The Future of Work in Central America and the Dominican Republic
© 2024 International Bank for Reconstruction and Development / The World Bank
1818 H Street NW
Washington DC 20433
Telephone: 202-473-1000
Internet: www.worldbank.org
This work is a product of the staff of The World Bank with external contributions. The findings,
interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World
Bank, its Board of Executive Directors, or the governments they represent.
The World Bank does not guarantee the accuracy, completeness, or currency of the data included in this
work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or
liability with respect to the use of or failure to use the information, methods, processes, or conclusions set
forth. The boundaries, colors, denominations, and other information shown on any map in this work do
not imply any judgment on the part of The World Bank concerning the legal status of any territory or the
endorsement or acceptance of such boundaries.
Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the
privileges and immunities of The World Bank, all of which are specifically reserved.
Rights and Permissions
The material in this work is subject to copyright. Because The World Bank encourages dissemination of
its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as
full attribution to this work is given.
Please cite the work as follows: Moroz, Harry and Mariana Viollaz. 2024. The Future of Work in Central
America and the Dominican Republic. Washington, DC: World Bank.
Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank
Publications, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail:
pubrights@worldbank.org.
Cover design: Mauricio Díaz
Contents
Abbreviations and Acronyms VIII
Acknowledgements and Data Notes X
Executive Summary XI
The Limited Reach of Technology in Central America and the Dominican Republic XI
CADR’s Incomplete Transition to the Future of Work XIII
The Factors Behind CADR’s Incomplete Transition XV
The Importance of Technological Change Abroad XVII
Facilitating Technological Progress That Benefits Workers XVIII
Chapter 1 Jobs in Central America and the Dominican Republic 1
Setting the stage: Labor Market Dynamics and Technological Progress 1
Growth and Employment Dynamics 3
Economic Growth 3
Productivity 4
Employment 5
Labor Supply 7
Labor Force Participation and Employment 7
Job Quality 9
Skills 10
Migration 12
Labor Demand 13
Chapter 2 The Labor Market Impacts of Technological Progress 16
Setting the Stage: The History of the Future of Work 16
Technology’s Labor Market Impacts: Changes in What Workers Do and How They Do It 17
Technology’s Labor Market Impacts: The Evidence from Advanced Economies 20
Technology’s Labor Market Impacts: The View Outside of Advanced Economies 22
Contents III
Chapter 3 Technological Progress and Labor Market Transformation
in CADR Countries 24
Setting the Stage: Technological Progress and Its Employment Impacts 24
Changes in Tasks 24
Changes in Tasks: Computerization 24
Changes in Tasks: Artificial Intelligence and Mobile Robotics 35
Changes in Working Arrangements 38
Changes in Working Arrangements: Remote Work 38
Changes in Working Arrangements: Platform Work 42
Chapter 4 Barriers to and Enablers of the Future of Work in CADR
Countries 46
Setting the Stage: Looking Beyond Technological Potential 46
The Channels of Employment Change 47
Employment Structure 48
Supply of Skills 53
Adoption and Diffusion of Technology 57
Globalization 63
Chapter 5 Policy Recommendations 68
Setting the Stage: Facilitating Technological Progress from Which Workers Can Benefit 68
Promote the Adoption and Diffusion of Technology by Building Firm Capabilities 70
Strengthen Pathways for Skills Development and Deployment 75
Deploy Labor Market Insight Tools 76
Build Foundations-driven, Demand-oriented Education and Training Systems 77
Design Digitally Enabled, Fit-for-purpose Intermediation Programs 83
Adapt Social Protection and Labor Market Policies to New Forms of Work 85
References 87
Contents IV
Boxes
Box 2.1: Competitive Markets and Technology Adoption 17
Box 2.2: Spatial Differences in the Labor Market Impacts of Technological Progress 21
Box 4.1: The Availability of Technology Skills in Costa Rica and Panama 56
Box 5.1: Promoting Competition, Rightsizing Regulations, and Expanding Access to Finance to
Facilitate Technology Adoption 71
Box 5.2: Green Jobs in CADR 72
Box 5.3: The Labor Market Benefits of Early Childhood Education 78
Box 5.4: Incorporating Digital Skills Training into Technical and Vocational Education and Training 79
Box 5.5: The Complex Interaction of Technology, Labor Markets, and Gender 80
Box 5.6: Using SkillCraft to Connect Disadvantaged Young People to the Labor Market 84
Box 5.7: Public Gig Work Platforms 85
Figures
Figure ES1: Lags in Technology Adoption in CADR Countries XII
Figure ES2: Use of Computers and the Internet at Work, 2021 XII
Figure ES3: Classifying Tasks by Their Automatability XIII
Figure ES4: Routine and Nonroutine Task Intensity in CADR Versus the United States, 2021 XIII
Figure ES5: Share of Jobs at High Risk of Automation, Original and Adjusted Measures, 2021 XIV
Figure ES6: Changes in the Routine Intensity of Work, 2011–19 XV
Figure ES7: Online Gig Workers, 2022 XV
Figure ES8: Evolution of the Task Content of Migrants in the United States, 1970–2021 XVIII
Figure 1.1: GDP Per Capita, 2002–21 3
Figure 1.2: GDP Growth, 2002–19 3
Figure 1.3: Labor Productivity Across Sectors, 2019 4
Figure 1.4: Labor Productivity Growth Across Sectors, 2002–19 5
Figure 1.5: Employment Growth, 2002–19 5
Figure 1.6: Change in Employment Share by Sector, 1960–2010 6
Figure 1.7: Employment Share by Sector, 1991–2019 6
Figure 1.8: Labor Force Participation Rate, 2021 7
Figure 1.9: Female Labor Force Participation Rate, 2021 7
Figure 1.10: Female-Male Gap in Labor Force Participation Rate, 2021 8
Figure 1.11: Unemployment Rate, 2002–21 8
Figure 1.12: Female-Male Gap in Unemployment Rate, 2002–21 9
Figure 1.13: Young People in NEET Status, 2021 9
Figure 1.14: Employment Share by Wage and Self-Employment, 2019 9
Figure 1.15: Employment Share by Wage Employment, 2002–19 9
Figure 1.16: Informality Rate, 2021 10
Figure 1.17: Schooling and Learning-Adjusted Schooling, 2019 11
Figure 1.18: Learning Poverty, 2019 11
Figure 1.19: Employment Rate by Education, 2021 11
Contents V
Figure 1.20: Informality Rate by Education, 2021 11
Figure 1.21: Employment Share by Skill Level, 2021 12
Figure 1.22: Change in Employment Share by Skill Level, 2011-21 12
Figure 1.23: Net Migration Rate, 2021 13
Figure 1.24: Migrants from CADR countries, 1990–2020 13
Figure 1.25: Share of CADR Outmigrants in the United States, 2020 13
Figure 1.26: Remittance Inflows, 2022 13
Figure 1.27: New Business Density, 2020 14
Figure 1.28: Firms Choosing Inadequate Education as Biggest Obstacle, 2016 14
Figure 2.1: Classifying Tasks by Their Automatability 18
Figure 2.2: The Labor Market Impacts of Technological Change in CADR 23
Figure 3.1: Change in Employment Share by Skill Level, circa 1980s to circa 2000s 25
Figure 3.2: Changes in the Routine Intensity of Work, 2011–19 26
Figure 3.3: RTI and Task Intensity of CADR Countries, 2021 27
Figure 3.4: RTI by Economic Sector, 2021 27
Figure 3.4: RTI by Economic Sector, 2021 (continued) 28
Figure 3.5: RTI by Sociodemographic Characteristics, 2021 29
Figure 3.6: Returns to Routine and Nonroutine Tasks, 2010s 29
Figure 3.7: Three-Year migrants in the United States, 1970–2021 30
Figure 3.8: Three-Year CADR Migrants in the United States, 1970–2021 30
Figure 3.9: Employment of Three-Year Migrants in Manufacturing Jobs in the United States, 1970–2021 31
Figure 3.10: Employment of Three-Year Migrants in Construction Jobs in the United States, 1970–2021 31
Figure 3.11: Change in Employment Share by Skill Level of Three-Year Migrants and United States
Nonmigrants in the United States, 1980–2021 32
Figure 3.12: Evolution of the Task Content of Three-Year Migrants in the United States, 1970–2021 33
Figure 3.13: Evolution of the Task Content of Three-Year CADR Migrants by Gender in the United
States, 1970–2021 34
Figure 3.14: Average Hourly Wage of Three-Year Migrants in the United States, 1980–2021 35
Figure 3.15: Susceptibility of the Workforce to Automation, 2021 35
Figure 3.16: Probability of Automation by Sociodemographic Characteristics, 2021 36
Figure 3.17: Share of Jobs at High Risk of Automation, Original and Adjusted Measures, 2021 37
Figure 3.18: Average Probability of Automation, 2021 38
Figure 3.19: Share of Workers with at Least 10 and 50 Percent of Tasks Exposed to GPTs, 2021 38
Figure 3.20: Share of Workers in Jobs with High Amenability to Working from Home, 2021 39
Figure 3.21: Share of Wage Workers Working from Home Prior to the Pandemic, 2019 40
Figure 3.22: Share of Self-Employed Working from Home Prior to the Pandemic, 2019 40
Figure 3.23: Share of Workers Working from Home During the Pandemic, 2021 41
Figure 3.24: Share of Workers Working from Home During the Pandemic, 2020 41
Figure 3.25: Global Demand for Online Gig Work, 2016–23 43
Figure 3.26: Online Gig Workers, 2022 44
Figure 3.27: Characteristics of Workana Freelancers, 2022 45
Figure 4.1: Factors Contributing to Cross-Country Variation in Routine Task Intensity, 2010–2021 48
Figure 4.2: Deindustrialization in CADR and High-Income Countries, 1991–2021 50
Contents VI
Figure 4.3: Change in the Industrial Share of Employment, 1991–2019 50
Figure 4.4: Digitally Deliverable Services Exports, 2005–21 51
Figure 4.5: Use of Digital Platforms, 2020 52
Figure 4.6: SME Engagement with Digital Tools, 2020 and 2022 53
Figure 4.7: Education of the Employed Population in CADR Countries, 2000–21 54
Figure 4.8: Share of Students Meeting Minimum Proficiency in Reading and Math, 2022 55
Figure B4.1.1: Penetration of Technology and Disruptive Technology Skills, 2022 56
Figure 4.9: The Digital Adoption Index, 2016 57
Figure 4.10: The Frontier Technology Readiness Index: ICT, 2022 57
Figure 4.11: The Contribution of ICT and Non-ICT Assets to Economic Growth, 2000s–10s 58
Figure 4.12: Lags in Technology Adoption in CADR Countries 59
Figure 4.13: Diffusion of the Internet and Cell Phones, 2021 59
Figure 4.14: Use of Computers and the Internet at Work, 2021 60
Figure 4.15: Use of Digital Platforms, 2021 60
Figure 4.16: Share of Jobs at High Risk of Automation by Sector, 2021 61
Figure 4.17: Imports of ICT Goods and ICT Services as Share of Total Trade, 2021 62
Figure 4.18: Prices of ICT Service Baskets, 2022 63
Figure 4.19: Net Inflows of Foreign Direct Investment, 1970–2022 64
Figure 4.20: Share of Output in GVCs by Sector, 2021 64
Figure 4.21: Relationship Between Routine Task Intensity Index and GVCs, 2010–2021 65
Figure B5.2.1: Share of Green Jobs in CADR Countries Overall and by Gender, 2021 73
Figure B5.2.2: Share of green jobs in CADR countries by sector, 2021 73
Figure B5.2.3: Green Talent in Costa Rica and Comparator Countries, 2023 74
Figure B5.5.1: Use of Computers and the Internet at Work by Gender, 2021 81
Figure B5.5.2: Female Enrolment in Tertiary Programs and in ICT- and Engineering-Related
Tertiary Degree Programs, 2021 82
Tables
Table ES1: Policies to Facilitate Technological Change and Mitigate the Negative Effects of
Resulting Disruptions XX
Table 1.1: Summary of the Main Labor Market Supply and Demand Issues in CADR 2
Table 1.2: The Biggest Obstacles Facing Firms 14
Table 3.1: Advantages and Disadvantages of Platform Work for Workers and Firms 42
Table 3.2: Primary Location-Based Platforms in CADR Countries 43
Table 3.3: Most Common Type of Online Gig Tasks in CADR Countries, 2017–23 44
Table 4.1: Examples of Digital Interventions Targeted to Smallholder Farmers in CADR 49
Table 4.1: Imports of Industrial Robots, 2021 60
Table 5.1: Policies to Facilitate Technological Change and Mitigate the Negative Effects of
Resulting Disruptions 69
Table 5.2: Drivers of Technology Adoption Among Firms 70
Table 5.3: Main Challenges Cited by SMEs When Using or Trying to Adopt Digital Platforms, 2022 75
Table 5.4: Tools for Identifying In-Demand Skills 77
Contents VII
Abbreviations and Acronyms
ACS American Community Survey
AI artificial intelligence
ARG Argentina
BOL Bolivia
BRA Brazil
CADR Central America and the Dominican Republic
CHL Chile
COL Colombia
CRI Costa Rica
DAI Digital Adoption Index
DOM Dominican Republic
ECE early childhood education
ECLAC Economic Commission for Latin America and the Caribbean
ECU Ecuador
FDI foreign direct investment
GDP gross domestic product
GNI gross national income
GPT generative pretrained transformers
GTI green task intensity
GSP global skills partnership
GTM Guatemala
GVC global value chains
HND Honduras
ICT information and communication technology
IFR International Federation of Robotics
ILO International Labour Organization
IPUMS Integrated Public Use Microdata Series, International
ISCO International Standard Classification of Occupations
KOR Republic of Korea
LAC Latin America and the Caribbean
LFP labor force participation
MEX Mexico
NEET not in education, employment, or training
Abbreviations and Acronyms VIII
NIC Nicaragua
OECD Organisation for Economic Co-operation and Development
OLI Online Labor Index
O*NET Occupational Information Network
PAN Panama
PERU Peru
PIAAC Program for the International Assessment of Adult Competencies
RTI routine task intensity
SCD Systematic Country Diagnostics
SEDLAC Socioeconomic Database for Latin America and the Caribbean
SLV El Salvador
SME small and medium-sized enterprise
SOC Standard Occupational Classification
TVET technical and vocational education and training
UE unemployment
UNCTAD United Nations Conference on Trade and Development
UNESCO United Nations Educational, Scientific and Cultural Organization
URY Uruguay
USA United States of America
USAID United States Agency for International Development
VEN Venezuela
WDI World Development Indicators
WFH work from home
WITS World Integrated Trade Solution
YES Youth Empowerment Service
Abbreviations and Acronyms IX
Acknowledgements
and Data Notes
This report was prepared by Harry Moroz and Mariana Viollaz. Diana Isabel Londoño Aguirre, Guillermo
Beylis, Guillermo Caballero Ferreira, Lily Franchini, Luis Laguinge, Maria Del Mar Gomez Ortiz, Marla
Hillary Spivack, and Daria Taglioni provided substantial contributions. The team received excellent
comments and advice from Marina Bassi, Diego Arias Cabballo, Lourdes Rodriguez Chamussy, Wendy
Cunningham, Janibeth Miranda, Miriam Montenegro, Alvaro Gonzalez De Pablo, Truman Packard, Viviana
Maria Eugenia Perego, Josefina Posadas, Jaime Saavedra, Alexandria Valerio, Deborah Elisabeth Winkler,
and William Wiseman.
Census, American Community Survey (ACS), and Current Population Survey data were obtained
from Integrated Public Use Microdata Series, International (IPUMS-International). The authors wish to
acknowledge the statistical offices that provided the underlying data. Household and labor force surveys
were obtained via the Socioeconomic Database for Latin America and the Caribbean (SEDLAC), as well
as from the statistical agencies of individual countries, as follows:
• Costa Rica: Encuesta Continua de Empleo
• Dominican Republic: Encuesta Nacional de Fuerza de Trabajo and Encuesta Nacional Continua de
Fuerza de Trabajo
• El Salvador: Encuesta de Hogares de Propósitos Múltiples
• Guatemala: Encuesta Nacional de Empleo e Ingresos
• Honduras: Encuesta Permanente de Hogares de Propósitos Múltiples
• Nicaragua: Encuesta Nacional de Hogares sobre Medición de Nivel de Vida
• Panama: Encuesta de Hogares and Encuesta de Mercado Laboral
The team utilized data on green tasks prepared by Julia Granata and Josefina Posadas as described
in Granata and Posadas (2022). Data on global value chains (GVCs) were provided by Daria Taglioni as
described in Borin, Mancini, and Taglioni (2021). Data on platform workers were provided by Namita Datta
as described in Datta and Chen (2023). The data on LinkedIn user profiles was provided by LinkedIn via
the Development Data Partnership.
The report uses comparisons to other countries in Central America and the Dominican Republic (CADR),
to Latin America and the Caribbean (LAC) and OECD regional averages, and to Korea to benchmark the
characteristics of CADR countries. See appendix A for more information on the selection of benchmarks.
Acknowledgements and Data Notes X
Executive Summary
Technological progress has the potential to cause significant disruption in labor markets. Advances
first in agricultural and industrial machinery, then computers, and now artificial intelligence (AI) have
enabled machines to undertake a growing range of tasks previously done by humans, which puts jobs
at risk. However, far from being a purely destructive force, technological progress and automation can
also generate employment. Where competition is robust, automation can create jobs as price declines
linked to productivity improvements create higher demand. New technologies can also generate entirely
new tasks, new types of jobs, and new industries. Information and communication technology (ICT) can
facilitate new working arrangements that are beneficial for workers and firms. This means that slow
progress towards the future of work could mean missing out on the opportunities created by technological
advancements.
