Te c h n o l o g y, S k i l l s , a n d G l o b a l i za t i o n : - PowerPoint PPT Presentation

Te c h n o l o g y, S k i l l s , a n d G l o b a l i za t i o n : E x p l a i n i n g I n t e r n a t i o n a l D i f fe r e n c e s i n Ro u t i n e a n d N o n - Ro u t i n e Wo r k U s i n g S u r v e y D a t a P i o t r L e w a n d o w s k i ( I B S , I Z A ) A l b e r t P a r k ( I E M S H K U S T, I Z A ) W o j c i e c h H a r d y ( I B S ) D u Ya n g ( C A S S )

Motivation: the shift away from routine tasks and towards non-routine tasks is a secular change on developed countries ’ labor markets Wor orker Tas asks in n the the EU EU28, , 199 1998-2014 20 15 10 5 0 -5 -10 -15 -20 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Non-Routine Analytical Non-Routine Interpersonal Routine Cognitive Non-Routine Manual Routine Manual Source: Autor, Price (2013) Source: own calculations

Four key factors explain differences in tasks over time and across countries • Technological progress (computers, ICT, robots, etc.) Autor, Levy, Murnane 2003, Spitz-Oener 2006, Autor & Dorn 2013, Michaels et al. 2013 • Glo Globalization (FDI, trade, and global value chains) Oldenski, 2012, Goos et al. 2014, Reijnders & de Vries 2018 • Str Structural ch change (sectoral composition) Bárány & Siegel, 2018; Du & Park, 2017, Hardy et al. 2018 • Su Supply of of sk skills ills (worker human capital, demographics) Salvatori, 2015; Hardy et al., 2018, Montresor, 2018

Task contents are usually measured with O*NET, the US database on occupational demands (Autor et al. 2003, Acemoglu & Autor 2011) Non on-routine cog ognit itive Routine Routine Non on-routine (an analy lytical cog ognitive manual manual l / interpersonal) Abstract thinking, Pace determined by Repeating the same Operating vehicles, creativity, problem equipment, controlling tasks, being exact or mechanized devices, Tas ask items solving /Guiding, machines and processes, accurate, structured manual dexterity, directing, motivating, making repetitive work spatial orientation communicating motions Rela lationship ip Automation tough or b/ b/w hu human Complementary Easy to automate Easy to automate unprofitable tasks an and IC ICT Specialists (e.g Production workers, e.g. Drivers, miners, Oc Occupations Office clerks, sellers, designers, engineers, machine operators, construction workers, rich in ri in these administrative workers, IT developers), assemblers and waiters and waitresses, tasks cashiers technicians, managers locksmiths porters, cooks

Limitations in the global study of tasks • Data: most countries lack information on worker tasks • Focus on occupational structure assuming the US occupation-specific tasks • Data: tasks are measured at the level of occupation with O*NET, the US database • Tasks in the same occupation may differ depending on workers ’ skills, tenure, etc. • Coverage: most research focused on the US and Western Europe • Story may be different in the middle-income and developing countries

The contribution of this paper • We construct task content measures which: • Are measured at the worker level and country-specific • Are consistent with the Acemoglu & Autor (2011) measures based on O*NET • Data from worker surveys in 42 countries, including high, middle, and low-income • Previous studies using survey data examine only richer or poorer countries, and define tasks in an ad-hoc fashion (De la Rica & Gortazar 2016, Marcolin et al. 2016, Dicarlo 2016) • We examine the contributions of technology, globalization, structural change, and skill supply to task differences across countries

Preview of our findings • The task contents of occupations are different around the world • The routine intensity of tasks is higher in less developed countries, also within particular occupations. • Cross-country differences in tasks can be attributed to differences in: • Technology – in 25%, even more for high-skilled occupations; • Globalization – in 20%, even more for low-skilled and offshorable occupations; • Supply of skills – in 20%.

We use three surveys which include comparable data on the skill use at work, literacy and labor market status PIAAC • 32 countries surveyed between 2011 and 2015 (OECD) • sample sizes: from 4000 (Russia) to 26000 (Canada) • 9 countries surveyed between 2011 and 2015 STEP • sample sizes: from 2400 (Ukraine) to 4000 (Macedonia) urban residents (World Bank) • representative for the survey areas CULS • 6 cities (Guangzhou, Shanghai, Fuzhou, Shenyang, Xian, Wuhan) in 2016 (Chinese Academy • sample size 15500 of Social Science) • representative for the survey area

