## wpiea2021011-print-pdf

## Source details

**Canonical URL:** [wpiea2021011-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021011-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021011-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021011-print-pdf.pdf.json)

---

### I. Introduction and research questions
- Context:
  - Automation, notably industrial robots, was accelerating prior to COVID-19 and may boost productivity while displacing workers, particularly low-skilled workers (Acemoglu and Restrepo, 2020).
  - COVID-19 has increased telework and forced automation; surveys show over 80 percent of large-company leaders accelerating digitalization and 50 percent indicating they will accelerate automation (World Economic Forum, 2020).
- Research questions:
  - Will COVID-19 accelerate robotization?
  - What will be the distributional impact of robotization following pandemics?
- Conceptual links:
  - Large shocks (recessions, pandemics) can catalyze reallocation toward labor-saving technologies by lowering opportunity costs of adjustment (Hall, 2005; Mortensen and Pissarides, 1994; Berger, 2012).
  - Pandemics can incentivize automation despite demand declines (Leduc and Liu, 2020); jobs with high automation potential and infection risk (e.g., for women with mid to low wages/education) are especially vulnerable (Chernoff and Warman, 2020).
  - Pandemics have been associated with persistent increases in inequality (Furceri, Loungani, Ostry and Pizzuto, 2020); this paper links that increase to accelerated robotization.

### II. Data and sample
- Robot data:
  - Source: International Federation of Robotics (IFR, 2018a).
  - Definition: industrial robot = “an automatically controlled, reprogrammable, multipurpose, manipulator programmable in three or more axes…”
  - IFR provides shipment and stock data for 75 countries with industry breakdown by ISIC; earliest data start in 1993.
  - Analysis focuses on industrial robots (excludes service robots).
- Employment, wages, capital:
  - Source: World Input-Output Database (WIOD), Socio-Economic Accounts (2016).
  - Labor data: 2000–2014 for 43 countries and 56 industries at 2-digit ISIC rev.4.
  - Combined IFR + WIOD coverage used: 18 industries in 40 countries for 2000-2014.
  - Robot data availability extends to 2018, permitting analysis up to four years after 2014.
- Inequality and pandemic measures:
  - Inequality: Standardized World Income Inequality Database (SWIID 8.2) for disposable net Gini coefficients.
  - Pandemics in coverage: SARS (2003), H1N1 (2009), MERS (2012), Ebola (2014).
  - H1N1: more than 6,000,000 confirmed cases across 148 countries and about 19,000 fatalities.
  - Pandemic measures: WHO-declaration dummy; alternatives include confirmed cases and deaths.

### III. Recent trends in robot usage
- Aggregate patterns:
  - Industrial robot use increased dramatically since the Global Financial Crisis (GFC), with acceleration over the last decade.
  - About two-thirds of new robots in recent years were installed in Asia, including about a half in China.
- Robot density leaders:
  - Korea, Singapore and Taiwan POC lead globally, followed by Germany and Japan.
- Industry concentration:
  - Robot usage concentrated in manufacturing; automotive manufacturing is the most automated sub-industry by far; electronics manufacturing is highly automated in some Asian economies.

### IV. Do pandemics accelerate robotization? — Empirical evidence
- Empirical approach:
  - Method: local projection method (Jordà, 2005) estimating impulse response functions for horizons k=0,...,4 (up to four years after shock).
  - Dependent variable: cumulative new robot installations between t and t+k normalized by employment at time t (new robots per 1000 employees).
  - Sample: unbalanced panel of 18 industries in 40 countries over 2000-2014 (robot data available to 2018).
  - Controls: three lags of dependent variable and pandemic dummy; industry and country fixed effects; five-year dummies; world real GDP growth. Robustness checks include log wages, capital-to-wage ratio, GDP per capita, share age 65+, trade and financial globalization. Standard errors clustered at country-industry pair.
- Main estimates (baseline impulse response):
  - In the second year after a pandemic: about 0.35 more new robots installed per 1000 employees (cumulative).
  - Four years after a pandemic: about 0.7 more new robots installed per 1000 employees (cumulative).
- Robustness and alternative specifications:
  - Including additional controls produces very similar results.
  - Excluding H1N1 yields similar coefficient sizes but less precise estimates due to fewer episodes.
  - Pandemic intensity (number of infection cases and deaths): more intense pandemics have greater impact on robot adoption.
  - Controlling for recessions yields broadly similar results.
  - Using log of robot adoption: four years after a pandemic, robot adoption is about 20 percent higher compared to a no-pandemic event.
- Nonlinearity by severity:
  - Smooth transition autoregressive model with severity indicator z (log confirmed cases, log deaths, or real GDP growth) and γ=3.5.
  - Finding: pandemics with severe health consequences (larger cases or deaths) and those accompanied by significant economic downturn have statistically significant positive effects on robot adoption; mild pandemics do not.