Advanced economies offer a model for how the future of work will look in Central America and the
Dominican Republic (CADR), but important differences in development stages mean that the labor
market impacts of technological progress are distinct now and are likely to continue to be in the
near future. Labor markets in CADR are better characterized by agricultural employment than by robot-
assisted manufacturing. This means that applying lessons from technology’s impacts on the labor markets
of advanced countries requires an assessment of both the technological potential to automate jobs in
CADR and the economic potential for new technologies to take hold and cause labor market disruptions.
The adoption and diffusion of technology as well as employment structure, skills supply, and globalization
will all shape the impact of technological progress on labor markets in the region. Technology adoption
abroad is also likely to affect CADR through changes in demand for labor that is embodied in the
movement of goods (offshoring), the movement of people (migration), and the movement of services
(digital trade in services).
This report examines the impact of computers, robots, AI, and improved ICT at work on labor markets
in CADR. The report focuses on these technologies as the most likely to have shaped labor markets
in the region in the recent past and the most likely to shape them in the near future. The report first
examines how technological progress within the region is shaping what workers do (their tasks) and
how they do it (their working arrangements). The report goes beyond the analysis of susceptibility to
automation to dissect the factors underlying recent labor market transformations and undercover the
extent to which technological change has played a role in these transformations. The report also examines
how technological progress outside of the region is shaping labor markets within it by investigating how
robot adoption in the United States is affecting the demand for CADR workers in CADR countries and for
CADR workers in the United States.
THE LIMITED REACH OF TECHNOLOGY IN CENTRAL AMERICA AND THE DOMINICAN
REPUBLIC
The penetration and diffusion of technology are limited in CADR countries . CADR countries rank
low on summary measures of technological progress like the World Bank’s Digital Adoption Index and
the Frontier Technologies Readiness Index compiled by the United Nations Conference on Trade and
Development. CADR countries are slower to adopt new technologies, though these lags have shortened
with each significant technological advancement (figure ES1). Even when technology is available, however,
diffusion is limited. For example, cellphones are prevalent in CADR countries but use of the internet varies
significantly across countries in the region. The Dominican Republic and Costa Rica have rates of internet
use that approach those of Korea and the United States. In all other CADR countries, in contrast, at least
one-third of the population did not use the Internet in 2021.
Executive Summary XI
FIGURE ES1: Lags in Technology Adoption in CADR Countries
Years
HND
NIC
GTM
Telephone
SLV
DOM
CRI
PAN
HND
GTM
Computer
PAN
NIC
SLV
CRI
NIC
HND
GTM
Internet
SLV
PAN
DOM
CRI
0 10 20 30 40 50 60 70 80
Source: Comin, Hobijn, and Rovito 2008.
Note: Lags are calculated as the years between a benchmark year and the year in which the United States had the same adoption as the CADR country
in the benchmark year. Data are not available for computers for the Dominican Republic. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Limited technology uptake is apparent among CADR workers and firms. In the agricultural sector, farm
machinery measured in total horsepower per hectare is around 10 percent of the level in the United
States in Guatemala, Honduras, and Nicaragua and 50 percent and 66 percent in Costa Rica and Panama,
respectively. Less than 15 percent of workers in all CADR countries work in jobs that are intensive in the
use of computers or the internet, compared with at least 25 percent in the United States (figure ES2).
Data on firms’ use of robots are not available for CADR countries, but trade data on imports of industrial
robots provide suggestive evidence that use of robotic technologies is not widespread. Costa Rica and
the Dominican Republic led CADR with imports of the equivalent of six robots and one robot for every
100,000 people in 2021, respectively. All other CADR count ries imported less than half a robot for every
100,000 people. The United States and Korea, in contrast, imported 13 and 21.
FIGURE ES2: Use of Computers and the Internet at Work, 2021
Percentage of workers in computer- and internet-intensive occupations
a. Computer use b. Internet use
30% 30%
25% 25%
20% 20%
15% 15%
10% 10%
5% 5%
0% 0%
USA PAN CRI DOM SLV NIC HND GTM USA PAN CRI DOM NIC SLV HND GTM
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The year is 2019 for Guatemala and Honduras; 2018 for Panama; and 2014 for Nicaragua. Computer- and internet-intensive occupations are defined
as occupations in the top 25 percent of computer and internet use at work as defined using PIAAC data from comparator countries. CRI = Costa Rica;
DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; USA = United States; SLV = El Salvador.
Executive Summary XII
CADR’S INCOMPLETE TRANSITION TO THE FUTURE OF WORK
Analyzing the intensity of employment in tasks that are more and less susceptible to automation
provides evidence of the impact of technology on labor markets . Certain types of tasks—routine
ones—are more susceptible to automation by computers and so their presence is associated with less
technological impact on the labor market (figure ES3). These routine tasks might be either manual repetitive
hands-on work like packaging and assembly or cognitive administrative work like data entry. Nonroutine
tasks, in contrast, are less susceptible to automation by computers and so their presence is associated
with a greater impact of technology on the labor market. These nonroutine tasks might be either manual
flexible hands-on work like groundskeeping and maintenance or cognitive, involving either knowledge
work like strategic planning or risk assessment, or people-oriented work like client relations or counseling.
FIGURE ES3: Classifying Tasks by Their Automatability
Higher-skilled
Manual Cognitive
Repetive hands-on work Administrative work
Routine Packaging Data entry
Assembly Call center operations
Less automatable
Loading and unloading Basic accounting
Flexible hands-on work Knowledge work
Carpentry Risk assessment, strategic planning, forecasting
Non-routine
Groundskeeping People-oriented work
Repair and maintenance Client relations, counseling, conflict resolution
Consistent with technology’s limited adoption in CADR, there is little evidence that computerization has
had substantial labor market impacts on most CADR countries. Workers in CADR do jobs that involve much
less of the knowledge and people-oriented tasks associated with technological progress than workers in the
United States (figure ES4). Though data do not allow for a distinction between routine and nonroutine manual
tasks, work in less-skilled hands-on jobs is much more common in CADR. Male, young, less-educated, and
rural workers are even more likely to be employed in these more routine, more automatable jobs of the past.
FIGURE ES4: Routine and Nonroutine Task Intensity in CADR Versus the United States, 2021
Standard deviations of tasks intensity from the United States’ average
0.30
0.20
0.10
0.00
-0.10
-0.20
-0.30
-0.40
-0.50
-0.60
-0.70
-0.80 GTM NIC HND SLV CRI DOM PAN
Knowledge work People-oriented work Administrative work Hands-on work
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The year is 2019 for Guatemala and Honduras; 2018 for Panama; and 2014 for Nicaragua. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Executive Summary XIII
Impacts from artificial intelligence and robots are also likely to be muted in the short term. Measures
of the potential for artificial intelligence and mobile robots1 to replace jobs in CADR that extend this
task analysis seem to show that large numbers of workers are at risk: at least 60 percent of workers
in all CADR countries work in jobs where at least some tasks they undertake are likely to disappear or
change substantially (figure ES5). However, this potential for automation appears much less dire when
small corrections are made for factors that might interrupt the translation of potential labor market impacts
of technological progress into actual impacts. When sectors that tend to have low rates of technology
use are excluded from the analysis, exposure to displacement by automation technologies declines
substantially to at most 46 percent of workers.
FIGURE ES5: Share of Jobs at High Risk of Automation, Original and Adjusted Measures, 2021
Percentage
100%
80%
60%
40%
20%
0%
Original Adjusted Original Adjusted Original Adjusted Original Adjusted Original Adjusted Original Adjusted Original Adjusted
CRI DOM GTM HND NIC PAN SLV
Low Medium High
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017; Frey and Osborne 2017.
Note: The year is 2019 for Guatemala and Honduras; 2018 for Panama; and 2014 for Nicaragua. The risk of automation is calculated according to the task-
based approach in Arntz, Gregory, and Zierahn (2016) and Egana del Sol et al. (2022) and then adjusted following Weller et al (2019). Low risk is a probability of
automation that is 30 percent or less; medium is above 30 but below 70 percent; and high is 70 percent or above. The adjustment assigns a risk of automation
equal to zero to workers in low-productivity sectors defined as self-employed workers with less than college education, wage employees and employers
in small-size firms, domestic workers, and workers who do not receive a labor income. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Recently, however, there is some evidence in CADR of a shift towards the knowledge- and people-
oriented jobs that characterize technological progress. In the last decade, the routine work that
characterizes jobs of the past declined in most of the region while the nonroutine knowledge and people-
oriented work that characterizes the employment of the future increased (figure ES6). Specifically, jobs
became more intensive in knowledge tasks in all CADR countries except Costa Rica and Honduras and
in people-oriented tasks in all except Costa Rica, Guatemala, and Honduras. Even with this progress,
however, jobs in CADR countries were much more intensive in the routine tasks of the past in 2021 than
those in the United States.
1 Mobile robots use sensors, artificial intelligence, and other technology to maneuver in order to perform nonroutine manual tasks (Frey and
Osborne 2017).
Executive Summary XIV
FIGURE ES6: Changes in the Routine FIGURE ES7: Online Gig Workers, 2022
Intensity of Work, 2011–19
Difference in index of routine task intensity (RTI) Percentage of employed population that participates in
online gig work activities
0.04 35%
0.02 30%
0
25%
-0.02
20%
-0.04
15%
-0.06
10%
-0.08
-0.10 5%
-0.12 0%
HND CRI PAN DOM SLV GTM CRI GTM PAN DOM SLV NIC HND
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017. Source: Datta and Chen 2023.
Note: The end year is 2018 for Panama. Nicaragua is excluded because Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
of lack of available data. CRI = Costa Rica; DOM = Dominican Republic; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
GTM = Guatemala; HND = Honduras; PAN = Panama; SLV = El Salvador.
There is also evidence that advancements in ICT are making new forms of work a normal part of
economic life in CADR, though their prevalence is varied. During the COVID-19 pandemic, remote and
platform work surged. In Costa Rica, for example, remote work increased from 8 percent of employment
in 2019 to 18 percent in 2021. Gig work is also increasingly becoming an option for workers in the region.
In 2022, more than 10 percent of workers reported doing online gig work in Costa Rica, the Dominican
Republic, Guatemala, and Panama (figure ES7). After the pandemic, however, remote work rates declined.
In the Dominican Republic, for example, telework fell from 5 percent of workers in 2020 to 1.7 percent
in 2022. Despite the high rates of online gig work in some countries, in El Salvador, Honduras, and
Nicaragua, only 6 percent or less of workers took on online gig jobs. The gig workers providing taxi and
delivery services in the Dominican Republic make up only 0.2 percent of the labor force. There is also
evidence that at least at present, platform jobs are accessible primarily to those who face fewer obstacles
in the labor market: platform workers tend to be younger, male, and more highly educated.
THE FACTORS BEHIND CADR’S INCOMPLETE TRANSITION
The transition of CADR countries to the future of work has been only partial, despite the existence
of the technology to automate many tasks and to enable remote and platform work. Employment is
evolving towards nonroutine tasks in CADR, but routine ones still dominate. Remote work surged during
the pandemic, but now seems to be diminishing. Platform work is present but makes up a small share of
total jobs in some countries. This highlights the importance of analyzing factors beyond the “technological
potential” for a job to be automated or performed remotely to understand which types of workers are
hired and in which types of capital investments are made.
Employment structure (how employment is distributed across sector), skills, adoption and diffusion
of technology, and globalization mediate the impact of technology on jobs. Factors that are key to
understanding changes in what work is being done and how work is being done include the adoption and
diffusion of technology (Do firms and workers use new technologies?); but also employment structure (Do
sectors that use technology dominate?); the supply of skills (Do workers have skills that are complementary
to technology?); and globalization (Is the economy open to technological influences?).
Executive Summary XV
Each of these factors plays a role in explaining the evolution of employment in CADR countries .
Examining the relationship between these factors and how intensive a job is in the routine tasks that are
associated with jobs of the past shows that each is important in explaining what workers do at work in
CADR countries. Across CADR:
• Less agricultural work is associated with employment that is less intensive in routine tasks.
• More education is associated with employment that is less intensive in routine tasks.
• Greater technology use is associated with employment that is less intensive in routine tasks.
• Greater global value chain participation is associated with employment that is more intensive in
routine tasks.
Adoption and diffusion of technology is not a primary factor explaining differences in the tasks that
workers do in CADR countries. Technology explains around 10 percent of the variation in the routine
intensity of employment over time in all CADR countries. This result is consistent with the limited reach
of technology in CADR. Instead, worker skills and employment structure are the most important factors.
Worker skills have improved substantially in CADR, but relatively low levels of tertiary education
and low educational quality impede uptake of newer technologies. Education levels have improved
throughout the CADR region, driven by a decline in workers who have primary education or less and an
increase in those who have secondary education. However, education levels in the higher-income CADR
countries are low relative to the most developed economies globally and those in the lower-income CADR
countries are low relative to the higher-income CADR countries. In Costa Rica, the Dominican Republic,
and Panama, around a quarter of workers are tertiary educated versus more than half of workers in
Korea and the United States. In the remaining CADR countries, 14 percent or less of workers have tertiary
education. Educational quality is also a problem. Even in the best performing country in the region, Costa
Rica, less than 60 percent of 15-year-olds meet the minimum proficiency standard on the Programme for
International Student Assessment (PISA) in reading and just 40 percent in math. In the poorest performing
CADR countries, less than 30 percent and 20 percent do, respectively. This compares with at least three-
quarters of students in both subjects in the OECD.
Workers in CADR countries also lag in the kinds of skills needed to accomplish the new tasks
associated with new technologies. In the Dominican Republic, firms cite inadequate digital skills as
a limitation when trying to fill vacancies. In El Salvador, digital startups report low quality training and
education as barriers to hiring talent. Data on technology skills taken from LinkedIn profiles in Costa Rica
and Panama show a significant lag in technology skills. Lack of management skills, which are increasingly
viewed as a prerequisite to technology adoption, may also be hindering adoption of technologies in
CADR. For example, based on data from the World Management Survey, the average management score
for firms in Nicaragua is 2.4, the seventh lowest among the 35 countries surveyed.
The prevalence of services sector employment across CADR countries implies less vulnerability
to automation, but at the expense of the dominance of low-productivity employment . Most CADR
countries are experiencing “premature deindustrialization” in which growth in the services share of
employment is happening at lower levels of development and at lower levels of peak manufacturing
employment than occurred in advanced countries. Indeed, the industrial share of employment contracted
between 1991 and 2019 in all CADR countries except Panama and Honduras. This implies a lower risk of
automation—low-paid services are harder to automate because they tend to involve more (nonroutine)
hands-on actions and more (nonroutine) people-oriented interactions—but raises concerns about future
growth because manufacturing has historically been an accelerator of economic development and a
generator of good jobs.
Executive Summary XVI
Continued advancements in ICT outside of the region could offer a way forward to improve the
productivity of the increasingly dominant services sector. Though globalization has not played
a significant role in the evolution of employment in CADR, connections to the global economy have
increased in recent decades. Foreign direct investment (FDI) inflows grew from a regionwide average of 1.0
percent of GDP between 1970 and 1989 to 4.5 percent between 2010 and 2022. This is, in part, the result
of the entry of CADR countries in manufacturing GVCs, particularly in textile and garments (Dominican
Republic, El Salvador, Guatemala, and Nicaragua), automotive (El Salvador, Honduras, and Nicaragua),
and medical devices (Costa Rica and the Dominican Republic). Digital technologies are opening new
opportunities for small- and medium-sized enterprises (SMEs) in CADR to access new markets and expand
their customer and supplier base, including to international markets. Meta and the World Bank’s Future
of Business Survey of SMEs shows that SMEs across the region use digital tools to facilitate online sales
and purchases: more than 40 percent say they use such technologies. Digital platforms in particular seem
capable of transformative impacts. Large shares of firms across CADR report that digital platforms had a
very or extremely important impact on their business.
THE IMPORTANCE OF TECHNOLOGICAL CHANGE ABROAD
Changes in the nature of work in CADR countries are being influenced not only be developments within
the region but also by how technological progress plays out in developed countries. Technological
progress abroad could affect CADR labor markets in several ways. First, automation abroad could drive a
process of reshoring in which automation-induced labor savings incentivize companies to bring jobs back
for domestic production (or to create new jobs domestically instead of abroad). Second, technological
developments in developed countries could alter the demand for migrants from CADR countries.
Technological advancements could replace the need for migrant workers, or they could increase and
change the kinds of migrant workers demanded, as suggested by recent research demonstrating a link
between technological progress and increased demand for manual service workers. Research conducted
for this report shows the following impacts of technological progress outside of CADR.
• Robot adoption in the United States has a negative effect on labor markets in most CADR countries,
likely as a result of reduced opportunities for offshoring. Between 2010 and 2019, robot adoption
in the United States is associated with a decline in labor force participation and employment of 0.4
percentage points for workers with medium education in the Dominican Republic and El Salvador.
In Costa Rica, the effects of robot adoption are channeled through an increase of 0.2 percentage
points in the unemployment rate of low- and medium-educated workers. Honduras is an exception.
Robot adoption led to an increase in the labor force participation of medium-educated workers of 0.2
percentage points. This may be linked to the reliance of Honduras’s export basket to the United States
on raw materials, which may mean that the country benefits when robot adoption leads to expansions
in demand.
