upgrading in compressed-wage economies (Germany). This institutional filter is not part of the standard ALM framework and constitutes the sharpest falsification of the routinization hypothesis in a comparative context. See Occupational Polarization. The deceleration in college attainment among men post-Vietnam (Autor 2010) is a supply-side amplifier: even as demand for skilled workers rose, male college supply stagnated ( attainment –), pushing up the college wage premium to (2009) from (1963). Rising automation demand stagnant supply large premium. RBTC is also incomplete as a causal account: the correlation between task content and employment growth is consistent with but does not prove that technology is the mechanism. Trade (especially the China shock post-2000) drove production job losses through a distinct channel. See China Trade Shock.
Mechanically, RBTC shifts the relative demand for labor at the task level. A firm that previously employed 10 clerks and 1 manager can shift to 2 clerks and 1 manager after introducing accounting software — the same manager is now more productive (complementarity), the 8 displaced clerks must find other work. Because routine tasks are concentrated at middle-skill-level occupations, the displaced workers are neither at the top (where demand is rising) nor the bottom (where supply expansion can absorb them) but in the middle — creating polarization in occupational shares. The equilibrium implication is that middle-skill wages also fall relative to both tails, unless the displaced workers reskill to abstract tasks (slow) or shift into service jobs (depressing low-skill wages via supply crowding).