← Papers

Industry robot penetration in China correlates with lower self-reported income gains, but workers who feel irreplaceable—especially in low-codifiability jobs—report substantially better perceived income outcomes, producing a polarized pattern across routine and high-skill occupations.

Artificial Intelligence Exposure, Perceived Job Replaceability, and Perceived Income Change: The Moderating Role of Task Codifiability—Evidence from the China General Social Survey

Rong Nie, Xiaomei Bai, Jiangmin Ding · August 03, 2026 · Mathematics

openalex correlational medium evidence 7/10 relevance Summary only summary available; pdf_status=paywall DOI Source PDF

Structured author observations

Linked only from stored provider relations; the raw author line above is never matched by name.

OpenAlex

Latest observation: August 25, 2026

Rong Nie provider ID
Xiaomei Bai provider ID
Jiangmin Ding provider ID

Semantic Scholar

Latest observation: August 25, 2026

Rong Nie provider ID
Xiaomei Bai provider ID
Jiangmin Ding provider ID

Higher industry robot density is associated with a lower probability that Chinese workers report higher household income year-over-year, but this association is strongly moderated by workers' perceived job replaceability and occupation task codifiability, producing polarization across skill/task groups.

Citation observations

Cumulative provider counts captured on specific dates; providers are never combined.

0 cumulative citations

OpenAlex · Observed August 25, 2026

View corpus context

0 cumulative citations

Semantic Scholar · Observed August 25, 2026

View corpus context

This study examines how artificial intelligence (AI) exposure is associated with Chinese residents’ perceived income change and how this association varies with Perceived Job Replaceability. Using a nationally representative sample of 6247 employed workers from the 2021 China General Social Survey matched with industry-level robot penetration data, we estimate ordered probit models and two-stage residual-inclusion control-function checks within an ordered-response framework. Three findings emerge. First, objective AI exposure is associated with a 4.2-percentage-point lower probability of reporting higher household income than last year for a one-standard-deviation increase in robot density, based on predicted-probability contrasts rather than raw ordered-probit coefficients. Second, workers reporting low Perceived Job Replaceability are 15.7 percentage points more likely to report higher household income than last year on the same probability scale. Third, task codifiability moderates this relationship: marginal-effect contrasts show that low Perceived Job Replaceability is associated with a 22.9 percentage-point increase in the probability of reporting higher household income than last year in low-codifiability occupations, compared with only 4.7 percentage points in high-codifiability occupations, indicating a polarization pattern in perceived income change outcomes. Heterogeneity analyses further show that medium-skill routine workers face the strongest negative exposure associations, whereas high-skill workers in low-codifiability occupations show the strongest positive low-replaceability contrast. Overall, the findings clarify how AI exposure, worker perceptions, and task structure are jointly associated with perceived income change in China and provide evidence relevant to more inclusive technological adjustment policies.

Summary

Main Finding

Objective AI exposure (industry robot density) is associated with lower odds of Chinese workers reporting higher household income compared with last year, but this association is strongly moderated by workers’ perceived job replaceability and task codifiability. Workers who feel their jobs are hard to replace report substantially better perceived income outcomes, especially in low-codifiability occupations, producing a polarization in perceived income-change outcomes across task types and skill groups.

Key Points

Data: nationally representative sample of 6,247 employed Chinese workers from the 2021 China General Social Survey, merged with industry-level robot penetration.
Primary association:
- A one-standard-deviation increase in robot density is associated with a 4.2 percentage-point lower probability of reporting higher household income than last year (reported as predicted-probability contrasts).
Perceived Job Replaceability:
- Workers reporting low Perceived Job Replaceability are 15.7 percentage points more likely to report higher household income than last year (same probability scale).
Interaction with task codifiability:
- Low Perceived Job Replaceability → 22.9 percentage-point increase in probability of reporting higher income in low-codifiability occupations.
- The same contrast is only 4.7 percentage points in high-codifiability occupations.
- This pattern indicates polarization: exposure-related downside concentrated among routine/medium-skill workers, upside concentrated among high-skill, low-codifiability workers who feel non-replaceable.
Heterogeneity:
- Medium-skill routine workers show the strongest negative association with AI exposure.
- High-skill workers in low-codifiability occupations show the largest positive effect of low perceived replaceability.

Data & Methods

Data sources:
- 2021 China General Social Survey (representative, n = 6,247 employed respondents).
- Industry-level robot penetration (robot density) matched to respondents’ industries.
Dependent variable:
- Self-reported perceived household income change relative to last year (ordered response).
Key independent variables:
- Objective AI exposure: industry robot density (standardized).
- Perceived Job Replaceability: respondent-level perception.
- Task codifiability: occupation-level measure.
Estimation approach:
- Ordered probit models to respect the ordinal nature of perceived income change.
- Two-stage residual-inclusion (control-function) checks implemented within an ordered-response framework to address endogeneity concerns.
- Reported effects primarily via predicted-probability contrasts (marginal-effect contrasts) rather than raw ordered-probit coefficients to improve interpretability.
Robustness / heterogeneity:
- Interaction analyses by task codifiability and skill/routine classification.
- Heterogeneity analyses across skill groups and occupation types.