• Computerization in the United States has shifted CADR migrants into less-skilled services and
construction jobs. Since 1970, the employment of migrants in the United States from non-CADR,
middle-income countries has become much more intensive in knowledge and people-oriented tasks
(figure ES8a). The employment of migrants from CADR countries, in contrast, has become somewhat
more intensive in flexible hands-on work and people-oriented work, but has not experienced much
change in knowledge work (figure ES8b). In sum, technological progress in the United States seems
to be pushing CADR workers towards less-skilled services and construction sector jobs where
nonroutine manual and interpersonal tasks dominate.
Executive Summary XVII
FIGURE ES8: Evolution of the Task Content of Migrants in the United States, 1970–2021
Task index (1970 = 0)
a. Migrants from other middle-income countries b. CADR migrants
0.60 0.40
0.40 0.20
0.20 0.00
0.00 -0.20
-0.20 -0.40
-0.40 -0.60
-0.60 -0.80
-0.80 -1.00
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
Administrative work People-oriented work Administrative work
Repetitive hands-on work Flexible hands-on work
Source: ACS 2000-2021; US Census 1970, 1980, 1990.
Note: CADR = Central America and the Dominican Republic.
• Robot adoption in the United States had no observable impact on overall migration flows from CADR
countries to the United States in the 2000s and 2010s. Overall, migrant flows did not seem to respond
to robot adoption in the United States during this period. However, robot adoption in the early 2000s
did increase the demand for high-skilled CADR migrants but in low-skilled nonroutine occupations.
Between 2000 and 2010, an increase of one robot per thousand industrial workers was associated
with a higher employment rate and a lower unemployment rate for high-educated CADR migrants.
This seems to be related to an increase in aggregate demand, particularly in food preparation and
building and grounds cleaning and maintenance, due to increased productivity associated with robot
adoption, as high-skilled CADR workers switched to these lower-skilled occupations. Robot adoption
in the United States between 2010 and 2019, however, led to less demand for high-educated CADR
migrants. The employment rate of high-educated CADR migrants declined 0.4 percentage points for
each additional robot, while the total number of high-educated CADR migrants fell by 53 migrants for
each additional robot adopted. This may reflect reduced incentives for high-skilled CADR migrants
to migrate.
FACILITATING TECHNOLOGICAL PROGRESS THAT BENEFITS WORKERS
The challenge for policymakers in CADR is to balance policies that enable technological progress and
the benefits it brings with policies that mitigate the effects of any ensuing disruptions. Technological
progress is a key component of development, allowing for increased productivity, new products and
services, and access to new markets. In many cases technological progress goes hand-in-hand with
better employment outcomes. CADR countries have not yet experienced significant disruptions from
technological progress, nor have they taken full advantage of its potential benefits. While many workers in
the region seem to be at risk of losing their jobs to machines, a number of factors related to employment
structure, skills, the use of technology, and globalization mean that this risk has not materialized and is
unlikely to do so in the near term. However, this also means that the region is not yet benefiting from the
potential gains associated with technological progress.
Executive Summary XVIII
Given the limited reach of technology in CADR countries, supporting technology adoption and
diffusion will be key to facilitating CADR countries’ continued development. Business advisory
and related services can promote technology uptake, improve the management skills that are often a
prerequisite of technology adoption, and improve linkages between small businesses and the digital
platforms that can open new markets.
From a labor market perspective, two priorities emerge for CADR countries to take advantage of the
benefits of technological progress while mitigating the downsides. First, pathways for developing skills
complementary to new technologies will be essential. Second, social protection and labor market policies
will need to be adapted to new working arrangements on one hand, and to the disruptions associated
with technological progress on the other. Within these adaptations lies a short-term opportunity to expand
access to social protection by developing partnerships among digital platforms, governments, and service
providers.
This will need to be done taking into account the unique circumstances in CADR countries. The region’s
more developed countries—Costa Rica, Panama, and, in many respects, the Dominican Republic—are
generally at a more advanced stage of structural change and technology adoption. This means that
different CADR countries have somewhat different priorities (table ES1). The region’s less advanced
countries need to focus more on building foundational structures and piloting new initiatives for skills
building and social protection, while the region’s more advanced countries can work to improve the
sophistication of existing systems.
Executive Summary XIX
TABLE ES1: Policies to Facilitate Technological Change and Mitigate the Negative Effects of
Resulting Disruptions
Objective 1: Promote the adoption and diffusion of technology by building firm capabilities
Building strong • Ensure quality infrastructure (e.g., electricity, internet service, mobile networks)
foundations • Promote competition, ensure regulations enable technology adoption and
diffusion, and expand access to finance
All CADR countries can work to CADR’s less developed countries CADR’s more developed
can focus more on countries can focus more on
Strengthen business advisory and Utilizing extension services to Targeting services through
technology extension services, and increase technology uptake in the assessments of market failures,
technology centers to promote agricultural sector strength of demand, and risk of
technology uptake among firms and overcrowding the market
improve management capabilities
Promote SME use of digital platforms by Piloting initiatives that develop Developing initiatives that help
increasing digital skills, awareness, and digital skills among SMEs connect SMEs to overseas
technology uptake markets
Objective 2: Strengthen pathways for skills development and deployment
Building strong • Invest in early childhood education and strengthen basic literacy and numeracy
foundations for school-age children
All CADR countries can work to CADR’s less developed countries CADR’s more developed
can focus more on countries can focus more on
Develop labor market insight tools Introducing or strengthening Deploying vacancy, skills profiling,
to collect, analyze, and disseminate labor force surveys and utilizing and other specialized surveys and
information about the labor market administrative data exploring novel sources of labor
market information (for example,
online job postings)
Build foundations-driven, demand- Piloting remedial skills and Identifying areas of growing
oriented education and training demand-driven training programs demand in real time, developing
systems that are designed to be lifelong that improve literacy and training programs in response,
and targeted to workers at greater risk numeracy and basic digital skills and incorporating other support
of labor market disruptions and that fill labor market demand services into these training
in strategic areas programs
Design digitally enabled, fit-for-purpose Building a public employment Expanding the public employment
intermediation programs that focus on services system that is a reliable services system to provide labor
overcoming geographic disparities and information source market intelligence, career and
information problems Exploring global skills skills guidance, job matching, and
partnerships (GSPs) to create safe referral services
and inclusive migration pathways
Objective 3: Adapt social protection and labor market policies to new forms of work
Building strong • Move away from reliance on traditional employer-employee relationships for
foundations financing and providing social protection
All CADR countries can work to CADR’s less developed countries CADR’s more developed
can focus more on countries can focus more on
Exploit the potential of platform work Avoid regulations that lead to In the short term, explore models
further labor market segmentation for expanding access to social
protection to platform workers
Monitor potential anti-competitive Develop models of the Develop more sophisticated
practices businesses, characteristics, and analyses of the particular
anti-competitive potential of anticompetitive practices of
digital platforms platforms, especially issues
related to data
Executive Summary XX
Chapter 1
Jobs in Central America and
the Dominican Republic
Labor markets in Central America and the Dominican Republic (CADR) face significant challenges despite
improvements in recent years. Structural change is slow in several countries. No matter the stage of
structural transformation, all CADR countries are experiencing strong growth in services employment. This
will require creating better quality jobs in a sector typically characterized by low-wage, low-productivity
work. Indeed, job quality is already a challenge. Lack of access to social protection is the norm and wage
growth has been stagnant in many countries. Advancements have been made in education levels, but
this human capital is underutilized: low female and youth labor force participation rates mean that a large
portion of the region’s human capital is not being deployed in the labor market. Large outmigration flows
from several CADR countries imply that human capital is better utilized elsewhere. The private sector
needs more dynamism to create high-quality jobs. Barriers to entry and firm growth translate into high
rates of small and less productive informal firms.
SETTING THE STAGE: LABOR MARKET DYNAMICS AND TECHNOLOGICAL PROGRESS
Understanding how the nature of work will evolve in the coming years in CADR first requires
understanding existing labor market dynamics. Much has been written about the potential impact of
technological progress on labor markets in developed countries, but less is known about the interplay
between technology and jobs in CADR where preexisting labor market challenges may mean a different
evolution. Much of the region is characterized by low female labor force participation, high unemployment,
limited social protection coverage, low skill levels, and challenging business environments. Understanding
these dynamics is important, both because they will help determine the future impact of technological
change in the region and because they have affected the impacts of technological changes that have
already occurred.
This chapter presents a high-level summary of labor markets in CADR . The chapter is drawn from
recent literature on jobs and labor markets in CADR countries supplemented by new analysis. The chapter
begins with a discussion of growth and employment dynamics, then moves on to discuss labor supply and
labor demand in turn. Table 1.1 summarizes the main labor market issues in CADR countries.
1
TABLE 1.1: Summary of the Main Labor Market Supply and Demand Issues in CADR
Country Macroeconomic context Challenges related to labor supply Challenges related to labor demand
CRI • Strong growth explained by growth in • Elevated unemployment (UE) rate • High tax wedge raises incentives for
labor productivity • Large gender labor force participation informal employment
• Substantial structural transformation (LFP) and UE gap; high youth UE and • Fragmented dialogue between workers
not in education, employment, or and employers
training (NEET) rates, esp. women
• Skills gaps, particularly in high-value
sectors
• High secondary school dropout rate
DOM • Strong growth explained by growth in • Large gender LFP and UE gap; high • Market concentration, special tax
labor productivity NEET rates, esp. women regimes, credit access problems, and
• Substantial structural transformation • Challenges with education quantity and unreliable electricity access disrupt firm
quality entry and growth
• Substantial outmigration • Multiple minimum wages with lagging
adjustment create labor market
distortions
GTM • Slow growth with limited labor • Low overall LFP • Barriers to entry and unfavorable
productivity growth • Large gender LFP gap; high NEET political and governance context create
• Structural transformation with smaller rates, esp. women lack of contestability and disincentivize
fall in agricultural employment • Challenges with education quantity, (formal) entry
esp. rural areas
• Limited internal mobility
• Substantial outmigration
HND • Slow growth with limited labor • Large gender LFP, UE gap; high NEET • Increases in minimum wage may
productivity growth rates, esp. women hamper formal job growth
• Structural transformation with smaller • Challenges with education quality, • High public sector wage premium
fall in agricultural employment particularly technical and vocational creates competition with private sector
education and training (TVET)
• Labor market mismatches
• Substantial outmigration
NIC • Slow growth with limited labor • Large gender and youth LFP gaps; high • Barriers to entry and growth
productivity growth youth UE; high NEET rates, esp. women disincentivize formality, leaving most
• Structural transformation with smaller • Substantial outmigration jobs in unproductive micro firms
fall in agricultural employment
PAN • Strong growth explained by growth in • Elevated UE rate • Challenge to shift to knowledge-driven
labor productivity • Large gender LFP gap; high youth UE, growth model; low investment in R&D
• Substantial structural transformation NEET rates, esp. women and ineffective diffusion of knowledge
• Challenges with education quality affect innovation
create skills gaps • Unequal access to infrastructure
• High secondary school dropout rate
SLV • Slow growth with limited labor • Low overall LFP rate • High minimum wages, tax wedge and
productivity growth • Large gender LFP gap; high NEET rigid regulations discourage formality
• Structural transformation rates, esp. rural women and reduce productivity
• Labor market mismatches • Crime disincentivizes formality and firm
• High public sector wages distort skilled growth and reduces profitability
labor • Lack of technology use by small firms
• Substantial outmigration, esp. higher- constrains productivity
skilled
Sources: IFC 2023 and OECD 2017c for Costa Rica; Abdullaev and Estevão 2013, OECD 2022a, USAID 2020, and Winkler and Montenegro 2021, for the Dominican Republic; Banegas and Winkler
2020 for El Salvador; Eberhard-Ruíz 2021 and USAID 2017a for Guatemala; Michel and Walker 2019 and USAID 2017b for Honduras; Castro-Leal and Porras-Mendoza 2020 and World Bank 2012b for
Nicaragua; OECD 2017a for Panama; Bashir, Gindling, and Oviedo 2012; Ulku and Zaoruak 2021; and World Bank 2022b for the Central America and the Dominican Republic (CADR) region as a whole.
Note: CRI = Costa Rica; DOM = Dominican Republic; esp. = especially; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; R&D = research and development; SLV = El Salvador.
Chapter 1 Jobs in Central America and the Dominican Republic 2
GROWTH AND EMPLOYMENT DYNAMICS
Costa Rica, the Dominican Republic, and Panama are CADR’s highest-income, highest-productivity
countries. El Salvador, Guatemala, Honduras, and Nicaragua, in contrast, have been relatively stagnant
in the last two decades with only modest productivity improvements. Deindustrialization characterizes
employment in the region: employment has shifted strongly out of agriculture and is now concentrated
in the services sector.
Economic Growth
CADR includes countries at diverse levels of economic development. The region’s sole high-income
country, Panama, has a gross domestic product (GDP) per capita that is five times higher than that of
Nicaragua and Honduras, the region’s least developed countries (figure 1.1). The region’s higher-income
countries grew more during the last two decades than the lower-income ones (figure 1.2). After beginning
the period with a similar GDP per capita to that of Panama, Costa Rica grew more slowly than its southern
neighbor, while the Dominican Republic caught up to Costa Rica. El Salvador, Guatemala, Honduras, and
Nicaragua all grew slowly with no substantial catch-up growth.
FIGURE 1.1: GDP Per Capita, 2002–21 FIGURE 1.2: GDP Growth, 2002–19
Purchasing Power Parity, constant 2017 international $ Percentage average annual growth
35,000 8.0%
30,000 7.0%
25,000 6.0%
20,000 5.0%
15,000 4.0%
10,000 3.0%
5,000 2.0%
0 1.0%
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
0.0%
PAN CRI DOM SLV PAN DOM CRI HND GTM NIC SLV
GTM NIC HND 2002-2019 2002-2011 2012-2019
Source: World Development Indicators data, World Bank. Source: World Development Indicators data, World Bank.
Note : CRI = Costa Rica; DOM = Dominican Republic; GDP = gross Note : CRI = Costa Rica; DOM = Dominican Republic; GDP = gross
domestic product; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; domestic product; GTM = Guatemala; HND = Honduras; PAN = Panama;
PAN = Panama; SLV = El Salvador. NIC = Nicaragua; SLV = El Salvador.
Labor productivity has driven growth in the region’s higher-income countries. CADR has a substantial
productivity gap with comparators (Ulku and Zaourak 2021). Previous analyses decomposing economic
growth into changes in labor productivity and in demographic and employment trends reveal two sets
of countries. Increasing labor productivity has been the main growth driver in Costa Rica, the Dominican
Republic, and Panama, explaining about 80 percent of per capita value-added growth between
2000 and around 2017 (Banegas and Winkler 2020; Castro-Leal and Porras-Mendoza 2020; Winkler
and Montenegro 2021). In El Salvador, Guatemala, Honduras, and Nicaragua, in contrast, economic
growth during this period was slow and mostly driven by changes in demographic factors rather than
labor productivity improvements (Banegas and Winkler 2020; Castro-Leal and Porras-Mendoza 2020;
Eberhard-Ruiz 2021; Michel and Walker 2019).
Chapter 1 Jobs in Central America and the Dominican Republic 3
Productivity
The relatively insignificant role of labor productivity in explaining growth in many CADR countries
indicates challenges for longer-term economic development. Productivity growth is the main engine of
sustained economic progress. While favorable demographic and employment trends can yield periods
of economic growth, improvements in labor productivity are necessary for sustained convergence with
advanced economies.
Costa Rica, the Dominican Republic, and Panama stand out as the leaders in productivity in CADR.
Productivity, proxied as value added per worker, is highest in Costa Rica, the Dominican Republic,
and Panama, though still well below the levels in aspirational benchmarks Korea, the Organisation
for Economic Co-operation and Development (OECD), and the United States (figure 1.3). Productivity
differences across countries are starkest in the industrial sector where Panama’s productivity is nearly
double that of Costa Rica and the Dominican Republic, which themselves have industrial productivity
that is at least twice that of the region’s remaining countries. Cross-country differences in services
sector productivity are also significant, though only two main groups—Costa Rica and Panama, on
one hand, and the region’s other countries, on the other—stand out. Differences are less stark in the
agricultural sector.
FIGURE 1.3: Labor Productivity Across Sectors, 2019
Value added per worker, constant 2015 US$
140,000
120,000
100,000
80,000
60,000
40,000
20,000
0
USA
OECD
KOR
DOM
CRI
LAC
PAN
GTM
SLV
NIC
HND
USA
OECD
KOR
PAN
CRI
DOM
LAC
GTM
SLV
NIC
HND
USA
OECD
KOR
PAN
CRI
LAC
DOM
GTM
SLV
HND
NIC
Agriculture Industry Services
Source: World Development Indicators data, World Bank.
Note: Data for the industrial sector for the Republic of Korea are for 2015. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras;
KOR = Korea; LAC = Latin America and the Caribbean; NIC = Nicaragua; OECD = Organisation for Economic Co-operation and Development; PAN = Panama;
SLV = El Salvador; USA = United States of America.
Productivity growth has been minimal in most CADR countries. Since 2002, agricultural productivity
has grown substantially in the Dominican Republic (figure 1.4a). Productivity growth has also been
significant in Costa Rica and Honduras, though agricultural productivity has fallen relative to 2002 levels
in Costa Rica in recent years. Agricultural productivity is lower than 2002 levels in El Salvador and
Panama, and growth has been minimal in Guatemala and Nicaragua. Panama stands out for the rapid
increase in its industrial productivity, which was nearly triple the 2002 level in 2019 (figure 1.4b). The
Dominican Republic is another standout with industrial productivity growth of 62 percent. Panama again
stands out for a substantial increase in services productivity, along with Costa Rica and the Dominican
Republic (figure 1.4c).