Representativeness of the data is limited in some countries. Bear that in mind when looking at the results STE TEP – urb rban su survey wit ith ad addit itional PIA IAAC limit lim itations in in so some countries • Belgium – Flanders • Bolivia – four main cities – La Paz, El Alto, Cochabamba and Santa Cruz de la • Russia – without Moscow municipal area Sierra (approx. 80% of urban population) • UK – England and Northern Ireland • Colombia – 13 main metropolitan areas • Indonesia – Jakarta • Georgia – no Abkhazia, South Ossetia • Singapore – only permanent residents • Lao PDR – both urban and rural, but we (approx. 75% of population) drop rural for consistency • China (CULS) – 6 cities

We construct our task measures on the US PIAAC and O*NET data Merge O*NET with the US PIAAC and calculate the Autor & Acemoglu (2011) task measures: non-routine cognitive analytical and personal, routine cognitive, manual Find combinations of PIAAC questions that approximate best the Autor & Acemoglu (2011) task measures across occupations in the US

We define task contents with these PIAAC / STEP items Non-routine Non Non on-routine Rou outi tine cog ognitiv ive Manual cogn ognit itive e analy lytic ical cogn ognit itive per ersonal Reading news Supervising Changing order of tasks – Physical tasks (at least once a month) reversed (not able) Presenting Reading professional titles (any frequence) Filling forms (at least once a month) (at least once a month) Task it item ems Solving problems Presenting – reversed (never) Programming (any frequence) Co Correla lation with ith O*N *NET 0.77 0.72 0.55 0.74 tasks

Example: the established Autor & Acemoglu (2011) measure contents calculated with O*NET data for the US 3 Non-routin ine cog ognitive an analytical 2 1 0 -1 -2 -3 111 122 133 142 212 215 222 226 233 241 251 262 265 313 321 325 333 341 351 412 422 441 513 516 523 532 612 622 713 723 741 752 811 814 817 831 834 912 932 951 962 Acemoglu & Autor (2011) measure ISC SCO occ occupatio ions

At the 3-digit occupation level in the US, the correlations between our measures and O*NET measures range from 0.55 to 0.77 3 Non-routin ine cog ognitive an analytical – cor orrelation 0.7 0.77 2 1 0 -1 -2 -3 111 122 133 142 212 215 222 226 233 241 251 262 265 313 321 325 333 341 351 412 422 441 513 516 523 532 612 622 713 723 741 752 811 814 817 831 834 912 932 951 962 Our measure Acemoglu & Autor (2011) measure ISC SCO occ occupatio ions

At the 3-digit occupation level in the US, the correlations between our measures and O*NET measures range from 0.55 to 0.77 3 Routine cog ognitive – correlation 0.5 0.55 2 1 0 -1 -2 -3 111 122 133 142 212 215 222 226 233 241 251 262 265 313 321 325 333 341 351 412 422 441 513 516 523 532 612 622 713 723 741 752 811 814 817 831 834 912 932 951 962 Our measure Acemoglu & Autor (2011) measure ISC SCO occ occupatio ions

We use the selected PIAAC / STEP questions to measure worker tasks in all 42 countries There is no unit of a task – we relate all countries to the US distribution: • 0 is the average level of a given task in the US • 1 is equivalent to the standard deviation of a given task in the US We also define routine task intensity (RTI) 𝑑𝑝𝑕 − ln 𝑜𝑠 𝑏𝑜𝑏𝑚𝑧𝑢𝑗𝑑𝑏𝑚 + 𝑜𝑠 𝑞𝑓𝑠𝑡𝑝𝑜𝑏𝑚 𝑆𝑈𝐽 = ln 𝑠 2 • RTI increases with the relative importance of routine tasks, • RTI decreases with the relative importance of non-routine tasks.

The more developed countries exhibit higher average values of non-routine tasks than the less developed countries Non-routine cognitive analytical Non-routine cognitive personal 0.2 0.2 R² = 0.59 FI NO SG SE SE NO DK NZ R² = 0.69 NZ 0.0 US 0.0 FI US JP GB DE GB SI KR AT DK Average value o f task Average value o f task NL CA CA NL SG AT EE BE EE CY CZ CN FR FR IL JP -0.2 -0.2 DE KR BE IE PL IE SK SK MK CL CL SI CO ES AM MK IL CY ES RU PL IT RU GE IT -0.4 -0.4 TR KE GR UA CO GR CZ LT BO BO KE AM CN LT GH LA -0.6 -0.6 ID TR GE GH -0.8 -0.8 0 20 40 60 80 100 0 20 40 60 80 100 GDP per capita ($k PPP) GDP per capita ($k PPP)