## wp18135 - Section 3 discusses the drivers and mechanisms which lie behind polarization in developed

## Source details

**Canonical URL:** [wp18135 - Section 3 discusses the drivers and mechanisms which lie behind polarization in developed](https://www.imf.org/-/media/files/publications/wp/2018/wp18135.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp18135.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp18135.pdf.json)

---

### Introduction and scope
- Focus: quantify "exposure to routinization" defined as the extent to which jobs are at risk of being automated by ICT capital or the propensity to adopt labor-displacing information technology.
- New, consistent and comparable measure of exposure to routinization across countries, industries, and over time.
- Dataset coverage: 160 countries at annual, biennial, quinquennial or decennial frequency between 1960 and 2015.
- Sample composition: 116 developing economies and 38 developed economies.
- Measurement approach:
  - Ordinal routine-task intensity (RTI) scores at 3-digit occupation level from Autor and Dorn (2013).
  - Scores weighted by employment shares to create time-varying country- and industry-level measures.
  - Assumption: intrinsic routinizability (RTI) is fixed across industries, countries and over time (noting limitations for very long horizons).

### Key empirical findings and stylized facts
- Developing economies are significantly less exposed to automation than developed economies and remain less exposed for the quarter century between 1990 and 2015.
- Time periods used: "initial" exposure 1990-95 and "subsequent" exposure 2010-15.
- Industry exposure patterns:
  - Industries with lowest exposure: Agriculture (lowest), Accommodation, Health services, Retail.
  - Industries with highest exposure: Manufacturing, Financial services, Transportation and storage.
- Convergence patterns:
  - Average exposures are lower in developing than in developed economies in all time periods shown (1990-95, 2000-05, 2010-15).
  - Developing economies: exposures have risen over time.
  - Developed economies: exposures have declined steadily over time.
  - In about half of industries over the next two decades after the early 1990s, average exposures in developing economies overtook developed economies (notably manufacturing, mining, transportation and storage).
- Predictive/asymmetric dynamics of initial exposure:
  - "The initial exposure to routinization is a powerful predictor of the subsequent exposure to routinization."
  - Developed economies: higher initial exposure → lower subsequent exposure (consistent with intensive displacement of routine labor and polarization).
  - Developing economies: higher initial exposure → less subsequent rise in exposure (structural transformation and other forces raised exposure despite possible mitigating forces).
- Occupational employment share impacts (1990 → 2015):
  - Developed economies:
    - An increase in initial exposure by 0.14 (one standard deviation) associated with declines of 2.7, 1 and 0.8 percentage points in employment shares of service workers, craft workers and machine operators respectively.
    - Same shock associated with increases of 3.5 and 4 percentage points in associate professionals and clerical workers respectively.
  - Developing economies:
    - Initial exposure significantly negatively associated with future employment-share growth for clerical workers, service workers and machine operators.
    - For agriculture, higher initial exposure associated with smaller declines in agricultural employment shares.
- Wages (manufacturing sector; illustrative sample of 48 countries):
  - "A unit increase in the initial exposure to routinization is associated with a decline of 11 percent in real manufacturing wages."