Chapter 1 Jobs in Central America and the Dominican Republic 4
FIGURE 1.4: Labor Productivity Growth Across Sectors, 2002–19
Change in value added per worker, percent, base year = 2002
a. Agriculture b. Industry C. Services
160% 200% 80%
70%
120% 150% 60%
50%
80% 100% 40%
30%
40% 50% 20%
10%
0% 0% 0%
-10%
-40% -50% -20%
2002
2004
2006
2008
2010
2012
2014
2016
2018
2002
2004
2006
2008
2010
2012
2014
2016
2018
2002
2004
2006
2008
2010
2012
2014
2016
2018
CRI DOM GTM NIC PAN SLV HND
Source: World Development Indicators data, World Bank.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Employment
Despite modest economic growth, CADR countries did generate new jobs in the last two decades .
Employment growth between 2002 and 2021 was highest in Honduras (2.5 percent), Nicaragua (2.4
percent), and Guatemala (2.3 percent), in part reflecting the lower contribution of labor productivity to
economic growth in these countries (figure 1.5). Across the region, employment growth was lower in the
2010s than in the 2000s.
FIGURE 1.5: Employment Growth, 2002–19
Percentage average annual growth
4.0%
3.5%
3.0%
2.5%
2.0%
1.5%
1.0%
0.5%
0.0%
HND NIC GTM DOM PAN CRI SLV
2002–19 2002–11 2012–19
Source: World Development Indicators data, World Bank.
Note: The period 2012–21 omits 2020, which was significantly affected by the COVID-19 pandemic. CRI = Costa Rica; DOM = Dominica; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Employment has shifted out of agriculture in CADR countries. The canonical path of economic
development known as structural transformation anticipates employment first shifting from
(low-productivity) agriculture to (higher-productivity) industry and then from industry to (lower-productivity)
services as industrial productivity increases. Consistent with this model, agricultural employment has
Chapter 1 Jobs in Central America and the Dominican Republic 5
declined in all CADR countries over the longer term. Though available irregularly for the last 60 years,
census data show a substantial decline in agricultural employment (figure 1.6). This decline continued
between 1991 and 2019, the period during which comparable survey data is available (figure 1.7). Still,
smaller declines in agricultural employment in Guatemala, Nicaragua, and Honduras mean that nearly a
third of employment remains in low-productivity agriculture.
FIGURE 1.6: Change in Employment Share by Sector, 1960–2010
Percentage point
60
40
20
0
-20
-40
-60
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
CRI: 1963-2011 DOM: 1960-2010 GTM: 1964-2002 HND: 1961-2001 NIC: 1971-2005 PAN: 1960-2010 SLV: 1992-2007 USA: 1960-2015
Source: Minnesota Populavwtion Center 2020.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador; USA = United
States of America.
FIGURE 1.7: Employment Share by Sector, 1991–2019
Percentage
a. Agriculture b. Industry c. Services
100% 100% 100%
90% 90% 90%
80% 80% 80%
70% 70% 70%
60% 60% 60%
50% 50% 50%
40% 40% 40%
30% 30% 30%
20% 20% 20%
10% 10% 10%
0% 0% 0%
CRI DOM GTM HND NIC PAN SLV CRI DOM GTM HND NIC PAN SLV CRI DOM GTM HND NIC PAN SLV
1991 1999 2009 2019
Source: World Development Indicators data, World Bank.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
CADR countries are deindustrializing. Deindustrialization—the shift of employment away from industry
and towards services—is a common phenomenon in advanced economies (see the United States in
figure 1.6) and considered to be the final stage of the structural transformation process. However, recent
evidence suggests that in developing countries in general and in Latin America in particular, the growth
in the services share of employment is happening at lower levels of development and at lower levels of
Chapter 1 Jobs in Central America and the Dominican Republic 6
peak manufacturing employment (Rodrik 2016).2 This seems to be the case in CADR countries where, at
odds with the typical model of structural transformation, employment has not shifted strongly into industry.
Over the long term, gains in the industrial share of employment have been modest and in the three most
recent decades, this share actually shrank or plateaued in all CADR countries except Panama (figures 1.6
and 1.7). In services, in contrast, employment has grown substantially in both the longer term and in the
three most recent decades.
LABOR SUPPLY
CADR countries face substantial challenges developing and deploying human capital in productive
employment. Low education levels combined with evidence of low educational quality raise concerns
about how well current and future workers will be able to complement and facilitate technological
progress and respond to changes in work resulting from technological progress. At the same time, low
labor force participation and high unemployment rates in some countries, particularly for women and
young people, combined with high levels of informality (proxied by access to social protection) and self-
employment, suggest that CADR countries struggle to deploy human capital.
Labor Force Participation and Employment
Labor force participation rates have generally improved in CADR countries in the last two decades,
with a few exceptions. Between 2002 and 2011, most countries in the region had labor force participation
rates below the Latin American and the Caribbean (LAC) average. This trend began to change around
2010: Costa Rica, the Dominican Republic, Honduras, Nicaragua, and Panama all had higher participation
rates in the 2012–19 period prior to the COVID-19 pandemic. In contrast, labor force participation rates
declined during this period in El Salvador and Guatemala, leaving these as the only CADR countries
with lower labor force participation rates in 2021 than the LAC average (figure 1.8). All CADR countries
experienced declines in participation in 2020 as a result of the pandemic and some of them remained
below prepandemic levels in 2021.
FIGURE 1.8: Labor Force Participation Rate, FIGURE 1.9: Female Labor Force
2021 Participation Rate, 2021
Percentage Percentage
66% 55%
64%
50%
62%
60% 45%
58% 40%
56%
35%
54%
52% 30%
NIC PAN DOM KOR HND CRI LAC OECD GTM SLV KOR PAN OECD DOM CRI LAC NIC HND SLV GTM
Source: World Development Indicators data, World Bank. Source: World Development Indicators data, World Bank.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; KOR = Korea; LAC = Latin America and the Caribbean; HND = Honduras; KOR = Korea; LAC = Latin America and the Caribbean;
NIC = Nicaragua; OECD = Organisation for Economic Co-operation and NIC = Nicaragua; OECD = Organisation for Economic Co-operation and
Development; PAN = Panama; SLV = El Salvador. Development; PAN = Panama; SLV = El Salvador.
2 The same is observed for sector value-added, though the phenomenon is more apparent for employment (Beylis et al. 2020). See also Felipe,
Mehta, and Rhee (2019).
Chapter 1 Jobs in Central America and the Dominican Republic 7
Women continue to participate in CADR labor markets at much lower rates than men. Female labor
force participation rates in Guatemala (36 percent) and El Salvador (44 percent) are low relative to both
other CADR countries and to the LAC regional average (50 percent) (figure 1.9). These are also the only
two countries where the female labor force participation rate declined between 2002 and 2021. The gap
between female and male labor force participation rates is large in all CADR countries (figure 1.10). Low
female labor force participation rates are associated with penalties arising from having children, being
married, and inadequate availability of formal wage employment, which themselves are linked to gender
norms and legal barriers (Almeida and Viollaz 2022; Eberhard-Ruiz 2021; Michel and Walker 2019). Other
factors include receipt of remittances and teenage pregnancy (Sousa and García-Suaza 2018; Winkler
and Montenegro 2021).
FIGURE 1.10: Female-Male Gap in Labor FIGURE 1.11: Unemployment Rate, 2002–21
Force Participation Rate, 2021
Percentage point Percentage
OECD KOR PAN CRI LAC DOM HND SLV NIC GTM 18%
0
16%
-5
14%
-10
12%
-15
-20 10%
-25 8%
-30 6%
-35 4%
-40 2%
-45 0%
CRI PAN LAC HND DOM OECD NIC SLV KOR GTM
-50 2002–11 2012–19 2020 2021
Source: World Development Indicators data, World Bank. Source: World Development Indicators data, World Bank.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; KOR = Korea; LAC = Latin America and the Caribbean; HND = Honduras; KOR = Korea; LAC = Latin America and the Caribbean;
NIC = Nicaragua; OECD = Organisation for Economic Co-operation and NIC = Nicaragua; OECD = Organisation for Economic Co-operation and
Development; PAN = Panama; SLV = El Salvador. Development; PAN = Panama; SLV = El Salvador.
During the last two decades, unemployment rates were low in CADR countries relative to the LAC
average. Between 2002 and 2019, every CADR country’s average unemployment rate was below LAC’s
(figure 1.11). Unemployment spiked during the pandemic in all CADR countries except the Dominican
Republic and remained elevated in 2021 in most countries. The gender gap in unemployment is largest in
Costa Rica, the Dominican Republic, and Honduras (figure 1.12). Female labor force participation increased
substantially during the last two decades in these countries, suggesting that women sought work in
greater numbers but faced challenges accessing employment.
Young people struggle to utilize their human capital in CADR labor markets. Young people have both
lower labor force participation rates and higher unemployment rates than older people in all CADR
countries. For example, the age gap in the unemployment rate ranges from 2 percentage points in
Guatemala and Honduras to more than 25 percentage points in Costa Rica, Nicaragua, and Panama. The
youth unemployment rate is also more volatile and susceptible to shocks. For example, young people
suffered a larger unemployment spike due to the COVID-19 pandemic. Rates of being outside of education
and training in addition to outside of the labor force (NEET rates) are also high in most countries (figure
1.13). Young women are more likely to be disengaged from these activities with double-digit gaps between
women and men in all CADR countries except Costa Rica and Panama.
Chapter 1 Jobs in Central America and the Dominican Republic 8
FIGURE 1.12: Female-Male Gap in FIGURE 1.13: Young People in NEET
Unemployment Rate, 2002–21 Status, 2021
Percentage point Percentage
10.0 50%
8.0 45%
40%
6.0
35%
4.0 30%
2.0 25%
20%
0.0
15%
-2.0 10%
-4.0 5%
-6.0 0%
CRI DOM HND LAC GTM SLV PAN OECD NIC KOR GTM DOM HND SLV NIC LAC PAN CRI OECD
2002–11 2012–19 2020 2021 Female Male Total
Source: World Development Indicators data, World Bank. Source: World Development Indicators data, World Bank.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; Note: Data is for 2019 for Guatemala and Honduras and 2014 for Nicaragua.
HND = Honduras; KOR = Korea; LAC = Latin America and the Caribbean; CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
NIC = Nicaragua; OECD = Organisation for Economic Co-operation and HND = Honduras; LAC = Latin America and the Caribbean; NEET = not in
Development; PAN = Panama; SLV = El Salvador. education, employment, or training; NIC = Nicaragua; OECD = Organisation
for Economic Co-operation and Development; PAN = Panama; SLV = El
Salvador.
Job Quality
Self-employment represents a large share of all jobs in most CADR countries. Wage employment is less
common than in the OECD and Korea in all CADR countries and is less common than the LAC average in
all except Costa Rica (figure 1.14). Self-employment is most common in Honduras where half of workers
are self-employed. Despite relatively low levels, wage employment increased in most CADR countries
in the last two decades, though these increases were uneven within and across countries (figure 1.15).
Men are more concentrated than women in self-employment in the region’s higher-income countries.
The relationship reverses in El Salvador, Guatemala, Honduras, and Nicaragua. These trends generally
hold over time, though the increase in self-employment in Panama in recent years has been driven by an
increasing share of self-employed women.
FIGURE 1.14: Employment Share by Wage FIGURE 1.15: Employment Share by Wage
and Self-Employment, 2019 Employment, 2002–19
Percentage Percentage
100% 80%
70%
80%
60%
60% 50%
40%
40% 30%
20%
20%
10%
0% 0%
OECD KOR CRI LAC SLV PAN GTM DOM NIC HND CRI DOM GTM HND NIC PAN SLV
Self-employment Wage employment 2002 2010 2019
Source: World Development Indicators data, World Bank. Source: World Development Indicators data, World Bank.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; KOR = Korea; LAC = Latin America and the Caribbean; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
NIC = Nicaragua; OECD = Organisation for Economic Co-operation and
Development; PAN = Panama; SLV = El Salvador.
Chapter 1 Jobs in Central America and the Dominican Republic 9
Informality is common in CADR with rates are as high as 80 percent in Honduras and Nicaragua (figure
1.16). The International Labour Organization’s (ILO) definition of informal workers, which includes workers
not receiving social protection as well as own-account and family workers, is used as a proxy measure for
job quality. Informal employment represents less than half of jobs only in Costa Rica, though even there
informality is high by international standards and informal workers struggle to shift to formal jobs (OECD
2017c). Women have higher informality rates in Costa Rica and El Salvador while men have higher rates in
the Dominican Republic, Honduras, and Panama.
FIGURE 1.16: Informality Rate, 2021
Percentage
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
HND NIC GTM SLV DOM PAN CRI
Female Male Total
Source: ILOSTAT.
Note: The years are 2012 for Nicaragua, 2017 for Honduras, and 2019 for Guatemala. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Wages have been stagnant outside of the CADR region’s higher-income countries. Analysis of hourly
wages reveals two groups of countries. First, Costa Rica and Panama have experienced wage growth
in recent years and have hourly wages above the LAC average. Second, the Dominican Republic, El
Salvador, and Honduras have stagnant hourly wages that are below the regional average. Women’s hourly
earnings are higher than men’s in Costa Rica, Guatemala, and Nicaragua while they are lower in the
Dominican Republic, El Salvador, Honduras, and Panama (Urquidi and Chalup 2023). This pattern persists
once characteristics like education and occupation are controlled for.
Skills
Improvements in education quantity in CADR are complicated by continued challenges related to
quality. CADR countries have made significant progress in schooling attainment. Data from the 2020
Human Capital Index show that across the region children can expect to complete on average 11 years of
schooling (figure 1.17). Even in the poorest performing countries in the region (Guatemala and Honduras)
the average child can still expect to complete nine years of school, indicating that they remain in school
through the end of basic education. However, learning outcomes are low across the region (World Bank
2022b; 2023b). Adjusting years of schooling for the quality of learning shows that children can only expect
to receive on average seven years of education. Learning poverty rates, the share of children who are out
of school or who cannot read a basic text by ten, are nearly 80 percent in every country with the exception
of Costa Rica and El Salvador (figure 1.18). When children fail to master basic skills early in school, it can be
difficult for them to build more advanced skills later, undermining the benefits of additional years spent in
the classroom (Belafi, Hwa, and Kaffenberger 2020).
Chapter 1 Jobs in Central America and the Dominican Republic 10
FIGURE 1.17: Schooling and Learning- FIGURE 1.18: Learning Poverty, 2019
Adjusted Schooling, 2019
Years Percentage of primary-age children
14 90%
12 80%
70%
10
60%
8
50%
6
40%
4 30%
2 20%
10%
0
CRI DOM SLV NIC PAN GTM HND
0%
Expected years of school Learning-adjusted years of school HND NIC GTM PAN DOM SLV CRI
Source: World Bank Human Capital Index. Source: World Development Indicators data, World Bank.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador. HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
More education is associated with better labor market outcomes and higher-quality jobs, translating
into better employment prospects in all CADR countries (figures 1.19 and 1.20). However, informality
rates among the most educated are still high—above a quarter of workers—in Guatemala and Honduras.
FIGURE 1.19: Employment Rate by FIGURE 1.20: Informality Rate by
Education, 2021 Education, 2021
PerCentage Percentage
100% 100%
90% 90%
80% 80%
70% 70%
60% 60%
50% 50%
40% 40%
30% 30%
20% 20%
10% 10%
0% 0%
CRI DOM GTM HND NIC PAN SAL CRI DOM GTM HND NIC SAL
Low Medium High
Source: SEDLAC (CEDLAS and The World Bank). Source: SEDLAC (CEDLAS and The World Bank).
Note: The years are 2014 for Guatemala and Nicaragua and 2019 for Note: The years are 2014 for Guatemala and Nicaragua and 2019 for
Honduras. Low is 8 years or less of formal education, medium is between Honduras. Low is 8 years or less of formal education, medium is between
9 and 13, and high is more than 13. CRI = Costa Rica; DOM = Dominican 9 and 13, and high is more than 13. CRI = Costa Rica; DOM = Dominican
Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; SLV = El
PAN = Panama; SLV = El Salvador. Salvador.
High-skilled employment is limited in CADR countries. Even the region’s higher-income countries have
high-skilled employment shares that are well below the OECD average (figure 1.21). Instead, low- and
mid-skilled jobs dominate, representing at least half of employment in all CADR countries. During the last
decade when comparable data is available, there has been little growth in high-skilled jobs as a share of
total employment while mid-skilled jobs have increased (figure 1.22). In all CADR countries, women have
a higher share of employment in high-skilled occupations compared to men and, in most countries, are
less concentrated in mid- and low-skilled jobs.
Chapter 1 Jobs in Central America and the Dominican Republic 11
FIGURE 1.21: Employment Share by Skill FIGURE 1.22: Change in Employment Share
Level, 2021 by Skill Level, 2011-21
Percentage Percentage point
100% 25
20
80%
15
60% 10
5
40%
0
-5
20%
-10
0% -15
OECD KOR PAN LAC CRI DOM NIC SLV HND GTM CRI DOM GTM HND PAN SLV
Low Medium High
Source: ILOSTAT. Source: ILOSTAT.