### V. Pandemics, robotization, and inequality — Evidence and magnitudes
- Empirical strategy for inequality:
  - Regress changes in log net Gini (G_{c,t+k} − G_{c,t−1}) on pandemic indicators interacted with robot adoption level (robot stock per 1000 employees), allowing coefficients to vary by robot density terciles (H, M, L).
  - Controls: three lags, country fixed effects, time dummies, global business cycle. Standard errors clustered at the country level.
- Key empirical findings:
  - Robot adoption increases following pandemics, especially when health impact is severe and associated with economic downturn.
  - The increase in inequality (net Gini) over the medium term is larger for economies with higher robot density and where new robot adoption has increased more.
  - The increase in inequality tends to be higher, and statistically significant, for cases with high robot density than with low density.
  - Similar results obtained when using the pace of robot adoption (cumulative sum of new robot installation over the next 5 years normalized by employment).
- Quantitative evidence and tests:
  - Sample for inequality analysis: 66 countries over 1993-2014.
  - Granger causality:
    - Null that pandemic dummy does not Granger-cause Gini is rejected at the 90 percent confidence level (p-value=0.03).
    - Null that Gini does not Granger-cause pandemic dummy is not rejected (p-value=0.22).
  - Figure 5 estimation: sample 66 economies, period 1993-2014, robotization groups are top, medium, bottom 1/3 by robot density or pace of adoption; dependent variable changes in net Gini in logs.
- Table A.1 sample descriptive statistics (Sample = 18 industries in 40 economies, period 2000-2014; Obs. = 4320 unless otherwise shown):
  - Annual robot installation: Med. 9.0; Mean 226.6; Std. dev. 1172.2.
  - Annual robot installation, per 1000 workers: Med. 0.1; Mean 0.8; Std. dev. 2.3.
  - Robot operating stock: Med. 74.5; Mean 1994.3; Std. dev. 10431.4.
  - Robot operating stock, per 1000 workers: Med. 8.1; Mean 70.1; Std. dev. 226.5.
  - Number of confirmed cases: Med. 0.0; Mean 22909.1; Std. dev. 228611.2.
  - Number of deaths related to pandemics: Med. 0.0; Mean 35.9; Std. dev. 255.3.
  - Real GDP growth: Med. 1.8; Mean 1.5; Std. dev. 3.0.
  - Net Gini SIID 8.2 2119 29.4 29.4 4.9
  - Pandemic events (N.Obs with Pand. Dumm. = 1): SARS (2003) = 78; H1N1 (2009) = 392; MERS (2012) = 164; Ebola (2014) = 72.

### VI. Mechanisms
- Distributional channel:
  - Acceleration in automation following pandemics can worsen inequality because robots disproportionately affect routine and manual tasks typically performed by mid- and low-skilled workers.
  - Low-skilled workers face higher displacement risk; robots create new tasks that benefit mainly high-skilled workers, reinforcing inequality dynamics (Acemoglu and Restrepo, 2020).
- Complementary findings:
  - Results are consistent with literature showing past pandemics have raised inequality and led to increases in the Gini coefficient.