### Drivers and mechanisms — how they differ by development status
- Computing cost trend:
  - "The real cost of computing power is estimated to have fallen at a staggering rate of more than 50 percent annually between 1969 and 2005 (Nordhaus 2007)."
- Four inter-related drivers discussed:
  1. Technological advancement and falling cost of automating routine tasks:
     - Hypothesis: greater decline in relative price of investment goods → more polarization (Autor, Levy and Murnane 2003).
  2. Size of routine employment share and imperfect substitutability:
     - Need a significant share of economy engaged in routine occupations and imperfect substitutability between routine and nonroutine tasks for measurable polarization.
  3. Skill bias of ICT:
     - ICT complementary with skilled labor → rising demand for skilled labor and displacement of middle-skilled routine workers into lower-paying service jobs, contributing to polarization.
  4. Trade–technology interaction (offshoring and GVCs):
     - Offshoring of routine-intensive tasks mutually reinforces automation effects; lowers relative demand for routine labor and contributes to polarization.
- How these operate differently in developing economies:
  - Relative price of investment goods:
    - Empirical statement: "whereas the relative price of investment declined by about 15 percent in developed economies since 1990, in developing countries they have risen 13 percent, albeit with significant cross-country heterogeneity."
    - Implication: displacement via falling capital goods prices likely weaker or absent in developing economies.
  - Occupational distribution:
    - ILO estimates about 40 percent employed in the primary sector in developing economies.
    - Developing economies concentrated in low-susceptibility jobs; high share in agricultural occupations and low shares in clerical and other high routine-intensive occupations.
  - Globalization and offshoring:
    - Offshoring of routine-intensive jobs to developing economies could raise demand for routine labor there (offsetting polarization forces); offshoring from developed countries lowers their routine demand.
    - Rising participation in global value chains could have helped raise routine exposures in developing economies and lower them in developed countries.
  - Adoption constraints in developing economies:
    - Adoption lags, insufficient information about new technologies, institutional barriers (property rights, misappropriation), limited supplier competition → higher price dispersion and lower marginal adoption.

### Empirical evidence: determinants and magnitudes (long-change regression framework 1990–2015)
- Regression setup:
  - Dependent variable: change in routine exposure.
  - Covariates: change in relative price of investment goods (PI), structural transformation (ST), participation in global value chains (Trade), initial middle-skill employment share (Skill).
  - Interaction with developing-economy dummy (DE) to allow differential impacts.
- Summary statistics (Table 1 excerpts)
  - Developed Economies:
    - Change in routine exposure: Min. -0.39; Max. 0.14; Mean -0.11; Median -0.11; St. Dev. 0.13.
    - Change in relative price of investment: Min. -0.74; Max. 0.12; Mean -0.17; Median -0.17; St. Dev. 0.17.
    - Initial middle-skill employment share: Min. 0.22; Max. 0.50; Mean 0.39; Median 0.41; St. Dev. 0.06.
    - Structural transformation: Min. -1.95; Max. 0.20; Mean -0.77; Median -0.70; St. Dev. 0.41.
    - Change in global value chain participation: Min. -0.09; Max. 0.42; Mean 0.13; Median 0.14; St. Dev. 0.08.
  - Developing Economies:
    - Change in routine exposure: Min. -0.28; Max. 1.30; Mean 0.20; Median 0.13; St. Dev. 0.31.
    - Change in relative price of investment: Min. -1.31; Max. 11.03; Mean 0.09; Median -0.05; St. Dev. 1.14.
    - Initial middle-skill employment share: Min. 0.03; Max. 0.76; Mean 0.25; Median 0.28; St. Dev. 0.13.
    - Structural transformation: Min. -2.42; Max. 1.20; Mean -0.57; Median -0.56; St. Dev. 0.58.
    - Change in global value chain participation: Min. -0.34; Max. 7.73; Mean 0.17; Median 0.11; St. Dev. 0.67.
- Main empirical findings (summary of Table 2 results and text):
  - Structural transformation (ST):
    - In developing economies, structural transformation robustly associated with increases in routine exposure.
    - For the median developing economy, a one standard deviation decrease of ST is associated with a 0.4 increase in routine exposure.
    - For the median developed economy, one standard deviation decline in ST lowers routine exposure by .005 (about 5 percent of the standard deviation of change in exposures).
  - Trade / Global Value Chain participation:
    - Trade strongly affected exposures in opposite directions:
      - In the median developed economy, an increase in Trade associated with a decline in exposure to routinization by about 0.13.
      - In the median developing economy, where Trade rose by 0.10, exposure to routinization increased by about 0.11.
  - Relative price of investment goods:
    - Declining relative price of investment a key contributor to falling routine exposures in developed economies; effect weaker or opposite in developing economies in the sample period.

