The effect of automation levels on US interstate migration


This study investigates the extent to which job process automation, which has resulted in wage inequality and job polarization in the USA and has affected US interstate migration over the past two decades. The level of automation in each state is calculated using data on the degree of automation of each occupation. In particular, this study examines how the difference in the levels among states explains the movement of migrants. The results show that people move to states with more automation in skilled occupations and less automation in unskilled occupations. This finding implies that automation has a complementary (substitution) effect on skilled (unskilled) occupations. The results also show that the former effect is larger and more robust than the latter one. Further analyses use migration flow data classified into several subgroups and find that both skilled and unskilled workers in most occupations move to states with more automation in skilled occupations and less automation in unskilled occupations.

This is a preview of subscription content, log in to check access.

Fig. 1


  1. 1.

    Autor and Dorn (2013) also examine the effect of computerization on the difference in education shares between migrant and non-migrant workers, but do not focus on the direction of migration flows.

  2. 2.

    Autor and Dorn (2013) assume that only high-skilled workers can migrate across regions, whereas low-skilled workers cannot. However, they state that the similar result holds when low-skilled workers can also migrate.

  3. 3.

    The previous title of the database was the Dictionary of Occupational Titles, which has been used in the literature of automation such as Autor et al. (2003), Autor and Dorn (2013), and Berger and Frey (2016).

  4. 4.

    Handel (2016) reports the detailed data collection method of O*NET.

  5. 5.

    Here, the mean, standard deviation, max, and min of the degree of automation are calculated for the standard occupational classification (SOC)-level occupations in O*NET-SOC, especially the occupations used for the analyses in this study.

  6. 6.

    The education level of each automated occupation can be identified using the “Typical education needed for entry” from the Occupational Projections and Training Data (Employment Projections: 2010-2020) published by the Bureau of Labor Statistics and complied/distributed by the National Crosswalk Service Center.

  7. 7.

    Again, the values are calculated for the SOC-level occupations in O*NET-SOC, especially those used for the analyses in this study.

  8. 8.

    The version of the O*NET for each year is following: O*NET 5.0 (based on O*NET-SOC2000) for 2003, O*NET 6.0 (based on O*NET-SOC2000) for 2004, O*NET 8.0 (based on O*NET-SOC2000) for 2005, O*NET10.0 (based on O*NET-SOC2006) for 2006, O*NET12.0 (based on O*NET-SOC2006) for 2007, O*NET13.0 (based on O*NET-SOC2006) for 2008, O*NET14.0 (based on O*NET-SOC2009) for 2009, O*NET15.0 (based on O*NET-SOC2009) for 2010, O*NET16.0 (based on O*NET-SOC2010) for 2011, O*NET17.0 (based on O*NET-SOC2010) for 2012, O*NET18.0 (based on O*NET-SOC2010) for 2013, O*NET19.0 (based on O*NET-SOC2010) for 2014, and O*NET20.0 (based on O*NET-SOC2010) for 2015.

  9. 9.

    We estimate the same specification model with AES, \(AES_{\mathrm{skilled}}\), and \(AES_{\mathrm{unskilled}}\) constructed using all occupations. Skill level is defined using the “required level of education” in O*NET for each year. This implementation does not change the sign or significance of the estimated coefficients in terms of automation, especially in the specification model in Sect. 6.1.

  10. 10.

    For example, “State-to-State Migration Flows: 2007” reports migration flows from 2006 to 2007.

  11. 11.

    We use “Migration status, 1 year (whether the person had changed residence since a reference point a year ago),” “State or country of residence 1 year ago”, and “Person weight,” which indicates how many people in the US population are represented by a given person to calculate the migration flows between each states and non-migrants for each state.

  12. 12.

    The Occupational Employment Statistics is a semiannual survey that estimates the number of jobs for SOC occupations in each state.

  13. 13.

    The occupations with missing values in the degree of automation are excluded from K.

  14. 14.

    This study uses the weekly average wage.

  15. 15.

    The data sources of the control variables are as follows. Population: 2000–2010 Intercensal Estimates and State Population Totals Tables 2010–2016 (the Bureau of the Census). Wage: Quarterly Census of Employment and Wages (the Bureau of Labor Statistics). Unemployment rate and employment growth: Local Area Unemployment Statistics (the Bureau of Labor Statistics). House price index: House Price Index Datasets (Federal Housing Finance Agency). Share of high-skilled workers: Occupational Employment Statistics (the Bureau of Labor Statistics) and Occupational Projections and Training Data (the Bureau of Labor Statistics, the National Crosswalk Service Center). Employment share by industry: Gross Domestic Product by State (the Bureau of Economic Analysis).

  16. 16.

    We also estimate the same specification model with zero flows. The dependent variable is redefined by \(\log ((P_{ij}+1)/(P_{ii}+1))\). This implementation does not change the conclusions of the analyses.

  17. 17.

    Tables 5 and 6 do not show all the decimal places used in the calculation for ease of viewing.