### VII. Policy implications and recommendations
- General observation:
  - Large shocks that combine health and economic crises can accelerate adoption of labor-saving technologies and contribute to jobless recoveries, with persistent employment and inequality consequences.
  - Pandemic severity heterogeneity matters: more severe health or economic impacts are more likely to accelerate robotization and its distributional consequences.
- Short-term policies:
  - Prevent labor market scarring through better and smarter targeting of limited fiscal support for people and firms; "Targeted support provides greater bang for the buck, both in protecting lives and livelihoods (Jurzyk et al., 2020)."
- Medium-term and structural responses:
  - Revamp education to meet demand for more flexible skill sets and lifelong learning.
  - Provide new training, especially targeted at the most adversely affected workers.
  - Reduce skill mismatches between workers and jobs.
  - Address medium-term social challenges through income redistribution and safety nets.
- Considered policy tools (literature surveyed):
  - Raising unemployment insurance benefits.
  - Introducing a universal basic income.
  - Increasing transfers to labor force non-participants.
  - Making the tax system more progressive.
  - Taxing robots.
- Tradeoffs and welfare implications:
  - Taxing robots may be optimal in the short run to protect routine workers who cannot acquire non-routine skills, but not optimal in the long run as it disincentivizes future acquisition of non-routine skills (Guerreiro, Rebelo and Teles, 2020).
  - Raising unemployment insurance benefits modestly can improve average welfare; introducing a universal basic income and increasing transfers to non-participants can impose large welfare losses on high-wage workers and be very costly in terms of aggregate income, whereas making the tax system more progressive can achieve much redistribution with smaller welfare losses (Jaimovich, Saporta-Eksten, Siu and Yedid-Levi, 2020).
  - Policies that soften labor market impacts of new technologies and support transitions can improve welfare and allow faster innovation while ensuring outcomes are welfare-improving for society (Korinek and Stiglitz, 2018).

*Source — wpiea2021011-print-pdf - References*

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

### References

### I. Introduction and related literature
- Context:
  - Automation, notably industrial robots, was accelerating prior to COVID-19 and may boost productivity while displacing workers, particularly low-skilled workers (Acemoglu and Restrepo, 2020).
  - COVID-19 has increased telework and forced automation, with surveys showing over 80 percent of large-company leaders accelerating digitalization and 50 percent indicating they will accelerate automation (World Economic Forum, 2020).
- Research questions:
  - Will COVID-19 accelerate robotization?
  - What will be the distributional impact of robotization following pandemics?
- Key conceptual links:
  - Large shocks (recessions, pandemics) can catalyze reallocation toward labor-saving technologies by lowering opportunity costs of adjustment (Hall, 2005; Mortensen and Pissarides, 1994; Berger, 2012).
  - Prior findings: pandemics can incentivize automation despite demand declines (Leduc and Liu, 2020); jobs with high automation potential and infection risk (e.g., for women with mid to low wages/education) are especially vulnerable (Chernoff and Warman, 2020).
  - Pandemics have been associated with persistent increases in inequality (Furceri, Loungani, Ostry and Pizzuto, 2020); this paper links that increase to accelerated robotization.

### II. Data
- Robot data:
  - Source: International Federation of Robotics (IFR, 2018a).
  - Definition: industrial robot = “an automatically controlled, reprogrammable, multipurpose, manipulator programmable in three or more axes…”
  - IFR provides shipment and stock data for 75 countries with industry breakdown by ISIC; earliest data start in 1993.
  - Analysis focuses on industrial robots (excludes service robots due to sparse coverage).
- Employment, wages, capital:
  - Source: World Input-Output Database (WIOD), Socio-Economic Accounts (2016), labor data 2000–2014 for 43 countries and 56 industries at 2-digit ISIC rev.4.
  - Combined IFR + WIOD coverage used: 18 industries in 40 countries for the period 2000-2014 (Table A.1).
  - Robot data availability extends to 2018, permitting analysis of impacts up to four years after 2014.
- Inequality data:
  - Source: Standardized World Income Inequality Database (SWIID 8.2) for disposable (net, post tax and transfer) Gini coefficients.
- Pandemic identification:
  - Four pandemics within coverage: SARS in 2003, H1N1 in 2009, MERS in 2012, Ebola in 2014.
  - H1N1: more than 6,000,000 confirmed cases across 148 countries and about 19,000 fatalities.
  - Pandemic measures: dummy when WHO declares a pandemic; alternative measures include confirmed cases and deaths.

### III. Recent trends in robot usage
- Aggregate patterns:
  - Industrial robot use increased dramatically since the Global Financial Crisis (GFC), with acceleration over the last decade.
  - About two-thirds of new robots in recent years were installed in Asia, including about a half in China.
- Robot density (stock per 1000 workers) leaders:
  - Korea, Singapore and Taiwan POC are global leaders, followed by Germany and Japan.
- Industry concentration:
  - Robot usage concentrated in manufacturing; automotive manufacturing is the most automated sub-industry by far. In some Asian economies robot density is also high in electronics manufacturing.