### Conclusions and policy implications
- Empirical synthesis:
  - Developing economies are significantly less exposed to routinization than developed counterparts, but exposures have converged over the last quarter century.
  - In economies with initially high exposure, strong polarization dynamics have led to lower subsequent exposures; in economies with low initial exposure, structural transformation and on-shoring of routine tasks raised subsequent exposure.
  - For developed economies: technological advancement and increased offshoring are proximate causes of declining routine exposures (polarization).
  - For developing economies: structural transformation and expanded vertical trade are main drivers raising exposures (anti-polarization forces during the sample period), but rising exposures imply growing future vulnerability.
- Risks and future outlook for developing economies:
  - Rapid technology adoption in some developing economies (e.g., rising industrial robotization in China) may increase automation impacts.
  - Potential for premature deindustrialization, erosion of middle-skilled employment early in convergence, adverse productivity and growth effects, and derailing income convergence (Rodrik 2016).
  - Preliminary modeling (Das and Hilgenstock 2018): polarization not imminent under some assumptions, but factors gathering force over ten to twenty years.
- Policy challenges and recommendations:
  - Increase supply of high-skilled labor.
  - Raise investment in technological proficiency.
  - Build safety buffers (Dao et al. 2017).
  - Address distributional consequences: technological change could shift income toward high-wage individuals, exacerbating inequality and raising policy trade-offs in developing economies.

### Appendix and data
- Exposures to offshorability constructed analogously to routinization exposures; online appendix referenced in source text for details.
- Database coverage: exposures constructed for 160 countries at aggregate level and 126 countries at industry level between 1960 and 2015 (various frequencies). Table A1 lists country coverage and years (as of 2018).

*Source: wp18135 - Section 3 discusses the drivers and mechanisms which lie behind polarization in developed economies (IMF working paper).*

### References .............................................................................................................

### wp18135 - References

### Introduction and scope
- Focus: quantify "exposure to routinization" defined as the extent to which jobs are at risk of being automated by ICT capital or the propensity to adopt labor-displacing information technology.
- New, consistent and comparable measure of exposure to routinization across countries, industries, and over time.
- Dataset coverage: 160 countries at annual, biennial, quinquennial or decennial frequency between 1960 and 2015.
- Sample composition: 116 developing economies and 38 developed economies.

### Contributions of the paper
- Constructed cross-country, cross-industry, and over-time measures of exposure to routinization by:
  - Using ordinal routine-task intensity scores at 3-digit occupation level (from Autor and Dorn (2013)).
  - Weighting scores by employment shares to create country- and industry-level measures that are time-varying.
- Documented stylized facts and cross-country heterogeneity in exposures, including differences between developing and developed economies.
- Empirically disentangled drivers of changes in exposures over the last quarter century, including:
  - Shifts in skill supply.
  - Decline in the price of capital goods.
  - Structural transformation.
  - Globalization of trade.
- Linked initial exposures to long-run changes in sectoral employment and wages and analyzed implications for labor markets in developing economies.

### Key empirical findings and stylized facts
- Developing economies are significantly less exposed to automation than developed economies and remain less exposed for the quarter century between 1990 and 2015.
- Dataset construction details:
  - Occupation-level routine scores (ordinal) assigned to all 3-digit census occupations.
  - Exposures aggregated by weighting routine scores with employment shares; measures are time-varying as employment shares change.
- Asymmetry in predictive power of initial exposure:
  - Among countries heavily exposed initially: higher initial exposure → lower subsequent exposure.
  - Among countries with initially low exposure: higher initial exposure → lower subsequent rise in exposure.
  - Asymmetry reflects polarization forces in high-exposure countries and structural transformation and globalization in low-exposure countries.
- Convergence of routine exposures across country groups:
  - Falling exposures in developed economies.
  - Rising exposures in developing economies.
- Interpretation of falling exposures in developed economies:
  - Consistent with polarization that reduces the share of middle-skilled workers (most exposed to ICT), lowering subsequent exposure for marginal workers.
- Interpretation of rising exposures in developing economies:
  - Evidence that automation has not displaced labor on a macro-significant scale; no broad evidence of polarization.
  - Rising exposures are consistent with structural transformation (expansion of manufacturing and routine tasks), supporting an alternative explanation to routinization for polarization patterns.
  - Heterogeneity: exposures have risen in most Asian economies, Developing Europe and many low-income economies but remained almost unchanged in Latin America.

