Routine-biased technological change reshuffled Indonesian jobs in waves from 2001–2019 and hit women differently: women saw higher displacement but often upgraded into interpersonal, non-routine roles, narrowing the gender pay gap temporarily—only for weakening wage returns to female-dominated jobs to undo much of that progress by 2015–2019.

Main Finding

Routine-biased technological change in Indonesia (2001–2019) unfolded episodically and unevenly across task types, producing gender-asymmetric labor market adjustments. Women faced greater exposure to routine-job displacement but shifted substantially into non-routine interpersonal occupations, which temporarily narrowed the gender wage gap (notably 2005–2010). Over time, however, declining market valuation of female-concentrated occupations weakened this equalizing channel and contributed to a renewed widening of the gap in 2015–2019. While within-occupation (residual) dynamics explain most of the gap’s level, task-based employment and wage channels crucially determine the timing and direction of gender inequality in the formal sector.

Key Points

• Routine displacement was episodic, not simultaneous:
  • 2001–2005: Relative contraction concentrated in routine cognitive jobs.
  • 2005–2010: Relative contraction shifted toward routine manual jobs.
  • 2015–2019: Renewed pressures on routine cognitive tasks.
  • These phases reflect interactions between technological change and macro- / institutional factors.
• Gender asymmetry in adjustment:
  • Women experienced higher exposure to routine-job displacement.
  • Women reallocated strongly into non-routine interpersonal occupations (occupational upgrading).
  • This reallocation aligns with task-demand shifts from technology and with entry of younger, more educated female cohorts.
• Effects on the gender wage gap:
  • Employment reallocation narrowed the gender wage gap especially in 2005–2010.
  • Over time, declining wage returns to female-concentrated occupations (unfavorable wage exposure) reduced the equalizing effect and helped widen the gap in 2015–2019.
• Quantitative balance:
  • Residual within-task-group wage and employment differences remain the dominant contributor to the gap’s magnitude.
  • Nevertheless, task-based employment shares and task-specific wage valuation are key drivers of temporal changes in gender inequality.

Data & Methods

• Data: Panel of formal wage workers in Indonesia covering 2001–2019 (labor-force / household survey sources of formal sector employment; task classification by occupation).
• Identification strategy:
  • Stacked first-difference estimations to trace within-cohort/time changes and reduce confounding from level differences across periods.
  • Dynamic shift-share decomposition to separate contributions of (a) employment reallocation across task groups, (b) changes in task-specific wages (market valuation), and (c) within-task (residual) effects to changes in the gender wage gap over time.
• Task taxonomy: Routine vs non-routine tasks, split further into cognitive, manual, and interpersonal categories to capture heterogeneous technological exposure.
• Robustness: Analysis leverages temporal variation and cohort/education composition to link task demand shifts to both technological change and labor supply dynamics (younger, more educated female entrants).

Implications for AI Economics

• Temporal dynamics matter: Automation/RBTC effects are episodic and task-specific. Cross-sectional or static analyses may miss sequenced displacement across task types and the timing of gender impacts.
• Task composition and occupational sorting are central to gendered outcomes:
  • Women’s tendency (or ability) to reallocate into non-routine interpersonal roles can temporarily mitigate wage gaps; however, long-run equality depends on sustained wage valuation of those roles.
• Market valuation (wage returns) can undo reallocation gains:
  • Technology-driven demand shifts combined with changing wage premia for female-concentrated tasks can reverse earlier equalizing trends. AI adoption that lowers returns to female-dominated tasks can exacerbate gender wage inequality even if employment shares shift favorably.
• Policy levers:
  • Invest in targeted training/reskilling toward non-routine and high-return tasks, with attention to occupations where women are concentrated.
  • Monitor and counteract declining wage valuation of female-concentrated roles (e.g., through pay transparency, collective bargaining support, minimum-wage and sectoral wage policies).
  • Design active labor-market policies timed to episodic displacement waves (early detection and rapid response).
• Measurement and research priorities for AI economics:
  • Use task-level, high-frequency data and dynamic decomposition methods to capture phased displacement and valuation changes.
  • Study firm-level AI adoption, occupational task redefinition, and demand-side wage-setting to identify causal channels.
  • Extend analysis to informal sectors and regional/industry heterogeneity in developing economies where formal-sector patterns may differ.
• Equity-minded deployment of AI:
  • Policymakers and firms should consider not only job counts but task composition and wage structures when assessing automation’s social impacts. Interventions that preserve or raise wages in occupations attracting displaced female workers will be critical to sustaining equalizing gains.