### IV. Do pandemics accelerate robotization?
- Empirical approach:
  - Method: local projection method (Jordà, 2005) estimating impulse response functions for horizons k=0,...,4 (up to four years after shock).
  - Dependent variable: cumulative new robot installations between t and t+k normalized by employment at time t (new robots per 1000 employees).
  - Sample: unbalanced panel of 18 industries in 40 countries over 2000-2014 (robot data available to 2018).
  - Controls: three lags of dependent variable and pandemic dummy; industry and country fixed effects; five-year dummies; world real GDP growth. Robustness checks include industry log wages, capital-to-wage ratio, GDP per capita, population aging (share age 65+), trade and financial globalization; standard errors clustered at country-industry pair.
- Main estimates (baseline impulse response):
  - In the second year after a pandemic: about 0.35 more new robots installed per 1000 employees (cumulative).
  - Four years after a pandemic: about 0.7 more new robots installed per 1000 employees (cumulative).
- Robustness checks and alternative specifications:
  - Including additional controls (log wages, capital-to-wage ratio, GDP per capita, trade and financial globalization, share age 65+) produces very similar results (Table A.3 and Figure A.1).
  - Excluding H1N1 yields similar coefficient sizes but less precise estimates due to fewer episodes (Figure A.1).
  - Using pandemic intensity measures (number of infection cases and deaths) shows more intense pandemics have greater impact on robot adoption (Figure A.2).
  - Controlling for recessions (recession dummy) yields broadly similar results (Figure A.3).
  - Using log of robot adoption: four years after a pandemic, robot adoption is about 20 percent higher compared to a no-pandemic event (Figure A.4).
- Nonlinearity by severity:
  - Specification: smooth transition autoregressive model allowing pandemic effects to vary with severity indicators z (log confirmed cases, log deaths, or real GDP growth), with weighting function F(z) and parameter γ=3.5.
  - Findings (Figure 4, Table A.3): pandemics with severe health consequences (larger cases or deaths) and those accompanied by significant economic downturn have statistically significant positive effects on robot adoption; mild pandemics do not show significant effects.

### V. Pandemics, robotization, and inequality
- Empirical strategy:
  - Regress changes in log net Gini (G_{c,t+k} − G_{c,t−1}) on pandemic indicators interacted with robot adoption level (robot stock per 1000 employees), allowing coefficients to vary by robot density/adoption terciles (high, medium, low).
  - Equation structure: coefficients on pandemic dummy vary with R_{c,t}^{q} ∈ {H, M, L}; controls include lags and fixed effects.
- Key finding summarized in preceding sections:
  - Robot adoption increases following pandemics, especially when health impact is severe and associated with economic downturn.
  - The increase in inequality (net Gini) over the medium term is larger for economies with higher robot density and where new robot adoption has increased more.
- Implication:
  - Distributional effects of COVID-19 could be sizeable through an acceleration of robotization; as automation accelerates from still low levels, it may become an increasingly important driver of inequality.

### Policy implications and concluding observations (as discussed in the chapter)
- Large shocks that combine health and economic crises can accelerate adoption of labor-saving technologies and contribute to jobless recoveries, with persistent employment and inequality consequences.
- The heterogeneity of pandemic severity matters: policy responses should consider that more severe health or economic impacts are more likely to accelerate robotization and its distributional consequences.
- Monitoring robot adoption and targeting policies to support displaced low-skilled workers will be important to mitigate inequality effects of pandemic-induced automation.

*Italic: Source — wpiea2021011-print-pdf - References*

### 2.3 robots per 1000 employees in the sample), intermediate (middle 1/3, M), and low (bottom

### wpiea2021011-print-pdf - 2.3 robots per 1000 employees in the sample), intermediate (middle 1/3, M), and low (bottom

### Key findings on inequality and robotization
- The increase in inequality tends to be higher, and statistically significant, for cases with high robot density than with low density (Figure 5, left).
- Similar results are obtained when using the pace of robot adoption: inequality rises faster following pandemics where robot adoption is at a faster pace (Figure 5, right).
- These results suggest that the robot adoption following pandemics could have a sizable impact on inequality.
- Results are robust when allowing smooth transitions where the transition is a function of robot density or the pace of robot adoption (Figure A.5).
- The findings are consistent with literature showing past pandemics have raised inequality and led to increases in the Gini coefficient.