### Drivers and mechanisms
- Decline in relative price of investment goods and its role:
  - The accelerated decline in the relative price of investment is a developed-economy phenomenon largely absent in developing countries (Dao et. al. 2017).
- Factor substitution and capital intensity:
  - Production in developing countries is less capital-intensive, potentially reflecting low relative price of labor to capital.
  - Aggregate elasticity of substitution between capital and labor estimated to be lower than one in developing economies (Dao et. al.).
- Role of structural transformation:
  - In developing economies, rising routine exposures reflect expansion of industrial base and concentration of routine tasks in manufacturing.
  - This pattern is evidence against premature deindustrialization in many cases, though substantial heterogeneity exists across developing economies.
- Globalization and trade:
  - Global value chain participation and trade globalization are analyzed as contributors to changes in routine exposures across countries.

### Implications for labor markets and future research directions
- Initial exposure measures offer exogenous variation useful for assessing long-run causal impacts of routinization on macroeconomic outcomes (e.g., labor share of income, sectoral employment, wages).
- Policy-relevant implications:
  - Divergent dynamics across developed and developing economies imply that policies addressing automation risks should be tailored to country-specific stages of development, factor prices, and structural transformation patterns.
- Open questions and considerations:
  - Whether developing economies will experience incipient polarization over time depends on interplay between technological diffusion, relative factor prices, elasticity of substitution, and structural transformation.
  - Heterogeneity across developing economies (e.g., Latin America vs. Asia) warrants disaggregated analysis.

*Source: wp18135 - References (IMF working paper content provided).*

### Section 3 discusses the drivers and mechanisms which lie behind polarization in developed

### wp18135 - Section 3 discusses the drivers and mechanisms which lie behind polarization in developed

### Exposure to routinization: concepts, measurement, and stylized facts
- Computing cost trend:
  - "The real cost of computing power is estimated to have fallen at a staggering rate of more than 50 percent annually between 1969 and 2005 (Nordhaus 2007)."
- Definition and role of routine tasks:
  - Routine tasks: "require methodical repetition of an unwavering procedure...exhaustively specified with programmed instructions and performed by machines" (Autor et al.).
  - Routine tasks concentrate in middle-skilled jobs (e.g., assembly line work, clerical work).
  - Non-routine low-skilled services (e.g., food services, child care) and non-routine high-skilled complex problem-solving tasks (e.g., scientists) differ in automability and complementarity with computer capital.
- Cross-country variation in polarization evidence:
  - Example: between 1993 and 2006, middle-skilled job share remained unchanged in Portugal, fell by about 5 percentage points in the Netherlands, and 15 percentage points in Austria.
  - Implication: routine exposures or automation rates vary across countries.
- Construction of aggregate routinization measures:
  - Base: Autor and Dorn (2013) RTI scores for 330 3-digit occupations.
  - Aggregation: average 3-digit scores into nine 1-digit ISCO-88 occupation categories; scores standardized to mean zero and standard deviation one.
  - Exposures computed by weighting occupation routinization scores by occupation employment shares at industry-country-time and country-time levels.
  - Important assumption: intrinsic routinizability (RTI) is fixed across industries, countries and over time (noting limitations for very long horizons).
- Stylized facts on exposures:
  - Time periods used: "initial" exposure 1990-95 and "subsequent" exposure 2010-15.
  - Industries with lowest exposure: Agriculture (lowest), Accommodation, Health services, Retail.
  - Industries with highest exposure: Manufacturing, Financial services, Transportation and storage.
  - Systematic evolution:
    - In Panel A (early 1990s), average routine exposures are higher in developed than in developing economies in every industry.
    - Over next two decades, in about half of industries average exposures in developing economies overtook developed economies (notably manufacturing, mining, transportation and storage).
  - Convergence of exposures by development stage:
    - Average exposures shown for 1990-95, 2000-05, and 2010-15:
      - "Average exposures are lower in developing than in developed economies in all time periods shown."
      - Developing economies: exposures have risen over time.
      - Developed economies: exposures have declined steadily over time.
    - Implication: convergence of world-wide routine exposures; potential rising automation risk for developing economies if trends continue.
  - Predictive power of initial exposure:
    - "The initial exposure to routinization is a powerful predictor of the subsequent exposure to routinization."
    - For developed economies: higher initial exposure → lower subsequent exposure (consistent with intensive displacement of routine labor).
    - For developing economies: higher initial exposure → less subsequent rise in exposure (structural transformation and other forces raised exposure despite possible mitigating forces).