Implications for AI Economics

Perceptions matter: Worker beliefs about replaceability strongly condition how AI exposure relates to perceived well-being. Policies addressing perceptions (transparent communication, career guidance) can alter subjective adjustment outcomes.
Task structure is key: Codifiability of tasks moderates exposure effects—routine, codifiable tasks concentrate downside risk, while low-codifiability tasks amplify the benefit of perceived non-replaceability. Sector- and task-targeted retraining and upskilling are therefore essential.
Distributional and polarization concerns: AI exposure may produce polarized perceived income-change outcomes across skill and task groups. Redistribution, targeted support for medium-skill routine workers, and inclusive adjustment policies are warranted.
Measurement and inference:
- Using predicted-probability contrasts from ordered-response models yields policy-relevant magnitudes for subjective outcomes.
- The study addresses endogeneity with control-function checks, but results are based on cross-sectional, self-reported perceived income change in China (2021). Caution is needed in interpreting causal direction and in generalizing to other countries or objective income measures.
Directions for policy and research:
- Design active labor-market policies focused on routine-task workers and strengthen pathways into low-codifiability, higher-skill roles.
- Evaluate interventions that alter perceived replaceability (e.g., certification, task redesign) and measure downstream effects on objective earnings and labor-market outcomes.
- Future research should use longitudinal data, stronger causal instruments, and firm-level measures to link perceived and realized income changes under AI adoption.

Assessment

Paper Typecorrelational Evidence Strengthmedium — Uses a nationally representative sample and an objective industry-level exposure measure (robot density), and shows clear, interpretable marginal-effect contrasts and rich interactions; however the design is cross-sectional with a subjective outcome, potential unobserved confounding, and no clearly exogenous source of variation in robot density, so causal claims remain tentative. Methods Rigormedium — Appropriate modeling choices for an ordinal dependent variable (ordered probit) and reporting of predicted-probability contrasts improve interpretability; the addition of two-stage residual-inclusion checks shows attention to endogeneity, and interaction/homogeneity analyses are comprehensive. Weaknesses include reliance on industry-level exposure (coarse measurement), self-reported perceived income change, cross-sectional data that limits causal inference, and no strong external instrument or longitudinal identification strategy. SampleNationally representative sample of 6,247 employed respondents from the 2021 China General Social Survey, merged with industry-level robot penetration (robot density) and occupation-level task codifiability measures; dependent variable is self-reported household income change relative to last year (ordinal). Themeslabor_markets human_ai_collab skills_training inequality adoption IdentificationCross-sectional association between respondent-level reported perceived household income change and industry-level robot density (standardized) merged into the 2021 China General Social Survey; estimated with ordered probit models and reported as predicted-probability contrasts; endogeneity concerns are addressed with two-stage residual-inclusion (control-function) checks within the ordered-response framework and extensive interaction/heterogeneity analyses (perceived job replaceability, task codifiability, skill/routine groups). No strong external instrument or natural experiment is reported. GeneralizabilityChina-only (2021) — institutional, labor-market, and pandemic-period specifics may limit external generalizability, Outcome is self-reported perceived household income change (subjective), not objective earnings or long-run outcomes, Exposure is industry-level robot density (coarse proxy) rather than firm- or worker-level AI adoption, Robot density captures automation broadly and may not reflect modern generative-AI exposures specifically, Cross-sectional design limits inference about dynamic or causal effects over time

Claims (7)

Claim	Direction	Outcome	Confidence & Evidence	Details
A one-standard-deviation increase in industry robot density is associated with a 4.2 percentage-point lower probability that Chinese workers report higher household income than the previous year. Wages	negative	Self-reported change in household income relative to the previous year	Reading fidelity high Study strength medium	n=6247 4.2 percentage-point lower probability 0.3
Workers reporting low perceived job replaceability are 15.7 percentage points more likely to report higher household income than the previous year. Wages	positive	Self-reported probability of higher household income relative to the previous year	Reading fidelity high Study strength medium	n=6247 15.7 percentage points more likely 0.3
The association between low perceived job replaceability and reporting higher income is substantially larger in low-codifiability occupations than in high-codifiability occupations: 22.9 percentage points versus 4.7 percentage points. Wages	positive	Probability of reporting higher household income than the previous year	Reading fidelity high Study strength medium	n=6247 22.9 percentage-point increase in low-codifiability occupations; 4.7 percentage-point increase in high-codifiability occupations 0.3
The interaction between perceived job replaceability and task codifiability produces a polarization in perceived income-change outcomes across task types and skill groups. Inequality	mixed	Self-reported change in household income relative to the previous year across occupational task and skill groups	Reading fidelity high Study strength medium	n=6247 0.3
Medium-skill routine workers show the strongest negative association between AI exposure and reporting higher household income. Wages	negative	Self-reported probability of higher household income relative to the previous year	Reading fidelity high Study strength medium	n=6247 0.3
High-skill workers in low-codifiability occupations show the largest positive association between low perceived job replaceability and reporting higher income. Wages	positive	Self-reported probability of higher household income relative to the previous year	Reading fidelity high Study strength medium	n=6247 0.3
The study finds an association rather than establishing a causal effect because it uses cross-sectional data and self-reported perceived income change, although two-stage residual-inclusion control-function checks were conducted to address endogeneity concerns. Other	mixed	Relationship between objective AI exposure, perceived job replaceability, and self-reported household income change	Reading fidelity high Study strength low	n=6247 0.15