### Quantitative results and statistical evidence
- Sample used for inequality analysis: 66 countries over 1993-2014.
- Granger causality test results:
  - Null that pandemic dummy does not Granger-cause Gini is rejected at the 90 percent confidence level (p-value=0.03).
  - Null that Gini does not Granger-cause pandemic dummy is not rejected (p-value=0.22).
- Figure 5 estimation details:
  - Sample: 66 economies over the period 1993-2014.
  - Method: local projection method (Jorda, 2005).
  - Robotization groups: top, medium, and bottom 1/3 by robot density or pace of adoption (cumulative sum of new robot installation over the next 5 years, normalized by employment).
  - Dependent variable: changes in net Gini in logs.
  - Controls: three lags of dependent variable and pandemic dummy, country fixed effects, time dummies, global business cycle. Standard errors clustered at the country level.
- Table A.1 descriptive statistics (Sample = 18 industries in 40 economies, period 2000-2014; Obs. = 4320 unless otherwise shown):
  - Annual robot installation: Med. 9.0; Mean 226.6; Std. dev. 1172.2.
  - Annual robot installation, per 1000 workers: Med. 0.1; Mean 0.8; Std. dev. 2.3.
  - Robot operating stock: Med. 74.5; Mean 1994.3; Std. dev. 10431.4.
  - Robot operating stock, per 1000 workers: Med. 8.1; Mean 70.1; Std. dev. 226.5.
  - Number of confirmed cases: Med. 0.0; Mean 22909.1; Std. dev. 228611.2.
  - Number of deaths related to pandemics: Med. 0.0; Mean 35.9; Std. dev. 255.3.
  - Real GDP growth: Med. 1.8; Mean 1.5; Std. dev. 3.0.
  - Net Gini SIID 8.2 2119 29.4 29.4 4.9
  - Pandemic events (N.Obs with Pand. Dumm. = 1): SARS (2003) = 78; H1N1 (2009) = 392; MERS (2012) = 164; Ebola (2014) = 72.

### Mechanisms linking pandemics, robot adoption, and inequality
- Acceleration in automation following pandemics can worsen inequality because robots do not affect all workers equally.
- Jobs most susceptible to automation involve routine and manual tasks, traditionally performed by workers with mid-level skills or low-skilled workers (Graetz and Michaels, 2018; Autor and Reynolds, 2020).
- Low-skilled workers, who tend to have lower earnings, face higher displacement risk from robots than high-skilled workers.
- Robots create new tasks but benefit mainly high-skilled workers, reinforcing inequality dynamics (Acemoglu and Restrepo, 2020).
- Through uneven impacts across skill levels, robot adoption can contribute to rises in inequality.

### Policy recommendations and considerations
- Short-term: prevent scarring in the labor market through better and smarter targeting of limited fiscal support for people and firms; "Targeted support provides greater bang for the buck, both in protecting lives and livelihoods (Jurzyk et al., 2020)."
- Medium-term and structural responses:
  - Revamp education to meet the demand for more flexible skill sets and lifelong learning.
  - Provide new training, especially targeted at the most adversely affected workers.
  - Reduce skill mismatches between workers and jobs.
  - Address medium-term social challenges through income redistribution and safety nets.
- Considered policy tools in the literature:
  - Raising unemployment insurance benefits.
  - Introducing a universal basic income.
  - Increasing transfers to labor force non-participants.
  - Making the tax system more progressive.
  - Taxing robots.
- Tradeoffs and welfare implications:
  - Taxing robots may be optimal in the short run to protect routine workers who cannot acquire non-routine skills, but not optimal in the long run as it disincentivizes future acquisition of non-routine skills (Guerreiro, Rebelo and Teles, 2020).
  - Raising unemployment insurance benefits modestly can improve average welfare; introducing a universal basic income and increasing transfers to non-participants can impose large welfare losses on high-wage workers and be very costly in terms of aggregate income, whereas making the tax system more progressive can achieve much redistribution with smaller welfare losses (Jaimovich, Saporta-Eksten, Siu and Yedid-Levi, 2020).
  - Policies that soften labor market impacts of new technologies and support transitions can improve welfare and allow faster innovation while ensuring outcomes are welfare-improving for society (Korinek and Stiglitz, 2018).