Note: Data are for 2019 for Guatemala, 2020 for Honduras, and 2014 for Note: Nicaragua is omitted because of lack of data. The end year is 2019
Nicaragua. Low skill is elementary occupations. Medium skill is clerical, for Guatemala and 2020 for Honduras. Low skill is elementary occupations.
service, and sales workers; skilled agricultural and trades workers; and Medium skill is clerical, service, and sales workers; skilled agricultural
plant and machine operators, and assemblers. High skill is managers, and trades workers; and plant and machine operators, and assemblers.
professionals, and technicians. CRI = Costa Rica; DOM = Dominican Republic; High skill is managers, professionals, and technicians. CRI = Costa
GTM = Guatemala; HND = Honduras; KOR = Korea; LAC = Latin America Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras;
and the Caribbean; NIC = Nicaragua; OECD = Organisation for Economic PAN = Panama; SLV = El Salvador.
Co-operation and Development; PAN = Panama; SLV = El Salvador.
Migration
Outmigration is a significant phenomenon in several CADR countries. Most CADR countries send more
migrants abroad than they receive (figure 1.23). On net, El Salvador sends the most people abroad as
a share of its population: 43 of every 10,000 people are outmigrants. The number of migrants abroad
has increased since 1990 for all CADR countries except Panama (figure 1.24). Four CADR countries—the
Dominican Republic, El Salvador, Guatemala, and Honduras—have 1 million or more migrants abroad. The
United States is the most common destination for all CADR migrants except for those from Nicaragua
(figure 1.25). Other common destinations for CADR migrants are Canada, Mexico, and Spain outside of the
region. Inside the region, higher-income CADR countries attract more CADR migrants and more migrants
overall.3 Remittance inflows are very large in several CADR countries at around one-fifth or more of GDP
in El Salvador, Guatemala, Honduras, Nicaragua (figure 1.26).
CADR migrants move for a multitude of reasons with a number of important consequences for the
region’s labor markets. Migration from CADR countries is driven by a range of elements (World Bank 2022f;
Voorend, Alvarado, and Oviedo 2021; OECD 2018b). These include “push” factors like lack of economic
opportunities, natural hazards, and violence domestically and “pull” factors such as better economic
opportunities and improved access to services in destination countries. Outmigration has complex labor
market effects, particularly in countries that send large numbers of migrants. Outmigrants from the region’s
sending countries tend to be more skilled than nonmigrants (Del Carmen and Sousa 2018; Winkler and
Montenegro 2021; World Bank 2022f). This raises concerns that investments in human capital at home are
benefiting labor markets abroad and potentially creating skills shortages. Remittances can also affect labor
supply decisions, particularly for women, by raising reservation wages and disincentivizing labor market
participation (Michel and Walker 2019; Sousa and García-Souza 2018; Winkler and Montenegro 2021). At the
same time, outmigration can, at times, raise wages at home by reducing competition for jobs (Gagnon 2011).
3 This report does not focus on CADR countries as destinations for migrants. For discussion of this topic, see Blyde (2020) and Voorend,
Alvarado, and Oviedo (2021) for Costa Rica; Hiller and Chatruc (2023) and OECD (2018) for the Dominican Republic; and Hausmann, Espinoza,
and Santos (2017) and OECD (2017c) for Panama.
Chapter 1 Jobs in Central America and the Dominican Republic 12
FIGURE 1.23: Net Migration Rate, 2021 FIGURE 1.24: Migrants from CADR countries,
1990–2020
Migrants per 10,000 people Number of migrants
20 1,800,000
1,600,000
10
1,400,000
0 1,200,000
1,000,000
-10
800,000
-20 600,000
400,000
-30 200,000
-40 0
1990 1995 2000 2005 2010 2015 2020
CRI DOM GTM HON
-50 PAN CRI HND DOM NIC GTM SLV
NIC PAN SAL
Source: World Development Indicators data, World Bank. Source: United Nations.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador. HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
FIGURE 1.25: Share of CADR Outmigrants in FIGURE 1.26: Remittance Inflows, 2022
the United States, 2020
Percentage Percentage of GDP
100% 30%
90%
25%
80%
70% 20%
60%
50% 15%
40%
10%
30%
20% 5%
10%
0% 0%
GTM SAL HON DOM PAN CRI NIC HON SAL NIC GTM DOM CRI PAN
Source: United Nations. Source: World Bank-KNOMAD.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; Note : CRI = Costa Rica; DOM = Dominican Republic; GDP = gross
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador. domestic product; GTM = Guatemala; HND = Honduras; NIC = Nicaragua;
PAN = Panama; SLV = EL Salvador.
LABOR DEMAND
Competition from the informal sector is a common obstacle facing firms across the CADR region.
The World Bank Enterprise Surveys offer insight into the main challenges for firm operations (table
1.2). Informality is a common challenge: more than 10 percent of firms cite the practices of the informal
sector, which could be a source of unfair competition, as their biggest obstacle in all CADR countries
except the Dominican Republic. Other issues are common to subsets of countries. Crime, corruption, or
political instability are top issues cited by firms in the Dominican Republic, Guatemala, and Panama. An
inadequately educated workforce is a top concern in higher-income Costa Rica and Panama, as well as
in El Salvador and Nicaragua. The Dominican Republic is the only country in which firms frequently report
an infrastructure concern: access to electricity.
Chapter 1 Jobs in Central America and the Dominican Republic 13
TABLE 1.2: The Biggest Obstacles Facing Firms
Country Biggest obstacles (Percentage of firms)
Costa Rica Access to finance (26%), practices of the informal sector (23%), inadequately
educated workforce (13%)
Dominican Republic Corruption (19%), electricity (15%), tax rates (11%)
El Salvador Access to finance (22%), practices of the informal sector (19%), inadequately
educated workforce (12%), business licensing (11%)
Guatemala Corruption (23%), political instability (21%), practices of the informal sector (17%),
crime, theft, and disorder (11%)
Honduras Access to finance (18%), practices of the informal sector (16%), tax rates (11%)
Nicaragua Practices of the informal sector (27%), access to finance (12%), inadequately
educated workforce (12%), customs and trade regulations (11%)
Panama Corruption (33%), practices of the informal sector (14%), inadequately educated
workforce (14%)
Source: World Bank Enterprise Surveys 2010, 2016, 2017, 2023.
Note: Obstacles shown are those that at least 10 percent of firms cite as their biggest obstacle. The years are 2010 for Costa Rica and Panama; 2016 for the
Dominican Republic, Honduras, and Nicaragua; 2017 for Guatemala; and 2023 for El Salvador.
FIGURE 1.27: New Business Density, 2020 FIGURE 1.28: Firms Choosing Inadequate
Education as Biggest Obstacle, 2016
New registrations per 1,000 working age population Percentage of firms
4.5 45%
4.0 40%
3.5 35%
3.0 30%
2.5 25%
2.0 20%
1.5 15%
1.0 10%
0.5 5%
0 0%
PAN OECD CRI KOR LAC DOM GTM SLV CRI HND GTM DOM LAC PAN NIC SLV
Source: World Development Indicators data, World Bank. Source: World Bank Enterprise Surveys 2010, 2016, 2017, 2023.
Note: Data are 2016 for Korea and 2018 for the OECD and the Dominican Note: Data are 2010 for Costa Rica and Panama; 2017 for Guatemala;
Republic. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; and 2023 for El Salvador. CRI = Costa Rica; DOM = Dominican Republic;
KOR = Korea; LAC = Latin America and the Caribbean; OECD = Organisation GTM = Guatemala; HND = Honduras; LAC = Latin America and the Caribbean;
for Economic Co-operation and Development; PAN = Panama; SLV = El NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Salvador.
Recent labor market assessments and survey data from the region identify idiosyncratic challenges
facing labor market demand in different CADR countries, but challenges for firms to enter the market,
survive, and grow are common across countries. With the exceptions of Costa Rica and Panama, all CADR
countries have new business entry rates well below the regional average and far below the rates of OECD
countries and high-income Korea (figure 1.27). In the Dominican Republic, older, less productive firms
dominate employment (Winkler and Montenegro 2021). In El Salvador and Nicaragua, small unproductive
firms are responsible for most employment but struggle to grow (Banegas and Winkler 2020; Castro-
Leal and Porras-Mendoza 2020). Across the region, firms that do enter tend to be informal and face high
Chapter 1 Jobs in Central America and the Dominican Republic 14
formalization costs, while formal firms tend to be older and less productive and may develop market
power that restricts entry (Eberhard-Ruiz 2021; OECD 2022a; Winkler and Montenegro 2021). The result
is that most job creation occurs in the informal sector, where jobs are lower quality and less productive.
Challenges finding workers with the right skills are also common. A higher share of firms in Costa Rica,
Honduras, Guatemala, and the Dominican Republic report an inadequately educated workforce as a major
constraint compared with the LAC average (figure 1.28). Nearly half of employers reported difficulties
finding new employees in 2019 in the Dominican Republic (USAID 2020). Labor market assessments done
by USAID find skills constraints to be an important obstacle facing growth industries in the Dominican
Republic, El Salvador, and Honduras (USAID 2017b, 2017c, and 2020). Even where firms are less likely to
report inadequate education as a constraint, skills issues arise. For instance, lack of skills is less frequently
cited as a barrier in El Salvador, though skills mismatch appears to undermine labor market demand
(Banegas and Winkler 2020).
Chapter 1 Jobs in Central America and the Dominican Republic 15
Chapter 2
The Labor Market Impacts
of Technological Progress
This chapter defines the analytical framework used in the report. Technological progress holds great
possibilities for growth, including growth that creates jobs. However, what people do at work (their tasks)
and how they do it (their working arrangements) are almost certain to change as new technologies are
adopted. Advanced economies offer a model for how the future of work will look in CADR, but important
differences in development stages mean that the labor market impacts of technological progress are
likely to be distinct now and in the near future. Employment structure, skills supply, the adoption and
diffusion of technology, and globalization will all shape the impact of technological progress on labor
markets in the region. Technology adoption abroad is likely to have its own impacts through changes
in demand for labor that is embodied in the movement of goods (offshoring), the movement of people
(migration), and the movement of services (digital trade in services).
SETTING THE STAGE: THE HISTORY OF THE FUTURE OF WORK
The kinds of jobs that people do have changed substantially over time. Many jobs that sound
commonplace now—network engineer, internet developer, web designer—appeared only in the 1990s
(Lin 2011). In fact, the majority of jobs done in the United States in 2018 did not exist in 1940 (Autor et al.
2022). Substantial changes have also occurred in what it means to do a certain job. Between 1950 and
2000, most—88 percent—of the changes in tasks in the United States’ labor market occurred within
occupations and job titles (Atalay et al. 2020). That is, the work done by a manager, a machinist, or a
cashier has evolved.4
Technological progress lies behind many of these changes. These labor market evolutions are closely
linked to the adoption of computers, ICT, and other technological developments (Lin 2011; Atalay et al.
2018). This means that the future of work has arrived, at least in the most advanced economies.
The future of work seems to be more distant in other countries, including in CADR. As Chapter 1
showed, labor markets in CADR are better characterized by agricultural employment than by robot-
assisted manufacturing. This means that learning lessons from technology’s impacts on the labor markets
of the most advanced countries of the world requires understanding not only the technological potential
to automate jobs in CADR but assessing the economic potential for new technologies to take hold and
cause significant labor market disruptions. This chapter lays out a framework for understanding the
potential labor market impacts of technological progress on CADR’s labor markets and the factors and
channels that will determine whether and how these impacts emerge.
4 For further discussion of within-occupation changes in tasks, see Autor, Levy, and Murnane (2003), Michaels, Ruach, and Redding (2019),
Ross (2017), and Michaels, Ruach, and Redding (2019) for the United States; Akcomak, Kok, and Romagosa (2016) for the United Kingdom;
and Bachmann et al. (2022), Spitz-Oener (2006), and Koomen and Backes-Gellner (2022), and Bachmann et al. (2022) for Germany.
16
TECHNOLOGY’S LABOR MARKET IMPACTS: CHANGES IN WHAT WORKERS DO AND
HOW THEY DO IT
Ongoing advancements in technology have prompted concern about widespread job displacement but
also raised hopes for improvements in productivity and new innovations that could spur job creation.
Technological progress has the potential to cause significant disruption in labor markets. Advances first in
agricultural and industrial machinery, then computers, and now AI have enabled machines to undertake
a growing range of tasks previously done by humans, which puts jobs at risk. However, far from being a
purely destructive force, technological progress and automation can also generate employment. Where
markets are competitive, automation can create jobs as price declines linked to productivity improvements
create higher demand (box 2.1). New technologies can also generate entirely new tasks, new types of jobs,
and new industries. ICT can facilitate new types of working arrangements that are beneficial for both
workers and firms. This creates concern that slow progress towards the future of work could mean missing
out on the opportunities created by technological advancements.5
BOX 2.1: Competitive Markets and Technology Adoption
Competitive markets are a crucial ingredient for benefiting from technology adoption. The
expansions in output that can result from technological improvements that increase productivity
and that can result in job creation depend, in part, on how well prices respond to the lower costs
linked to technology (Dutz, Almeida, and Packard 2018). Firms are more likely to respond with
lower costs, leading to higher demand and output, when markets are competitive. A virtuous
cycle can be created, as more competition is also linked to greater technology adoption. However,
many markets in LAC are characterized by limited competition and the operation of cartels that
undermine competitive pressures and so productivity as well (Licetti et al. 2021). Overly regulated
product markets characterize Central America generally (Cirera, Cunha, and Lee 2022). In
El Salvador and Honduras, taxes and subsidies as well as market concentration result in low
competition. Policies that can help enhance the competitive environment include those that lower
barriers to external trade, that facilitate domestic competition by avoiding regulatory distortions
on firm entry and exit, and that increase access to finance.
Note: LAC = Latin America and the Caribbean.
No matter the ultimate impact on employment levels, technological progress is altering what people
do at work and how they work. The deployment of technology in the workplace has two impacts on
work. First, technology can add, eliminate, or change the tasks that employers demand and the skills
that workers need to undertake these tasks.6 Second, technology can change how work is organized.
To understand these two impacts, take the example of telephone operators.7 In the middle of the
twentieth century, telephone operators physically routed telephone calls by inserting plugs into jacks.
Computerization of switchboards eliminated the need for operators to undertake certain tasks—
for example, manual switching—and as a result, changed the kinds of skills required for the job, with
5 Acemoglu and Restrepo (2019) formalizes these different potential labor market effects of technology as the displacement effect of
(automation) technologies that substitute labor with capital, the reinstatement effect of (nonautomation) technologies that create new tasks
with a comparative advantage for labor, and the productivity effect of (automation and nonautomation) technologies that permits more flexible
allocation of tasks to factors of production. The first reduces the labor share of value added and the second increases the labor share of
value added and labor demand and results in a positive productivity effect. The nature of the productivity effect determines the reinstatement
effect’s impact on labor demand. See also Acemoglu and Restrepo (2018). Agrawal, Gans, and Goldfarb (2019) undertake a similar exercise
focused specifically on AI. Brynjolfsson and Mitchell (2017) define a set of “nontechnological factors” that mediate the labor market impacts
of technology.
6 This perspective on tasks and skills follows Acemoglu and Autor (2011).
7 This example elaborates on the discussion of telephone operators in Bresnahan (1999).
Chapter 2 The Labor Market Impacts of Technological Progress 17
interpersonal skills—communicating effectively with callers—becoming more important. Computerization
also led to a substantial decline in the number of telephone operators. For example, in the United States,
the number of operators fell from around 400,000 in the 1960s to around 4,000 in 2022. At the same
time, the organization of telephone operators shifted, as computerization allowed for decentralization
from a telephone company to individual businesses with in-house operators, some of which have now
been replaced by voice-enabled chatbots. Notably, in the United States, current telephone operators
experienced negative employment effects but future cohorts of the young women who were the most
common telephone operators were not impacted but rather shifted to comparable jobs as typists and
servers as demand increased in these areas (Feigenbaum and Gross 2022).
In a first stage of digitally enabled automation, computers facilitated the automation of routine tasks.
Certain types of tasks—routine ones—are more susceptible to automation by computers (Acemoglu
and Autor 2011; Autor, Levy, and Murnane 2003). These routine tasks might be either cognitive (that is,
involving thinking or analysis) or manual (that is, involving physical labor) but are suitable for automation
because they “follow explicit programmed rules” and “can be exhaustively specified with programmed
instructions and performed by machines.” Routine manual tasks can be thought of as repetitive hands-on
work like packaging and assembly that requires recurring actions in circumstances that do not change
much (figure 2.1). Routine cognitive tasks consist of administrative work like data entry that do not involve
much problem solving or critical thinking. Nonroutine tasks, in contrast, “cannot at present be described in
terms of a set of programmable rules” and so are less susceptible to automation (Autor, Levy, and Murnane
2003). Again, these could be higher-skilled cognitive or lower-skilled manual tasks. Nonroutine manual
tasks involve flexible hands-on work like groundskeeping and maintenance that requires physical labor
in varying settings. Nonroutine cognitive tasks may be either analytical, involving knowledge work like
strategic planning or risk assessment, or interpersonal, involving people-oriented work like client relations
or counseling. This model of routine-biased technological change predicts that as the costs of computing
decline, repetitive hands-on and administrative jobs—that is, jobs that involve lots of routine manual and
routine cognitive tasks—will decline as well.8
FIGURE 2.1: Classifying Tasks by Their Automatability
Higher-skilled
Manual Cognitive
Repetive hands-on work Administrative work
Routine Packaging Data entry
Assembly Call center operations
Less automatable
Loading and unloading Basic accounting
Flexible hands-on work Knowledge work
Carpentry Risk assessment, strategic planning, forecasting
Non-routine
Groundskeeping People-oriented work
Repair and maintenance Client relations, counseling, conflict resolution
A second stage of digitally enabled automation raises concerns about displacement of the nonroutine
tasks once thought safe from automation. This second stage, driven by AI and mobile robotics9, has
the potential to automate nonroutine tasks including high-skilled knowledge work like forecasting and
high-skilled people-oriented work like counseling (Brynjolfsson and Mitchell 2017). For instance, the latest
iteration of chatbots like ChatGPT allows users to make requests in natural language. A user might ask for
8 Routine-biased technological change is a more sophisticated version of skill-biased technological change. Skill-biased technological change
posits that technological change augments the productivity (and wages) of high-skilled workers relative to less-skilled workers and so favors
higher-skilled employment growth (Mondolo 2021; Katz and Murphy 1992; Sebastian and Biagi 2018).