  1. Accetturo A, Dalmazzo A, De Blasio G (2014) Skill polarization in local labor markets under share-altering technical change. J Reg Sci 54(2):249–272

    Article  Google Scholar 

  2. Acemoglu D, Restrepo P (2017) Robots and jobs: evidence from US labor markets, NBER Working Paper No. 23285

  3. Autor DH (2015) Why are there still so many jobs? The history and future of workplace automation. J Econ Perspect 29(3):3–30

    Article  Google Scholar 

  4. Autor DH, Dorn D (2013) The growth of low-skill service jobs and the polarization of the US labor market. Am Econ Rev 103(5):1553–1597

    Article  Google Scholar 

  5. Autor DH, Levy F, Murnane RJ (2003) The skill content of recent technological change: an empirical exploration. Quart J Econ 118(4):1279–1333

    Article  Google Scholar 

  6. Berger T, Frey CB (2016) Did the computer revolution shift the fortunes of U.S. cities? Technology shocks and the geography of new jobs. Reg Sci Urban Econ 57:38–45

    Article  Google Scholar 

  7. Davies PS, Greenwood MJ, Li H (2001) A conditional logit approach to U.S. state-to-state migration. J Reg Sci 41(2):337–360

    Article  Google Scholar 

  8. Executive Office of the President of the United States of America (2016) Artificial intelligence, automation, and the economy. Technical representative, Washington, DC

  9. Frey CB, Osborne MA (2017) The future of employment: how susceptible are jobs to computerisation? Technol Forecast Soc Change 114:254–280

    Article  Google Scholar 

  10. Fu Y, Gabriel SA (2012) Labor migration, human capital agglomeration and regional development in China. Reg Sci Urban Econ 42(3):473–484

    Article  Google Scholar 

  11. Gaspar J, Glaeser EL (1998) Technology and the future of cities. J Urban Econ 43:136–156

    Article  Google Scholar 

  12. Goos M, Manning A (2007) Lousy and Lovely Jobs: the rising polarization of work in Britain. Rev Econ Stat 89(1):118–133

    Article  Google Scholar 

  13. Greenwood MJ (1997) Internal migration in developed countries. Handb Popul Family Econ 1(PART B):647–720

    Article  Google Scholar 

  14. Greenwood MJ, Hunt GL (1989) Jobs versus amenities in the analysis of metropolitan migration. J Urban Econ 25(1):1–16

    Article  Google Scholar 

  15. Handel MJ (2016) The O*NET content model: strengths and limitations Stärken und Grenzen des O*NET-Models. J Labour Mark Res 49(2):157–176

    Article  Google Scholar 

  16. McFadden D (1973) Conditional logit analysis of qualitative choice behavior. In: Zarembka P (ed) Frontiers in econometrics. Academic Press, New York

    Google Scholar 

  17. Michaels G, Natraj A, Reenen JV (2014) Has ICT polarized skill demand? Evidence from eleven countries over twenty-five years. Rev Econ Stat 96(1):60–77

    Article  Google Scholar 

  18. Modestino AS, Dennett J (2013) Are American homeowners locked into their houses? The impact of housing market conditions on state-to-state migration. Reg Sci Urban Econ 43(2):322–337

    Article  Google Scholar 

  19. Molloy R, Smith CL, Wozniak A, Perspectives E, Smith L (2014) Internal migration in the United States. J Econ Perspect 25(3):173–196

    Article  Google Scholar 

  20. Pingle JF (2007) A note on measuring internal migration in the United States. Econ Lett 94(1):38–42

    Article  Google Scholar 

  21. Plantinga AJ, Détang-Dessendre C, Hunt GL, Piguet V (2013) Housing prices and inter-urban migration. Reg Sci Urban Econ 43(2):296–306

    Article  Google Scholar 

  22. Poncet S (2006) Provincial migration dynamics in China: borders, costs and economic motivations. Reg Sci Urban Econ 36(3):385–398

    Article  Google Scholar 

  23. Sasser AC (2010) Voting with their feet: relative economic conditions and state migration patterns. Reg Sci Urban Econ 40(2–3):122–135

    Article  Google Scholar 

  24. US Census Bureau (2018) Understanding and using American Community Survey data: What all data users need to know. Tech. Rep. July, Washington, DC

  25. Zabel JE (2012) Migration, housing market, and labor market responses to employment shocks. J Urban Econ 72(2–3):267–284

    Article  Google Scholar 

Download references


The author would like to thank Takatoshi Tabuchi, Marcus Berliant, Michael Pflüger, Yasuhiro Sato, and Atsushi Yamagishi. I am also indebted to an anonymous referee and the editor-in-chief, Martin Andersson, for their helpful comments and suggestions. The author also thanks the participants of the JEA meeting at Ritsumeikan University, the Asian Seminar in Regional Science at National Taiwan University, and of seminars at Tohoku University, Kyushu Sangyo University, Kyoto University, the University of Tokyo, and Kagawa University.

Author information



Corresponding author

Correspondence to Chigusa Okamoto.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Okamoto, C. The effect of automation levels on US interstate migration. Ann Reg Sci 63, 519–539 (2019).

Download citation

JEL Classification

  • J24
  • R23