### Methods and robustness
- Robot adoption analysis:
  - Sample: 18 industries in 40 economies, period 2000-2014 (impulse responses estimated with local projection method (Jorda, 2005)).
  - Dependent variable in robot adoption regressions: new robot installations per 1000 employees in cumulative terms between t and t+k.
  - Pandemic dummy defined from Furceri et al. (2020).
  - Controls: three lags of dependent variable and pandemic dummy, industry and country fixed effects, five-year dummies, global business cycle; additional specifications include industry-level and macro-level controls, aging control, and excluding H1N1.
  - Standard errors clustered at the country-industry pair level.
- Multiple robustness checks reported across figures and tables (e.g., impact by pandemic severity, by cases and mortality, interaction with recessions, log specifications, smooth transition specifications).

*Source: wpiea2021011-print-pdf - 2.3 robots per 1000 employees in the sample), intermediate (middle 1/3, M), and low (bottom*

### References

### References

### Automation, Robots, and Labor Market Polarization
- Acemoglu, Daron (1991). “Changes in Unemployment and Wage Inequality: An Alternative Theory and Some Evidence”. The American Economic Review, Vol. 89, No. 5 (Dec., 1999), pp. 1259-1278
- Acemoglu, Daron, and David Autor (2011). “Skills, Tasks and Technologies: Implications for Employment and Earnings.” In Handbook of Labor Economics, Vol. 4b, edited by O. Ashenfelter, R. Layard, and D. Card, 1043–1171. Amsterdam: Elsevier.
- Acemoglu, Daron, and Pascual Restrepo (2020). "Unpacking Skill Bias: Automation and New Tasks." AEA Papers and Proceedings, 110: 356-61.
- Autor, David H (2015). Why are there still so many jobs? The history and future of workplace automation. Journal of Economic Perspectives 29, no. 3:3–30.
- Autor, David H. and David Dorn (2013). The growth of low-skill service jobs and the polarization of the US labor market. American Economic Review 103, no. 5:1553–1597.
- Autor, David, H., Lawrence F. Katz, and Melissa S. Kearney (2006). "The Polarization of the U.S. Labor Market." American Economic Review, 96 (2): 189-194.
- Goos, M. and A. Manning (2007). Lousy and lovely jobs: The rising polarization of work in Britain. Review of Economics and Statistics 89 (1), 118-133.
- Goos, M., A. Manning, and A. Salomons (2009). Job polarization in Europe. American Economic Review: Papers & Proceedings 99 (2), 58-63.
- Goos, M., A. Manning, and A. Salomons (2014). Explaining job polarization: Routine-biased technological change and ofshoring. American Economic Review 104 (8), 2509-26.
- Jaimovich, Nir and Henry E. Siu (2020). Job polarization and jobless recoveries. Review of Economics and Statistics 102, no. 1:129–147.
- Jaimovich, Nir and Saporta Eksten, Itay and Siu, Henry and yedid-levi, yaniv (2020). The Macroeconomics of Automation: Data, Theory, and Policy Analysis (May 2020). NBER Working Paper No. w27122, Available at SSRN: https://ssrn.com/abstract=3597854
- Graetz, Georg, and Guy Michaels (2018). “Robots at Work”. Review of Economics and Statistics Volume 100. Issue 5. December 2018 p.753-768.
- International Federation of Robotics (2018a). “World Robotics 2018–Industrial Robots.” https://ifr.org.
- International Federation of Robotics (2018b). “ World Robotics 2018–Service Robots.” https://ifr.org.
- Guerreiros, J, S Rebelo and P Teles (2020). “Should Robots Be Taxed?” NBER Working Paper 23806, Sep 2017, revised January 2019.

### COVID-19, Forced Automation, Inequality, and Recovery Dynamics
- Autor, David and Elisabeth Reynolds ( 2020). “The Nature of Work after the COVID Crisis: Too Few Low-Wage Jobs”.
- Caselli Mauro, Andrea Fracasso and Silvio Traverso (2020). Mitigation of risks of covid-19 contagion and robotisation: Evidence from Italy. Covid Economics 17:174–188.
- Chernoff, Alex W and Warman, Casey (2020). "COVID-19 and Implications for Automation,” NBER Working Paper No. W27249, July 2020. http://www.nber.org/papers/w27249
- Ding, Lei and Julieth Saenz Molina (2020). “Forced Automation” by COVID-19? Early Trends from Current Population Survey Data,” Federal Reserve Bank of Philadelphia September 2020: https://www.philadelphiafed.org/-/media/frbp/assets/community-development/discussion-papers/forced-automation-by-covid-19.pdf
- Furceri, Davide, Prakash Loungani, Jonathan D. Ostry, and Pietro Pizzuto (2020). Will COVID-19 affect inequality? Evidence from past pandemics. COVID Economics Vetted and Real-Time Papers, Issue 12, 138–157. London: Centre for Economic and Policy Research.
- Leduc, Sylvain and Zheng Liu (2020). Can pandemic-induced job uncertainty stimulate automation? Working Paper 2020-19, Federal Reserve Bank of San Francisco.
- Ma, Chang and Rogers, John H. and Zhou, Sili (2020). “Modern Pandemics: Recession and Recovery” Available at SSRN: https://ssrn.com/abstract=3565646 or http://dx.doi.org/10.2139/ssrn.3565646
- Jurzyk, Emilia, Medha Madhu Nair, Nathalie Pouokam, Tahsin Saadi Sedik, Anthony Tan, Irina Yakadina (2020). “COVID-19 and Inequality in Asia: Breaking the Vicious Cycle”, IMF Working Paper WP/20/217.
- Saadi Sedik, Tahsin and Rui Xu (2020). “A Vicious Cycle: How Pandemics Lead to Economic Despair and Social Unrest” IMF Working Paper WP/20/216.