### Drivers and mechanisms of polarization — how they differ by development status
- Four inter-related drivers discussed:
  1. Technological advancement and falling cost of automating routine tasks:
     - Hypothesis: greater decline in relative price of investment goods → more polarization (Autor, Levy and Murnane 2003).
  2. Size of routine employment share and imperfect substitutability:
     - Need a significant share of economy engaged in routine occupations and imperfect substitutability between routine and nonroutine tasks for measurable polarization.
  3. Skill bias of ICT:
     - ICT complementary with skilled labor → rising demand for skilled labor and displacement of middle-skilled routine workers into lower-paying service jobs, contributing to polarization.
  4. Trade–technology interaction (offshoring and GVCs):
     - Offshoring of routine-intensive tasks is mutually reinforcing with automation; lowers relative demand for routine labor and contributes to polarization.
- How these operate differently in developing economies:
  - Relative price of investment goods:
    - Decline in relative price of investment goods predominantly a developed economy phenomenon since 1990.
    - Empirical statement: "whereas the relative price of investment declined by about 15 percent in developed economies since 1990, in developing countries they have risen 13 percent, albeit with significant cross-country heterogeneity."
    - Implication: displacement via falling capital goods prices likely weaker or absent in developing economies.
  - Occupational distribution:
    - Developing economies concentrated in low-susceptibility jobs: ILO estimates about 40 percent employed in the primary sector.
    - Low routine exposure driven by high share in agricultural occupations and low shares in high routine-intensive occupations such as clerical workers.
    - Implication: favorable shocks to capital goods prices may not yield observable displacement in near term.
  - Globalization and offshoring:
    - Offshoring of routine-intensive jobs to developing economies could raise demand for routine labor there (offsetting polarization forces); offshoring from developed countries lowers their routine demand.
    - Evidence hint: rising participation in global value chains could have helped raise routine exposures in developing economies and lower them in developed countries.
  - Adoption constraints in developing economies:
    - Possible adoption lags, insufficient information about new technologies, institutional barriers (property rights, misappropriation), limited supplier competition → higher price dispersion and lower marginal adoption.