9 Mobile robots use sensors, artificial intelligence, and other technology to maneuver in order to perform nonroutine manual tasks (Frey and
Osborne 2017).
Chapter 2 The Labor Market Impacts of Technological Progress 18
programming code, courteous written invitations, or a short essay on an historical event and receive a rapid,
plausible response. Automation of these tasks need not involve the explicit programmed rules that constrain
automation under the traditional routine-biased task model. To acknowledge the expansion in computer
capacity to undertake more complex tasks, Frey and Osborne (2017) created a new framework to evaluate
the impact of technological progress in which only “engineering bottlenecks to computerization” hold back
automation. These bottlenecks are tasks involving perception and manipulation (that robots struggle to
undertake) and tasks involving creative and social intelligence (that computers struggle to undertake).
AI and mobile robotics seem to make most occupations susceptible to automation of at least some
of their tasks. Research on the susceptibility of employment to automation finds that only 14 percent of
employment is at high risk of automation in the OECD (Nedelkoska and Quintini 2018).10 Still, an additional
32 percent of jobs are at medium risk, meaning that they involve some tasks that will likely disappear
because of automation. More recent literature focused specifically on AI emphasizes the disruption that
AI is likely to bring within jobs.11 For instance, Brynjolfsson, Mitchell, and Rock (2023) finds that nearly
every occupation in the United States has some tasks that are suitable for machine learning, though no
occupation consists of only these activities. Looking specifically at the subset of machine learning large
language models called generative pretrained transformers (GPTs) of the type that underpins ChatGPT,
Eloundou et al. (2023) estimates that about 80 percent of the workforce in the United States could have
at least 10 percent of their tasks affected by GPTs. In short, the content of specific jobs is likely to change
substantially because of AI and other new technological developments, even if entire jobs do not disappear.
Alongside these developments in digitally enabled automation, continued improvements in ICT are
transforming how work is organized. Firms arise because they organize tasks that would otherwise
be expensive to arrange using separate contracts (Coase 1937). Organizing production, supply chains,
and workers via a firm’s physical infrastructure and internal organizational structure may involve lower
transaction costs than coordinating remote workers or contracting work out. For example, jobs that
involve teamwork, jobs that are difficult to monitor, and jobs that are intensive in manual tasks involve
significant transaction costs that may make remote provision or contracting challenging (Mas and Pallais
2020).12 However, technology lowers these transaction costs, in turn reshaping the boundaries of the firm.
Moreover, the cost of these technologies themselves has generally been declining in recent decades
making them more attractive (Oettinger 2011).
Two transformations in working arrangements are emerging. The World Development Report 2019: The
Changing Nature of Work outlines how technology is remaking firms.13 First, improved ICT is reshaping the
geographic boundaries of the firm by facilitating remote work. Jobs can be done outside of big cities and
developed countries in places where labor and other costs are cheaper. Online labor markets for contract
workers are already dominated by exchanges between developed and developing countries in which
firms in the former employ workers in the latter (Agrawal et al. 2015). Second, the improved technology
associated with platform work is reshaping the functional boundaries of the firm. Firms are increasingly
able to contract tasks outside, including for tasks like software design or marketing that were previously
done internally. Overall, investments in ICT are linked to less vertical integration, smaller firms, and greater
reliance on (offshore) markets for purchase of services (Abramovsky and Griffith 2006; Brynjolfsson et al.
1989; Hitt 1999).
10 Frey and Osborne (2017) estimate that nearly half of employment in the United States is at high risk of automation. Results for other developed
countries following the Frey and Osborne (2017) occupation-based methodology include 42 percent of employment in Canada (Lamb 2016);
35 percent in the United Kingdom (Deloitte 2014); and 33 percent in Singapore (CSF 2015). However, Frey and Osborne (2017)’s focus on
occupations fails to account for differences in the automatability of tasks within occupations. Nedelkoska and Quintini (2018) addresses this
concern. See also Arntz, Gregory, and Zierahn (2016) and Pouliakas (2018). The McKinsey Global Institute, the World Economist Forum, and PwC
have also produced frequently cited measures of automatability using alternative methodologies (Manyika 2017a, 2017b; WEF 2018; PwC 2018).
11 For the analysis of AI generally, see Felten, Raj, and Seamans (2018); Felten, Raj, and Seamans (2021); Grace et al. (2018); Gries and Naudé
(2022); Kogan et al. (2021); Lassébie and Quintini (2022); Martínez-Plumed et al. (2020); Meindl, Frank, and Mendonça (2021); Tolan et al.
(2021); and Webb (2020). For analysis of machine learning specifically, see Brynjolfsson and Mitchell (2017) and Brynjolfsson, Mitchell, and
Rock (2018, 2023). For analysis of generative pretrained transformers, a type of large language model, see Eloundou et al. (2023) and Felton,
Raj, and Seamans (2023).
12 Remote work can also have negative externalities on other workers (Mas and Pallais 2020).
13 See also “How Technology Is Redrawing the Boundaries of the Firm,” The Economist, January 8, 2023.
Chapter 2 The Labor Market Impacts of Technological Progress 19
TECHNOLOGY’S LABOR MARKET IMPACTS: THE EVIDENCE FROM ADVANCED
ECONOMIES
In advanced economies, there is evidence of both job losses and jobs gains from these waves of
technological progress. The process of routine-biased technological change described above is
associated with job loss as automation replaces jobs and job gains as price declines linked to lower capital
costs induce additional product demand and, in turn, additional labor demand (Autor 2022; Montobbio
et al. 2022; Gregory, Salomons, and Zierahn 2022). A recent review of the empirical literature uncovers
several themes (Montobbio et al. 2022).14 First, a set of studies finds that technological change has been
associated with job loss in some areas, particularly in traditional manufacturing in the case of process
innovations, but job gains in others, particularly in high-tech and knowledge-intensive industries where
product innovations are more frequent. Second, recent research on the impact of robots finds that firms
adopting robots tend to increase employment, but primarily by expanding at the expense of smaller and
less innovative firms. This explains in part why, at the aggregate level, some studies find negative impacts
of robots on employment (for example, Acemoglu and Restrepo 2020) while others find no negative
impacts (for example, Graetz and Michaels 2018). Again, where negative impacts are found, they tend to
be present in manufacturing. Third and finally, the firm and its level of innovation plays an important role
in mediating the impacts of technological change. In general, more innovative firms that invest in new
technologies tend to increase employment (Bessen, Denk, and Meng 2022; Hirvonen, Stenhammer, and
Tuhkuri 2022).
The labor market impacts of technological change are felt most by less-skilled workers. Across
different technologies and different measures of technological progress, high-skilled workers are found
to benefit from technological change while less-skilled workers are found to lose out (Hötte, Somers,
and Theodorakopoulos 2023). For instance, in many but not all developed countries, routine-biased
technological change has led to employment polarization, which is a hollowing out of middle-skill jobs.15
Computers replace (low- and mid-skilled) jobs intensive in routine cognitive and manual tasks, complement
(high-skilled) jobs intensive in more advanced nonroutine cognitive (analytical and interpersonal) tasks like
problem solving and coordination, and in some countries, increase (low-skilled services) jobs intensive in
basic nonroutine manual and cognitive tasks like dexterity and interpersonal communication (Autor and
Dorn 2013; Autor et al. 2022; Autor, Levy, and Murnane 2003). While other factors, such as offshoring
that moves routine (industrial) jobs abroad, have played a role, automation has generally been found to
be dominant (Autor and Dorn 2013; Goos, Manning, and Salomons 2009; Goos, Manning, and Salomons
2014; Michaels, Natraj, and Van Reenen 2014).16 Inequality impacts are expected, given the larger impact
on less-skilled workers (Berg, Buffie, and Zanna 2018; IMF 2018). Workers in routine jobs in the United
States have lost wages, which is linked to increased wage inequality (Acemoglu and Restrepo 2022;
Bachmann, Cim, and Green 2019; Cortes 2016; Ross 2017). The negative wage effects of robots in the
United States are primarily concentrated at the bottom and middle of the wage distribution (Acemoglu
and Restrepo 2020). The distribution of skills across geographies has implications for regional equity as
well (box 2.2).
14 Mondolo (2021) and Hötte, Somers, and Theodorakopoulos (2023) are also recent literature reviews of the links between technological
change and labor markets.
15 Employment polarization is found in the United States (Autor and Dorn 2013; Autor, Katz, and Kearney 2006, 2008); Canada (Green and
Sand 2015); West Germany (Dustmann, Ludstock, and Schönberg 2009; Spitz-Oener 2006); the United Kingdom (Goos and Manning 2007);
and Europe (Goos, Manning, and Salomons 2009, 2014; OECD 2017b; Oesch and Menés 2011). Employment polarization may lead to wage
polarization in the case of additional demand for low-wage workers or additional wage inequality, for example, if displaced middle-skilled
workers shift to lower-skilled jobs (Autor 2022). Findings of wage polarization are generally restricted to the 1990s in the United States (Autor
and Dorn 2013; Autor, Katz, and Kearney 2008; Green and Sand 2015). Wage polarization is not found in Canada and Germany (Antonczyk,
DeLeire, and Fitzenberger 2018; Dustmann, Ludstock, and Schönberg 2009; Green and Sand 2015; Koomen and Backes-Gellner 2022).
16 For additional evidence linking technological change to routine and nonroutine tasks, see Akerman, Gaarder, and Mogstad (2015); de Vries
et al. (2020); Gaggl and Wright (2017); Michaels, Natraj, and Van Reenen (2014); and Spitz-Oener (2006).
Chapter 2 The Labor Market Impacts of Technological Progress 20
BOX 2.2: Spatial Differences in the Labor Market Impacts of Technological
Progress
Automation may benefit larger cities. In the United States, larger cities tend to be more
specialized in the kinds of managerial and technical tasks that are less likely to be automated than
smaller cities, which are more likely to experience disruption from automation (Frank et al. 2018).
Indeed, the computer revolution in the United States shifted jobs to cities specialized in analytical
and interactive skills, which tend to be present in larger cities (Berger and Frey 2016).
However, changes in working arrangements could work in the opposite direction, benefiting
smaller cities to the detriment of larger ones. The kinds of jobs that are more suitable to remote
provision tend to be concentrated in cities, meaning that more remote working possibilities could
mean a shift away from the biggest cities. During the COVID-19 pandemic, for example, business
services workers in very dense United States cities were more likely to work from elsewhere than
workers in less dense cities (Althoff et al. 2022). This had knock-on effects for consumer services
workers who rely on spending by business services workers to support their jobs (Althoff et al.
2022; Barrero, Bloom, and Davis 2021). Additionally, platform work can offer income-generating
opportunities where jobs are lacking, as long as good digital infrastructure and digital devices
are available. Recent World Bank research on online gig workers shows that around two-thirds
of these workers in Latin America and the Caribbean are in small cities and towns (Datta, Namita
and Chen 2023).
However, the latest advancements in AI may have different impacts across the skills distribution,
including benefiting less-skilled workers. AI-linked technological changes may have more disruptive
effects among high-skilled occupations. Recent research estimating the exposure of tasks to disruption by
AI and machine learning generally finds that exposure is higher in higher-skilled, higher-wage occupations
(Eloundou et al. 2023; Felten, Raj, and Seamans 2023; Meindl, Frank, and Mendonça 2020; Tolan et
al. 2021; Webb 2020).17 Industries likely to be impacted include information processing, legal services,
and securities and investments (Eloundou et al. 2023; Felten, Raj, and Seamans 2023). These sectors
are typically associated with high-skilled knowledge work. Notably, recent research suggests that AI
assistants increase the productivity of less-skilled workers within occupations, in effect substituting for
worker education and experience (Agrawal, Gans, and Goldfarb 2023). This has been found in the case
of customer support agents, taxi drivers, software developers, and college-educated workers completing
writing tasks (Brynjolfsson, Li, and Raymond 2023; Kanazawa et al. 2022; Noy and Zhang 2023; Peng et
al. 2023).
Other distributional impacts tend to be context specific. For instance, exposure to ICT in Europe
between 2010 and 2018 was beneficial for young and prime-aged workers but bad for older women,
while exposure to robots had adverse impacts on prime-aged men (Albinowski and Lewandowski 2022).
Robot adoption in the United States has larger negative wage and employment effects on men than on
women (Acemoglu and Restrepo 2020). Recent research shows that women who were initially more
exposed to automation than men in the United States shifted to high-skill, high-wage occupations more
than men (Cortés et al. 2023).
17 Brynjolfsson, Mitchell, and Rock (2023) find that both low- and high-wage jobs are suitable for machine learning but show a negative
correlation between suitability and wage.
Chapter 2 The Labor Market Impacts of Technological Progress 21
TECHNOLOGY’S LABOR MARKET IMPACTS: THE VIEW OUTSIDE OF ADVANCED
ECONOMIES
The applicability of this model of how technology impacts labor markets to CADR depends on several
factors. Each of the technologies described above is available around the world. As a result, labor markets
everywhere are susceptible to similar disruptions from technological progress. In fact, the relatively large
share of less-skilled labor in developing countries means that a larger population may be at risk (Schlogl
and Sumner 2018). However, several key factors beyond the existence of a technology determine its
ultimate impact on labor markets. These factors include the adoption and diffusion of technology (Do
people and firms use new technologies?) as well as employment structure (Do sectors that use technology
dominate?), the supply of skills (Are there workers with skills that are complementary to technology?),
and globalization (Is the economy open to technological influences?). Labor markets in developing
countries are also susceptible to technological progress outside of their borders, which could occur even
if technological progress is concentrated in just a few advanced economies, or even in a few superstar
firms (Korinek, Schindler, and Stiglitz 2021). Technology adoption abroad may alter the demand for labor
from developing countries that is embodied in the movement of goods (offshoring), the movement of
people (migration), and the movement of services (digital trade in services).
Evidence of the impact of technological progress on employment in developing countries is mixed so
far. In many parts of the world, declining computing costs have led to a decline in occupations intensive in
routine manual and cognitive tasks (World Bank 2016). However, most developing countries do not show
signs of the employment polarization that characterizes the United States and Western Europe (Maloney
and Molina 2016; Maloney and Molina 2019; Martins-Neto et al. 2021). In Latin America there is evidence
of growth in high-skilled jobs and loss of low-skilled jobs but not of the hollowing out of middle-skilled
jobs except in Brazil and Mexico (Busso and Hincapié 2017; Maloney and Molina 2019; Messina and Silva
2018; Messina and Silva 2021). This is more consistent with skill-biased technological change (technology
benefiting more highly skilled workers) than routine-biased technological change. Several recent studies
have found evidence of a negative link between robots and employment in developing countries. Robots
have been linked to lower employment and wages in China and more negative employment effects in
developing than developed economies (Carbonero, Ernst, and Weber 2020; Giuntella and Wang 2019;
Giuntella, Lu, and Wang 2022). In China, these effects are concentrated among less-skilled workers and
are larger for male, prime-age, and older workers (Giuntella, Lu, and Wang 2022). However, positive
impacts have been found in Indonesia and in cross-country research focusing on manufacturing operators
and assemblers (Calì and Presidente 2021; Maloney and Molina 2019).
Results focused on technology generally have been more positive . Most firms in the 11 primarily
developing economies surveyed by the World Bank’s Firm-Level Adoption of Technology Survey report
maintaining the same number of workers after adopting more sophisticated technologies (Cirera, Comin,
and Cruz 2022). These firms also tend to generate more jobs and may even increase their share of
unskilled workers. Cross-country research on adoption of digital technologies in the manufacturing sectors
of 88 developing countries finds that adoption is labor augmenting (Cusolito, Lederman, and Peña 2020).
Focusing on four countries in Latin America and the Caribbean, Dutz, Almeida, and Packard (2018) find
that technological development leads to job growth, including of low-skilled workers, particularly when
skills complementary to technology are in abundance. In Costa Rica, product innovation and process
innovation are found to be linked to employment growth (Monge-Gonzalez et al. 2011).
Chapter 2 The Labor Market Impacts of Technological Progress 22
The remainder of this report looks at the impact of adopting computers, robots, AI, and improved ICT
at work on labor markets in CADR.18 The report focuses on these technologies as the most likely to shape
labor markets in the region in the near future. It looks at technological progress both within and outside
of the region. It first examines how technological progress within the region is shaping what workers do
(their tasks) and how they do it (their working arrangements). The report goes beyond the typical analysis
of susceptibility to automation to dissect the factors underlying recent labor market transformations and
uncover the extent to which technological change has played a role. It also examines how technological
progress outside of the region is shaping labor markets within it by investigating how robot adoption in
the United States is affecting the demand for CADR workers working in CADR countries (by changing
incentives for offshoring) and for those working in the United States (by changing demand for migrants).