### Labor Market Dynamics, Unemployment, and Wage/Policy Models
- Bernanke, B. S. (2003). The jobless recovery: http://www.federalreserve.gov/boarddocs/speeches/2003/200311062/default.htm
- Hall, Robert E (2005). “Employment Fluctuations with Equilibrium Wage Stickiness.” American Economic Review 95 (1): 50–65.
- Mortensen, Dale T., and Christopher A. Pissarides (1994). “Job Creation and Job Destruction in the Theory of Unemployment.” Review of Economic Studies 61 (3): 397–415.
- Hershbein, Brad and Lisa B. Kahn (2018). Do recessions accelerate routine-biased technological change? Evidence from vacancy postings. American Economic Review 108, no. 7:1737–1772.
- Groshen, E. L. and S. Potter (2003). Has structural change contributed to a jobless recovery? Current Issues in Economics and Finance, Federal Reserve Bank of New York 9 (8), 1-7.
- Jaimovich, Nir and Henry E. Siu (2020). Job polarization and jobless recoveries. Review of Economics and Statistics 102, no. 1:129–147.
- Tenreyro, S., G. Thwaites ( 2016). “Pushing on a String: US Monetary Policy Is Less Powerful in Recessions.” American Economic Journal: Macroeconomics, 8 (4): 43-74.

### Inequality Measurement, International Production, and Digital/Regional Forces
- Korinek, Anton and Stiglitz, Joseph, (2018). Artificial Intelligence and Its Implications for Income Distribution and Unemployment, NBER Chapters, in: The Economics of Artificial Intelligence: An Agenda, pages 349-390, National Bureau of Economic Research, Inc.
- Solt, Frederick (2020). “Measuring Income Inequality Across Countries and Over Time: The Standardized World Income Inequality Database.” Social Science Quarterly 101(3):1183-1199. SWIID Version 8.3, May 2020.
- Timmer, Marcel. P., Erik Dietzenbacher, Bart Los, Robert Stehrer, and Gaaitzen J. de Vries (2015). “An Illustrated User Guide to the World Input-Output Database: The Case of Global Automotive Production.” Review of International Economics 23 (3): 575–605.
- International Monetary Fund (2018). “The Digital Revolution in Asia: Disruptor or New Growth Engine (or Both)?” Asia and Pacific Department Regional Economic Outlook Background Paper No. 4. Washington, DC, October.
- World Economic Forum (2020). “The Future of Jobs Report 2020”: https://www.weforum.org/reports/the-future-of-jobs-report-2020.

### Econometric Methods and Modeling Techniques
- Granger, Clive W. J. & Terasvirta, Timo (1993). "Modelling Non-Linear Economic Relationships," OUP Catalogue, Oxford University Press, number 9780198773207.
- Teräsvirta, T. (1998). Modelling economic relationships with smooth transition regressions. In Handbook of Applied Economic Statistics; Marcel Dekker: New York, 1998; 507–552.
- Jordà, O. (2005). “Estimation and inference of impulse responses by local projections,” ` American Economic Review, 95, 161–182.
- Leduc, Sylvain and Zheng Liu (2020). Can pandemic-induced job uncertainty stimulate automation? Working Paper 2020-19, Federal Reserve Bank of San Francisco.

*References section from wpiea2021011-print-pdf*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021011-print-pdf.pdf_