### Empirical evidence: determinants and magnitudes
- Long-change regression framework (1990–2015) includes:
  - Dependent variable: change in routine exposure.
  - Covariates: change in relative price of investment goods (PI), structural transformation (ST), participation in global value chains (Trade), initial middle-skill employment share (Skill).
  - Interaction with developing-economy dummy (DE) to allow differential impacts.
- Summary statistics (Table 1 excerpts):
  - Developed Economies:
    - Change in routine exposure: Min. -0.39; Max. 0.14; Mean -0.11; Median -0.11; St. Dev. 0.13.
    - Change in relative price of investment: Min. -0.74; Max. 0.12; Mean -0.17; Median -0.17; St. Dev. 0.17.
    - Initial middle-skill employment share: Min. 0.22; Max. 0.50; Mean 0.39; Median 0.41; St. Dev. 0.06.
    - Structural transformation: Min. -1.95; Max. 0.20; Mean -0.77; Median -0.70; St. Dev. 0.41.
    - Change in global value chain participation: Min. -0.09; Max. 0.42; Mean 0.13; Median 0.14; St. Dev. 0.08.
  - Developing Economies:
    - Change in routine exposure: Min. -0.28; Max. 1.30; Mean 0.20; Median 0.13; St. Dev. 0.31.
    - Change in relative price of investment: Min. -1.31; Max. 11.03; Mean 0.09; Median -0.05; St. Dev. 1.14.
    - Initial middle-skill employment share: Min. 0.03; Max. 0.76; Mean 0.25; Median 0.28; St. Dev. 0.13.
    - Structural transformation: Min. -2.42; Max. 1.20; Mean -0.57; Median -0.56; St. Dev. 0.58.
    - Change in global value chain participation: Min. -0.34; Max. 7.73; Mean 0.17; Median 0.11; St. Dev. 0.67.
- Main empirical findings (summary of Table 2 results and text):
  - Structural transformation:
    - In developing economies, structural transformation robustly associated with increases in routine exposure.
    - For the median developing economy, a one standard deviation decrease of ST is associated with a 0.4 increase in routine exposure.
    - For median developed economy, one standard deviation decline in ST lowers routine exposure by .005 (about 5 percent of the standard deviation of change in exposures).
  - Trade / Global Value Chain participation:
    - Trade strongly affected exposures in opposite directions:
      - In the median developed economy, an increase in Trade associated with a decline in exposure to routinization by about 0.13.
      - In the median developing economy, where Trade rose by 0.10, exposure to routinization increased by about 0.11.
  - Relative price of investment goods:
    - Declining relative price of investment a key contributor to falling routine exposures in developed economies; effect weaker or opposite in developing economies in the sample period.
- Occupational employment share dynamics (Tables 3 and 4; summary):
  - Initial routine exposure predicts subsequent changes in 1-digit occupational employment shares (1990 → 2015).
  - Developed economies:
    - Higher initial exposure associated with declines in employment shares of routine-intensive occupations (service workers, craft workers, machine operators) and increases for some less routine occupations (associate professionals, clerical workers).
    - Quantified: an increase in initial exposure by 0.14 (one standard deviation) associated with declines of 2.7, 1 and 0.8 percentage points in employment shares of service workers, craft workers and machine operators respectively; and increases of 3.5 and 4 percentage points in associate professionals and clerical workers respectively.
  - Developing economies:
    - Initial exposure significantly negatively associated with future employment-share growth for clerical workers, service workers and machine operators (i.e., higher initial exposure → smaller increases).
    - For agriculture, higher initial exposure associated with smaller declines in agricultural employment shares.
- Wages (manufacturing sector; illustrative sample of 48 countries):
  - "A unit increase in the initial exposure to routinization is associated with a decline of 11 percent in real manufacturing wages."

### Conclusions and policy implications
- Empirical synthesis:
  - Developing economies are significantly less exposed to routinization than developed counterparts, but exposures have converged over the last quarter century.
  - In economies with initially high exposure, strong polarization dynamics have led to lower subsequent exposures; in economies with low initial exposure, structural transformation and on-shoring of routine tasks raised subsequent exposure.
  - For developed economies: technological advancement and increased offshoring are proximate causes of declining routine exposures (polarization).
  - For developing economies: structural transformation and expanded vertical trade are main drivers raising exposures (anti-polarization forces during the sample period), but rising exposures imply growing future vulnerability.
- Risks and future outlook for developing economies:
  - Rapid technology adoption in some developing economies (e.g., rising industrial robotization in China) may increase automation impacts.
  - Potential for premature deindustrialization, erosion of middle-skilled employment early in convergence, adverse productivity and growth effects, and derailing income convergence (Rodrik 2016).
  - Preliminary modeling (Das and Hilgenstock 2018): polarization not imminent under some assumptions, but factors gathering force over ten to twenty years.
- Policy challenges and considerations:
  - Increase supply of high-skilled labor, raise investment in technological proficiency, and build safety buffers (Dao et al. 2017).
  - Distributional concerns: technological change could shift income toward high-wage individuals, exacerbating inequality and raising policy trade-offs in developing economies.

### Appendix and data
- Exposures to offshorability constructed analogously to routinization exposures; online appendix referenced in source text for details.
- Database coverage: exposures constructed for 160 countries at aggregate level and 126 countries at industry level between 1960 and 2015 (various frequencies). Table A1 lists country coverage and years (as of 2018).

*Source: wp18135 - Section 3 discusses the drivers and mechanisms which lie behind polarization in developed economies (IMF working paper).*

### REFERENCES

### REFERENCES

### Labor market polarization, skills, and technology
- Acemoglu, D, and D. H. Autor. 2011. “Skill, Tasks and Technologies: Implications for Employment and Earnings.” In Handbook of Labor Economics, Volume 4. Elsevier: Amsterdam. 
- Autor, D. H., and D. Dorn. 2013. "The growth of low-skill service jobs and the polarization of the US labor market." American Economic Review 103 (5): 1553–1597. 
- Autor, D., H., Frank Levy, and R. J. Murnane. 2003. “Computer-based technological change and skill demands: Reconciling the perspectives of economists and sociologists.” In Low-wage America: How employers are reshaping opportunity in the workplace. Russell Sage Foundation, New York. 
- Autor, D.H, L. F. Katz, and M. S. Kearney. 2006. “The Polarization of the U.S. Labor Market.” NBER Working Paper No. 11986. National Bureau of Economic Research. Cambridge. 
- Bárány, Z and C. Siegel. 2018. “Job polarization and structural change”. American Economic Journal: Macroeconomics, 10, 57-89.  
- Beaudry, P., D. Green, and B. M. Sand. 2016.  "The Great Reversal in the Demand for Skill and Cognitive Tasks," Journal of Labor Economics 34, no. S1 (Part 2, January 2016): S199-S247. 
- Firpo, S., N. M. Fortin, and T. Lemieux. 2011. “Occupational tasks and changes in the wage structure”. University of British Columbia. 
- Goos, M., and A. Manning. 2007. “Lousy and Lovely Jobs: The Rising Polarization of Work in Britain.” The Review of Economics and Statistics 89 (1): 118–133. 
- Goos, M, A. Manning, and A. Salomons. 2014. “Explaining Job Polarization: Routine-Biased Technological Change and Offshoring.” The American Economic Review 104 (8): 2509–2526. 
- Ikenaga, T. and Kamibayashi, R. 2016. “Task Polarization in the Japanese Labor Market: Evidence of a Long-Term Trend”. Industrial Relations, 55, 267–293. 
- Ikenaga, T. and Kamibayashi, R. 2016. “Task Polarization in the Japanese Labor Market: Evidence of a Long-Term Trend”. Industrial Relations, 55, 267–293.
- Karabarbounis, L, and B. Neiman. 2014. “The Global Decline of the Labor Share.” The Quarterly Journal of Economics 129 (1): 61–103. 
- Katz, L and K. Murphy. 1992. “Changes in Relative Wages, 1963–1987: Supply and Demand Factors.” The Quarterly Journal of Economics 107 (1): 35–78. 
- Levy, F, and R. Murnane. 1996. “With What Skills Are Computers a Complement?” The American Economic Review, 86, 258-62.  
- Michael, G, A. Natraj, and J.Van Reenen. 2013. “Has ICT Polarized Skill Demand? Evidence from Eleven Countries over 25 Years.” NBER Working Paper No. 16138. 
- Reijnders, L. and de Vries, G. 2017. “Job Polarization in Advanced and Emerging Countries”. University of Groningen Working Paper.  
- Spitz‐Oener, A. 2006. "Technical Change, Job Tasks, and Rising Educational Demands: Looking outside the Wage Structure," Journal of Labor Economics 24, no. 2 (April 2006): 235-270. 
- Frey, C and M. Osborne. 2017. “The Future of Employment: How Susceptible are Jobs to Computerisation". Technological Forecasting and Social Change, 114, 254-280.  

### Automation, robotics, and welfare implications
- Eden, M and P. Gaggl. 2015. “On the Welfare Implications of Automation.” Policy Research Working Paper 7487. World Bank, Washington, D.C. 
- International Federation of Robotics, 2017. “World Robotics 2017”. 
- Das, M and B. Hilgenstock. 2018. “When Cheap Labor meets Cheap Robots: the Future of labor markets in Developing Economies”, mimeo, IMF.  
- World Bank. 2016. World Development Report 2016: Digital Dividends. Washington DC.  

### Globalization, offshoring, and trade
- Blinder, A. and A. Krueger, 2013. "Alternative Measures of Offshorability: A Survey Approach," Journal of Labor Economics, University of Chicago Press, vol. 31(S1), pages S97 - S128. 
- Feenstra, R. C. 2007. “Globalization and its Impact on Labor.” Presented as the Global Economy Lecture, Vienna Institute for International Economics Studies 
- Feenstra, R. C., and G. H. Hanson. 1996. "The impact of outsourcing and high-technology capital on wages: estimates for the United States, 1979–1990." The Quarterly Journal of Economics 114 (3): 907–940. 
- Goldberg, P. and N. Pavcnik. 2007. “Distributional Effects of Globalization in Developing Countries”. Journal of Economic Literature, 45, 39-82.  
- Koopman R., Z. Wang and S. Wei. 2014. “Tracing Value-Added and Double Counting in Gross Exports”. American Economic Review, 104, 459-94.  
- Maloney, W. and C. Molina. 2016. Are Automation and Trade Polarizing Developing Country Labor Markets, Too? Policy Research Working Paper 7922. World Bank, Washington DC. 
- Comin, D and M. Mestiere. 2013. “If Technology Has Arrived Everywhere, Why Has Income Diverged?” working paper, Harvard Business School. 
- Dao, M, M. Das, Z. Koczan and W. Lian. 2017. “Why is Labor Receiving a Smaller Share of Global Income? Theory and Evidence”, IMF WP Series  

### Growth, structural transformation, and methodology
- Herrendorf, B, C. Harrington, and A.Valentinyi. 2014. “Growth and Structural Transformation,” in Philippe Aghion and Steven N. Durlauf, eds., Handbook of Economic Growth, Vol. 2, 855–941.  
- Hsieh, C-T and P. J. Klenow. 2007. "Relative Prices and Relative Prosperity." American Economic Review,97, 562-585. 
- Nordhaus, W. 2007. Two centuries of productivity growth in computing. The Journal of Economic History, 67, 128-159.  
- Rodrik, D.  2016. "Premature deindustrialization," Journal of Economic Growth, 21, 1, 1-33. 
- Foster, A and M. Rosenzweig. 1995. “Learning by Doing and Learning from Others: Human Capital and Technical Change in Agriculture”. Journal of Political Economy, 103, 1176-1209. 
- Hallward-Driemeier, M. and G. Nayyar. 2018. “Trouble in the Making? The Future of Manufacturing-Led Development”. Washington DC: World Bank.  

### Institutional reports and IMF works
- International Labor Organization. 2014. “World of Work Report 2014: Developing with Jobs”.  
- International Monetary Fund. 2018A. “Labor Force Participation in Advanced Economies: Drivers and Prospects”, Chapter 3, World Economic Outlook, forthcoming April 2018. 
- International Monetary Fund. 2018B. “Youth Labor Market Prospects in Emerging Markets and Developing Economies: Drivers and Policies”, Staff Discussion Note, forthcoming. 
- International Monetary Fund. 2018C. “Technology and the Future of Work”. Forthcoming. 

*Source: wp18135 - REFERENCES (wp18135 - REFERENCES).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp18135.pdf_