Figure 2.2 depicts the analytical framework described in the previous paragraphs.
FIGURE 2.2: The Labor Market Impacts of Technological Change in CADR
Factors influencing the evolution of work in CADR Impacts on the future
of work in CADR
Outside CADR
O shoring
Technology adoption
Migration
∆ tasks
Technological
progress
Inside CADR ∆ work arrangements
Technology adoption
Employment
Employment structure
Wages
Nontech factors Skills
Globalization
Note: CADR = Central America and the Dominican Republic.
18 Despite the focus on these three types of technology, previous waves of technological change like agricultural and industrial mechanization
remain relevant in the region, particularly because the shift to nonfarm work expected from agricultural mechanization and improved
agricultural inputs has been limited in several countries in CADR (FAO 2022). The report also focuses specifically on technologies that are
used at work. Despite this, a broad set of home, health, and other technologies have important labor market impacts, particularly for women.
Improved internet access has been shown to increase labor force participation and employment rates in developing countries (Chiplunkar
and Goldberg 2022; Hjort and Poulsen 2019). Home appliances and contraception are linked to increased female labor force participation in
Guatemala and other Latin American countries (Almeida and Viollaz 2022; Cubas 2016; Gasparini and Marchionni 2015). Similar results are
found for developed countries. For the impact of home technologies, see Dettling (2017), Cavalcanti and Tavares (2008), Coen-Pirani, León,
and Lugauer (2010), and Greenwood, Seshadri, and Yorukoglu (2005). For health technologies, see Bailey (2006), Albanesi and Olivetti (2016),
and Goldin and Katz (2002).
Chapter 2 The Labor Market Impacts of Technological Progress 23
Chapter 3
Technological Progress and
Labor Market Transformation
in CADR Countries
This chapter describes the impacts of technology on labor markets in CADR that are observable to date.
Over the longer term, there is little evidence that technological progress has shifted employment away
from the occupations that are most susceptible to automation, though trends towards the knowledge and
people-oriented work that is less automatable are apparent in the last decade. Even with these shifts,
employment in El Salvador, Guatemala, Honduras, and Nicaragua is much more intensive in routine,
automatable tasks than other CADR countries and the United States. A large share of employment in
every CADR country seems to be at risk of disruption from AI and robots, but this share is much lower
once sectors with little technological penetration are excluded. Changes in working arrangements are
apparent in CADR, but their extent is limited.
SETTING THE STAGE: TECHNOLOGICAL PROGRESS AND ITS EMPLOYMENT IMPACTS
Evidence from advanced economies shows labor markets in flux, but how technology is affecting
labor markets in CADR is less certain. In advanced economies, technological progress is generating
increased demand for cognitive and socioemotional skills, advanced technical skills, and digital skills.
Automation is reducing the demand for workers to perform routine and even some knowledge and
people-oriented nonroutine tasks. Working arrangements are shifting as digitization creates opportunities
for people to work from home and to find work on job platforms. COVID-19 accelerated some of these
trends, particularly the shift to new types of working arrangements. However, evidence for these trends is
scarcer for CADR. This chapter looks at how tasks and working arrangements are changing in the region.
CHANGES IN TASKS
Certain types of tasks are more susceptible to disruption by computers, AI, and robots. Evidence of
the impact of technological progress on labor markets can be observed in how occupations and tasks of
different types evolve over time. Changes in tasks are observable in changes in occupational shares over
time, as well as by breaking down occupations into different types of tasks that are more and less likely to
be automated. This can be done for both computers, which automate routine tasks, and for AI and robots,
which automate both routine and nonroutine tasks.
Changes in Tasks: Computerization
Long-term changes in occupations in CADR countries do not show signs of decline in jobs associated
with routine, automatable tasks. The first wave of digitally enabled automation described in Chapter
2 involves rules-driven computers that are well suited to take on the routine repetitive hands-on and
administrative work previously done by workers. Looking at census data for CADR countries permits
24
analysis of long-term employment shifts that may be related to technological change, though this can only
be done for broad occupational categories. The data do not show evidence of a decline in the jobs most
associated with routine, automatable tasks between the 1980s and the 2000s. For instance, employment
generally increased for plant and machine operators and assemblers, an occupation typically considered
to be intensive in routine tasks. Declines in crafts and related trades workers, another occupation typically
considered to be routine-intensive, were observed in Costa Rica and El Salvador, but in none of the other
CADR countries. Ordering occupations by their average skill level, proxied by average years of education
in 1980, shows that no CADR country saw a substantial decrease in employment among mid-skilled
occupations (figure 3.1). That is, unlike in advanced economies, changes in occupations in CADR countries
during the last 30 years do not show signs of employment polarization. Instead, shifts in employment
patterns are more consistent with so-called skill-biased technological change: employment changes were
increasing in skill level.
FIGURE 3.1: Change in Employment Share by Skill Level, circa 1980s to circa 2000s
Percentage points
20
15
10
5
0
-5
-10
-15
-20
-25 CRI DOM GTM HND NIC PAN SLV USA
Low Medium High
Source: Minnesota Population Center 2020.
Note: The periods are 1984 to 2011 for Costa Rica, 1981 to 2010 for the Dominican Republic, 1981 to 2002 for Guatemala, 1988 to 2011 for Honduras, 1995
to 2005 for Nicaragua, 1980 to 2010 for Panama, 1992 to 2007 for El Salvador, and 1980 to 2010 for the United States. Occupations at the oneInternational
Standard Classification of Occupations (ISCO) one-digit ISCO-08 level are ranked by average years of education circa 1980 and combined into low (the three
occupations with the lowest average years of education), high (the three occupations with the highest average years of education), and medium (the remaining
occupations) skill levels. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador;
USA = United States of America.
In more recent years, there is evidence of a shift away from the routine tasks that are easier to
automate. Evidence from LAC shows that there has been a shift away from routine tasks and towards
nonroutine ones across the region (Beylis et al. 2020; Gasparini et al. 2021). A similar pattern is evident in
CADR countries. The routine task intensity (RTI) score measures how intensive employment is in routine
tasks.19 The measure is standardized with respect to the value of the United States to allow for cross-
country comparisons. The RTI shows that employment in all CADR countries except Costa Rica and
Honduras became less intensive in routine tasks in the previous decade (figure 3.2a).
Employment became more intensive in knowledge and people-oriented tasks in many CADR countries.
The routine intensity of occupations measured by the RTI can be broken down into different types of tasks:
nonroutine analytical (knowledge work), nonroutine interpersonal (people-oriented work), and routine
cognitive (administrative work) tasks. During the last decade, jobs became more intensive in nonroutine
analytical tasks in all CADR countries except Costa Rica and Honduras and in nonroutine interpersonal
tasks in all except Costa Rica, Guatemala, and Honduras (figure 3.2b and figure 3.2c). Routine cognitive
19 The RTI is constructed from data on tasks from the Program for the International Assessment of Adult Competencies (PIACC) and data on
employment levels. See appendix B for a detailed description of the methodology to calculate the RTI.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 25
tasks became more prevalent in three of the six countries with data available, though these increases
were relatively small and were outweighed by the growing intensity of the nonroutine tasks (figure 3.2d).
Manual tasks, which are not directly incorporated into the RTI because of the inability to distinguish routine
from nonroutine ones in the available data, experienced relatively small changes (figure 3.2e).20
FIGURE 3.2: Changes in the Routine Intensity of Work, 2011–19
Difference in RTI and task indexes
a. RTI b. Nonroutine analytical c. Nonroutine interpersonal
0.1 0.1 0.1
0.05 0.05 0.05
0 0 0
-0.05 -0.05 -0.05
-0.1 -0.1 -0.1
-0.15 -0.15 -0.15
-0.2 -0.2 -0.2
HND CRI PAN DOM SLV GTM DOM SLV PAN GTM CRI HND SLV DOM PAN HND CRI GTM
d. Routine cognitive e. Manual
0.1 0.1
0.05 0.05
0 0
-0.05 -0.05
-0.1 -0.1
-0.15 -0.15
-0.2 -0.2
HND DOM SLV PAN CRI GTM GTM PAN CRI SLV HND DOM
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The end year is 2018 for Panama. Nicaragua is excluded because of lack of available data. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; PAN = Panama; RTI = routine task intensity; SLV = El Salvador.
Despite these recent shifts, employment in the region remains intensive in routine work, particularly in
CADR’s less developed countries. The RTI shows that employment in El Salvador, Guatemala, Honduras,
and Nicaragua is much more intensive in routine tasks than employment in the United States as well as
that in the other CADR countries (figure 3.3). While differences in the intensity of routine cognitive tasks
(administrative work) are small across countries, all CADR countries show a much lower prevalence of
nonroutine analytical and interpersonal tasks (knowledge and people-oriented work) than the United
States, with particularly large differences in El Salvador, Guatemala, Honduras, and Nicaragua. In these four
countries, the size of the differences and the size of the workforce in nonroutine-intensive employment
means a much higher RTI and more routine employment overall. Manual tasks are more prevalent than in
the United States in all CADR countries.
20 These results are generally consistent with previous studies that include CADR countries (Banegas and Winkler 2020; Winkler and
Montenegro 2021; World Bank 2022b). See appendix B for a comparison.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 26
FIGURE 3.3: RTI and Task Intensity of CADR Countries, 2021
Standard deviations from the United States’ average
0.50
0.30
0.10
-0.10
-0.30
-0.50
-0.70
GTM NIC HND SLV CRI DOM PAN
RTI NRA NRI RC M
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The years are 2019 for Guatemala and Honduras, 2018 for Panama, and 2014 for Nicaragua. CRI = Costa Rica; DOM = Dominican Republic;
GTM = Guatemala; HND = Honduras; M = manual; NIC = Nicaragua; NRA = nonroutine analytical; NRI = nonroutine interpersonal; PAN = Panama; RD = routine
cognitive; RTI = routine task intensity; SLV = El Salvador.
Jobs with routine tasks are most common in agriculture and domestic work and less common in
education and health. In all CADR countries with data available, agriculture and domestic work have the
highest RTI (figure 3.4a–f). Employment in these sectors tends to be less intensive in nonroutine cognitive
tasks. On the other extreme, employment in education and health is less routine. Public administration is
also among the least routine intensive sectors, especially in Nicaragua and Panama.21
FIGURE 3.4: RTI by Economic Sector, 2021
Standard deviations from the United States’ average
a. Dominican Republic b. El Salvador
0.60 0.80
0.60
0.40
0.40
0.20
0.20
0.00
0.00
-0.20 -0.20
-0.40 -0.40
ind e
ch stry
tru y
n
po e
ion
s
ion tion
me alth
k
re
try
y
on
ns erce
n
es
ion ion
me alth
k
ice
r
str
or
or
o
r
c
tio
Co ust
ltu
ltu
c
tra mer
cti
cti
us
t
t
cw
cw
rvi
u
u
he
e
rta
a
Sk orta
a
c a serv
& omm
u
u
ind
ind
ind
tru
h
str
str
ric
ric
se
m
sti
sti
&
&
Ed mini
ini
ns
ns
p
Co
Ag
Ag
ch
ch
ch
Pu illed
d
ns
C
dm
Co
ille
te
-te
te
-te
tra
d
at
at
Do
Do
w-
w-
Sk
ca
gh
gh
uc
uc
&
Lo
Lo
bli
bli
Hi
Hi
Ed
ies
ies
Pu
ilit
ilit
Ut
Ut
21 In general, these patterns are consistent with previous findings for other LAC countries (Gasparini et al. 2021).
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 27
FIGURE 3.4: RTI by Economic Sector, 2021 (continued)
c. Guatemala d. Honduras
1.00 0.80
0.80 0.60
0.60
0.40
0.40
0.20
0.20
0.00
0.00
-0.20 -0.20
-0.40 -0.40
e. Nicaragua f. Panama
1.00 0.60
0.50
0.80
0.40
0.60 0.30
0.40 0.20
0.10
0.20
0.00
0.00 -0.10
-0.20 -0.20
-0.30
-0.40
-0.40
-0.60 -0.50
ind e
try
ns try
Co on
po e
ion
s
ion tion
me alth
k
e
try
try
on
ce
n
es
ion
h
k
ice
or
or
ur
c
tio
alt
ur
vic
tra mer
er
cti
cti
us
s
us
us
t
at
cw
cw
ult
ult
u
he
he
rta
a
rta
c a serv
mm
ind
tru
ind
ind
tru
str
str
er
ric
ric
m
po
sti
sti
&
&
ds
Ed mini
ini
ns
Co
Ag
Ag
ch
ch
ch
ch
Pu illed
ion
me
ns
ns
dm
Co
Co
ille
te
-te
te
-te
tra
d
at
at
Do
Do
w-
w-
Sk
Sk
ca
gh
gh
uc
uc
&
&
Lo
Lo
bli
bli
Hi
Hi
Ed
ies
ies
Pu
ilit
ilit
Ut
Ut
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The year is 2019 for Guatemala and Honduras, 2018 for Panama, and 2014 for Nicaragua. Sector is not available for Costa Rica. RTI = routine task intensity.
Male, young, less-educated, and rural workers tend to be employed in more routine-intensive jobs.
When looking at the sociodemographic characteristics of workers, some consistent patterns emerge.
Men, young workers, and those in rural areas are more likely to be employed in jobs that are more routine
intensive (figure 3.5a–c). Workers with low and medium levels of education are employed in jobs that are
more routine intensive than more highly educated workers (figure 3.5d).22 The disaggregation by type of
employment, in contrast, does not show a consistent pattern across countries (figure 3.5e).
Jobs that are intensive in nonroutine tasks have higher returns. Regression estimates show that
nonroutine analytical and interpersonal tasks (knowledge and people-oriented work) contribute positively
to hourly wages in all CADR countries (figure 3.6a–b).23 This finding is in line with previous literature
for developed and developing countries (Saltiel 2020; Autor and Handel 2013). The expected increase
in hourly wages from an increase of half a standard deviation in the RTI ranges from 21 percent in El
Salvador to 35 percent in Nicaragua for nonroutine analytical tasks and from 23 percent in the Dominican
Republic to 36 percent in Honduras for nonroutine interpersonal tasks. The returns to routine cognitive
tasks (administrative work) are not statistically significant (and so are not shown). Manual tasks, in contrast,
have negative returns in all countries except the Dominican Republic where the expected change in
hourly wages is very close to zero and not statistically significant (figure 3.6c).24
22 This pattern is consistent with previous evidence for Argentina, Brazil, Chile, Colombia, Mexico, and Peru (Gasparini et al 2021).
23 This finding is in line with previous literature for developed and developing countries (Saltiel 2020; Autor and Handel 2013).
24 The finding on returns to manual tasks differs from Saltiel (2020), which finds a positive return to manual tasks in a set of nine low- and middle-
income countries using STEP data. Saltiel (2020) uses variation at the individual level, while we use variation at the occupation level.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 28
FIGURE 3.5: RTI by Sociodemographic Characteristics, 2021
Standard deviations from the United States’ average
a. Gender b. Age c. Geography
0.5 0.5 0.7
0.6
0.4
0.4
0.5
0.3
0.3 0.4
0.2 0.3
0.2
0.1 0.2
0.1 0.1
0
0
-0.1 0
-0.1
-0.2 -0.1 -0.2
CRI DOM GTM HND NIC PAN SLV CRI DOM GTM HND NIC PAN SLV CRI DOM GTM HND NIC PAN SLV
Women Men 15–24 25–65 Urban Rural
d. Education e. Employment type
0.6 0.5
0.4 0.4
0.2
0.3
0
0.2
-0.2
0.1
-0.4
-0.6 0
-0.8 -0.1
CRI DOM GTM HND NIC PAN SLV CRI DOM GTM HND NIC PAN SLV
Low Medium High Wage employee Self-employed
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The year is 2019 for Guatemala and Honduras, 2018 for Panama, and 2014 for Nicaragua. Low education is less than 9 years of school, medium
is between 9 and 13 years, and high is 14 or more. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua;
PAN = Panama; RTI = routine task intensity; SLV = El Salvador.
FIGURE 3.6: Returns to Routine and Nonroutine Tasks, 2010s
Percentage change in hourly wages from a 0.5 SD change in the task measure
a. Nonroutine analytical b. Nonroutine interpersonal c. Manual
40% 40% 0%
35% -5%
35%
-10%
30% 30%
-15%
25% 25% -20%
20% 20% -25%
-30%
15% 15%
-35%
10% 10%
-40%
5% 5% -45%
0% 0% -50%
NIC PAN GTM HND CRI DOM SLV HND CRI NIC SLV GTM PAN DOM DOM CRI PAN GTM SLV NIC HND
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: Models are estimated by ordinary least squares (OLS) controlling for gender, age and its square, indicators for educational level, and year fixed effects. Each
bar shows the expected percentage change in hourly wages due to a change in the task measure equal to 0.5 standard deviations of the corresponding task
measure in each country. All models include years with comparable data for each country: 2011 to 2021 for Costa Rica; 2017 to 2021 for the Dominican Republic;
2013 to 2021 for El Salvador; 2010 to 2015 and 2017 to 2019 for Guatemala; 2015 to 2019 for Honduras; 2010 to 2012 for Nicaragua; and 2011 to 2014, 2016, and
2018 for Panama. Robust standard errors are clustered at the country and International Standard Classification of Occupations (ISCO) two-digit occupation level.
CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SD = standard deviation; SLV = El Salvador.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 29
Computerization and CADR Workers in the United States
The importance of CADR migration means that some CADR workers are exposed to the impacts of
technology on labor markets in advanced economies, particularly in the United States. Chapter 1 briefly
introduced the substantial role that migration, especially to the United States, plays for CADR countries.
This means that some CADR workers have been a part of the significant transformation of labor markets
described in Chapter 2. This section looks specifically at whether CADR workers in the United States have
experienced the same kinds of labor market impacts as other migrant groups and as nonmigrant United
States workers.
Migration to the United States from CADR countries has risen over time and CADR migrants now
make up about 15 percent of all recent migrants to the United States. Between 1970 and 2000, there
was a sharp increase in migrants to the United States from all countries including CADR countries (figure
3.7). Since the 2000s, migration from low-income countries generally increased until 2017, after which
migration declined, while migration from middle- and high-income countries declined during most of
this period. Migration from CADR countries, in contrast, increased sharply throughout the entire 2010s,
and except for a COVID-19-related drop in 2020, did not experience a decline after 2017 as occurred in
every other country group. In 2021, there were nearly 450,000 migrants from CADR in the United States
who had arrived in the last three years and around 1.3 million who had arrived in the last decade.25 This
represents about 15 percent of total migrants of both types in the United States. CADR migrants make
up about 0.2 percent of the working age population versus 0.04 percent for migrants from low-income
countries, 0.9 percent for other middle-income countries, and 0.2 percent for other high-income
countries.
FIGURE 3.7: Three-Year migrants in the FIGURE 3.8: Three-Year CADR Migrants in
United States, 1970–2021 the United States, 1970–2021
Number of migrants Number of migrants
800,000 3,500,000 160,000
700,000 140,000
3,000,000
600,000 120,000
2,500,000
500,000 100,000
2,000,000
400,000 80,000
1,500,000
300,000 60,000
1,000,000
200,000 40,000
100,000 500,000 20,000
0 0 0
2000
2005
2006
2009
2002
2020
2008
2003
2004
2007
2001
2010
1990
2012
2015
2016
2019
1980
2021
2018
1970
1980
1990
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2013
2014
1970
2017
2011
CADR LICs HICs MICs (right axis) CRI SAL GTM HON
NIC PAN DOM
Sources: ACS 2000-2021; US Census 1970, 1980, 1990. Sources: ACS 2000-2021; US Census 1970, 1980, 1990.
Note: CADR = Central America and the Dominican Republic; HICs is high- Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
income countries; LICs = lower-income countries; MICs is middle-income HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
countries.
25 The remainder of the analysis will focus on three-year migrants, as trends are very similar between the two groups.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 30
The Dominican Republic, El Salvador, Guatemala, and Honduras are the main senders of migrants to
the United States among CADR countries. These four countries have been responsible for more than 90
percent of the migrants from CADR countries since the early 2000s. During this period, migration from all
four countries generally increased (figure 3.8). Migrants from CADR countries are younger than those from
other countries (30 versus the mid-30s). Nearly half are women. Most have less than secondary education.
This educational pattern has been fairly stable over time, though the share of migrants with secondary
education has increased slightly. In contrast, the skill level of migrants from other countries has increased
substantially over time: at least 30 percent of migrants from low-, middle-, and high-income countries have
tertiary education versus 13 percent in CADR countries.
Migrants from CADR and non-CADR countries have shifted out of jobs that involve routine tasks.
During the last 50 years, migrants from CADR countries have been concentrated in just a few less-skilled
occupations.26 Migrants from non-CADR countries are less concentrated and have substantial shares of
workers in high-skilled jobs.27 Employment in manufacturing jobs declined precipitously for CADR migrants
in the last 50 years from nearly 50 percent of employment in 1970 to 8 percent in 2021, consistent with
the automation of routine manual tasks (repetitive hands-on work) (figure 3.9). This decline also occurred
among non-CADR migrants, but manufacturing jobs made up a much smaller share of their jobs in 1970,
meaning that the overall shift was less drastic. Employment in office and administrative support jobs,
which are intensive in routine cognitive tasks, also declined for both CADR and non-CADR migrants.
FIGURE 3.9: Employment of Three-Year FIGURE 3.10: Employment of Three-Year
Migrants in Manufacturing Jobs in the United Migrants in Construction Jobs in the United
States, 1970–2021 States, 1970–2021
Percentage of CADR and non-CADR employment Percentage of CADR and non-CADR employment
50% 30%
45%
25%
40%
35%
20%
30%
25% 15%
20%
10%
15%
10%
5%
5%
0% 0%
1970 1980 1990 2000 2010 2021 1970 1980 1990 2000 2010 2021
CADR migrants Non-CADR migrants
Sources: ACS 2000-2021; US Census 1970, 1980, 1990. Sources: ACS 2000-2021; US Census 1970, 1980, 1990.
Note: CADR = Central America and the Dominican Republic. Note: CADR = Central America and the Dominican Republic.
CADR migrants shifted into jobs intensive in nonroutine manual tasks while non-CADR migrants
shifted into higher-skilled knowledge jobs. Employment in construction occupations, which are intensive
in nonroutine manual tasks, increased sharply among CADR migrants and is now the top occupational
group by employment share (figure 3.10). In contrast, there has been little change in employment in
construction among non-CADR migrants. Employment also increased for CADR migrants in several other
occupations intensive in nonroutine manual tasks including food preparation and serving, building and
grounds cleaning, and transportation and material moving. Non-CADR migrants generally saw more
26 These occupations have made up 5 percent or more of employment at least once during this period. These are food preparation and serving,
building and grounds cleaning, sales, office and administrative support, construction, production, and transportation and material moving.
27 These include management, computer and mathematics, and education jobs.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 31
muted employment increases in these occupations while experiencing employment growth in higher-
skilled occupations as well, including in management, computers and mathematics, and education.
CADR migrants contributed to the expansion in low-skilled employment observed in the United States
in the last several decades—one side of employment polarization in the United States—but did not
contribute to the expansion in high-skilled employment—the other side of polarization. CADR migrants
have shifted strongly from middle- to low-skilled jobs, unlike the pattern observed for migrants from
other middle- and high-income countries and for nonmigrant US workers (figure 3.11a–e). In fact, CADR
migrants and migrants from lower-income countries are responsible for much of the increase in low-skilled
employment observed in the United States in recent decades. This is consistent with recent research
showing that offshoring and automation in the United States has increased demand for low-skilled
services that are intensive in nonroutine manual tasks (Mandelman and Zlate 2022). Migrants from other
middle- and high-income countries, on the other hand, have benefited from the same complementarity
between technology and skills that nonmigrant United States workers have enjoyed. The importance
of CADR migrants and of migrants from low-income countries is such that the picture of routine-biased
technological change becomes much murkier for nonmigrant US workers (figure 3.11e).
FIGURE 3.11: Change in Employment Share by Skill Level of Three-Year Migrants and United
States Nonmigrants in the United States, 1980–2021
Percentage points
a. CADR b. LICs c. MICs
15 15 10
8
10
10 6
5 4
5
2
0
0
0
-5 -2
-5 -4
-10
-6
-10 -15 -8
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
d. HICs e. United States
15 2.5
2
10 1.5
1
5 0.5
0
0 -0.5
-1
-5 -1.5
-2
-10 -2.5
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Sources: ACS 2000-2021; US Census 1970, 1980, 1990.
Note: Employment shares are calculated for each occupation and year and then ordered in 1 percent bins by mean occupational wage. These bins are then
summed to 10 percent bins. CADR = Central America and the Dominican Republic; HICs = high-income countries; LICs = lower-income countries; MICs = middle-
income countries.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 32
Overall, employment of CADR migrants is evolving away from routine tasks and becoming somewhat
more intensive in the flexible hands-on and people-oriented work common in construction and
services. The task-based measures of routine and nonroutine employment described earlier in the
chapter provide a summary picture of the changing composition of the employment of CADR workers in
the United States. Since 1970, the employment of nonmigrant US workers and migrants from middle- and
high-income countries has become more intensive in nonroutine interpersonal and nonroutine analytical
tasks (knowledge and people-oriented work) and less intensive in routine cognitive and manual and
nonroutine manual tasks (administrative, repetitive hands-on, and flexible hands-on work) (figure 3.12c–
e). The employment of migrants from CADR countries (and from low-income countries), in contrast, has
become more intensive in nonroutine manual tasks (flexible hands-on work) (figure 3.12a–b). In sum,
technological progress in the United States seems to be pushing CADR workers towards less-skilled
services and construction sector jobs where nonroutine manual and interpersonal tasks dominate.
FIGURE 3.12: Evolution of the Task Content of Three-Year Migrants in the United States,
1970–2021
Task index (1970 = 0)
a. CADR b. LICs c. MICs
0.40 0.60 0.60
0.20 0.40 0.40
0.00 0.20 0.20
-0.20 0.00 0.00
-0.40 -0.20 -0.20
-0.60 -0.40 -0.40
-0.80 -0.60
-0.60
-1.00 -0.80 -0.80
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
d. HICs e. United States
1.00 0.50
0.40
0.50 0.30
NR analytical
0.20
0.00 0.10 NR interpersonal
0.00 R cognitive
-0.50 -0.10
R manual
-0.20
-1.00 -0.30 NR manual
-0.40
-1.50 -0.50
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
Sources: ACS 2000-2021; US Census 1970, 1980, 1990.
Note: CADR = Central America and the Dominican Republic; HICs = high-income countries; LICs = lower-income countries; MICs = middle-income countries;
NR = nonroutine; R = routine.
Patterns are similar across gender and country of origin with a few notable exceptions. The increase in
nonroutine manual tasks is only apparent for male migrants from CADR: women have only experienced an
increase in the intensity of employment in nonroutine interpersonal tasks (figure 3.13a–b). This is consistent
with the growth in construction employment among CADR workers being heavily biased towards men.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 33
Focusing on the four CADR countries with the most migrants in the United States, employment has become
more intensive in nonroutine interpersonal tasks among migrants from all of them and in nonroutine
manual tasks among migrants from all except the Dominican Republic (figure 3.13c–f). Nonroutine
analytical skills have become more important for migrants from El Salvador, while routine manual skills
have become more important for migrants from Guatemala. In the case of El Salvador, these changes
relate to growth in occupations like education and training, management, and business, and a related
increase in migrants with tertiary education. Guatemala shows nearly the opposite picture: employment
increased in construction occupations while the educational level of migrants from Guatemala declined.
FIGURE 3.13: Evolution of the Task Content of Three-Year CADR Migrants by Gender in the
United States, 1970–2021
Task index (1970 = 0)
a. Women b. Men c. Dominican Republic
0.4 0.6 0.6
0.2 0.4 0.4
0.2 0.2
0
0
-0.2 0
-0.2
-0.4 -0.2
-0.4
-0.6 -0.4
-0.6
-0.8 -0.6 -0.8
-1 -0.8 -1
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
d. El Salvador e. Guatemala f. Honduras
0.6 0.8 0.8
0.6 0.6
0.4
0.4 0.4
0.2
0.2
0.2
0 0
0
-0.2
-0.2
-0.2
-0.4
-0.4 -0.4 -0.6
-0.6 -0.6 -0.8
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
1970
1990
2001
2003
2005
2007
2009
2011
2013
2015
2017
2019
2021
NR analytical NR interpersonal R cognitive R manual NR manual
Sources: ACS 2000-2021; US Census 1970, 1980, 1990.
Note: NR = nonroutine; R = routine.
Wages of CADR migrants have barely increased since 1970. The growing predominance of CADR
migrants in low-skilled (low-paid) occupations is apparent in the lack of wage growth of this group (figure
3.14). Real wages have grown 12 percent for CADR migrants since 1970 and only 3 percent since 1990.
This contrasts with growth of 62 percent, 55 percent, and 41 percent for migrants from low-, middle-, and
high-income countries, respectively. This suggests that on one hand, technological progress in the United
States has created additional employment opportunities for CADR migrants. On the other hand, these
opportunities are in less-skilled jobs, and the growing supply of comparable low-skilled CADR workers
may have increased competition, depressing wage growth.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 34
FIGURE 3.14: Average Hourly Wage of Three-Year Migrants in the United States, 1980–2021
Real 2010 US$
35
30
25
20
15
10
5
0
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
00
20 1
02
03
04
05
06
07
08
09
10
11
12
13
14
15
16
17
18
19
20
21
0
20
20
20
20
19
20
20
20
20
19
20
20
20
20
19
19
20
19
19
19
19
19
19
20
19
19
19
19
19
19
19
19
20
19
20
20
20
19
20
20
20
CADR LICs MICs HICs
Sources: ACS 2000-2021; US Census 1970, 1980, 1990.
Note: CADR = Central America and the Dominican Republic; HICs = high-income countries; LICs = lower-income countries; MICs = middle-income countries.
Changes in Tasks: Artificial Intelligence and Mobile Robotics
Estimates of the susceptibility of employment to the latest technological developments suggest that
a substantial share of jobs is at risk of disruption. The second wave of digitally enabled automation
described in chapter 2 involves AI and robots that are capable of doing both routine and nonroutine tasks.
Because these technologies are nascent, research in this area tends to involve prospective measures of
what AI and robots could do. These measures are calculated in a similar manner to those used to identify
the routine and nonroutine tasks suitable for computerization. Tasks are identified that AI and robots are
unlikely to be able to carry out, enabling an estimation of the share of workers doing tasks that are likely to
be automated. Using this approach, between 9 and 36 percent of workers are at high risk of automation
in CADR countries (figure 3.15).28 However, nearly all workers in all CADR countries are at least at medium
risk, meaning that some tasks they undertake are likely to disappear or change substantially. This contrasts
with the OECD where 14 percent of jobs are at high risk and 46 percent are at least at medium risk.
FIGURE 3.15: Susceptibility of the Workforce to Automation, 2021
Percentage of employment
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
GTM NIC HND SLV PAN CRI DOM
Low Medium High
Sources: Frey and Osborne 2017; SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The years are 2019 for Guatemala and Honduras; 2018 for Panama; and 2014 for Nicaragua. The risk of automation is calculated according to the
task-based approach in Arntz, Gregory, and Zierahn (2016) and Egana del Sol et al. (2022). Low risk is a probability of automation that is 30 percent or less;
medium is above 30 but below 70 percent; and high is 70 percent or above. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras;
NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
28 See appendix C for a detailed description of the methodology to calculate the probability of automation.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 35
Male, young, less-educated, and rural workers have a higher risk of automation. In all CADR countries,
the occupations where women, adult, and urban workers are employed and the tasks they perform at
work result in a lower average risk of automation compared to men, younger, and rural workers (figure
3.16a–c). The average risk of automation declines with education and is lower for wage employees than
self-employed workers (figure 3.16d–e). The services sector has the lowest risk of automation among
economic sectors while agriculture has the highest (figure 3.16f).
FIGURE 3.16: Probability of Automation by Sociodemographic Characteristics, 2021
Percentage average probability
a. Gender b. Age c. Geography
80% 80% 80%
70% 70% 70%
60% 60% 60%
50% 50% 50%
40% 40% 40%
30% 30% 30%
20% 20% 20%
10% 10% 10%
0% 0% 0%
CRI DOM GTM HND NIC PAN SLV CRI DOM GTM HND NIC PAN SLV CRI DOM GTM HND NIC PAN SLV
Women Men 15-24 25-65 Urban Rural
d. Education e. Employment type f. Sector
80% 80% 90%
70% 70% 80%
70%
60% 60%
60%
50% 50%
50%
40% 40%
40%
30% 30%
30%
20% 20% 20%
10% 10% 10%
0% 0% 0%
CRI DOM GTM HND NIC PAN SLV CRI DOM GTM HND NIC PAN SLV DOM GTM HND NIC PAN SLV
Low Medium High Wage employee Self-employed Agriculture Industry Services
Sources: Frey and Osborne 2017; SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The years are 2019 for Guatemala and Honduras; 2018 for Panama; and 2014 for Nicaragua. The risk of automation is calculated according to the task-
based approach in Arntz, Gregory, and Zierahn (2016) and Egana del Sol et al. (2022). Low risk is a probability of automation that is 30 percent or less; medium
is above 30 but below 70 percent; and high is 70 percent or above. Low education is less than 9 years of school, medium is between 9 and 13 years, and high is
14 or more. Data on sector is not available for Costa Rica. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua;
PAN = Panama; SLV = El Salvador.
However, the susceptibility of employment to automation in CADR countries is much less dire when
small corrections are made for factors that might interrupt the translation of potential labor market
impacts of technological progress into actual impacts. As described in chapter 1, CADR countries are
characterized by high rates of informal employment and high rates of self-employment. These large
informal and self-employed workforces reflect private sectors that have less mature organizational
structures. These sectors also tend to be less capital-intensive and so less affected by technological
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 36
advancement (Weller, Gontero, and Campbell 2019). When workers in these sectors are excluded from
measures of the probability of automation, exposure to displacement by automation technologies
declines substantially (figure 3.17).29 The share of workers at low risk of automation rises from single digits
in the original measure to about half in all countries.
FIGURE 3.17: Share of Jobs at High Risk of Automation, Original and Adjusted Measures, 2021
Percentage
100%
80%
60%
40%
20%
0%
Original Adjusted Original Adjusted Original Adjusted Original Adjusted Original Adjusted Original Adjusted Original Adjusted
CRI DOM GTM HND NIC PAN SLV
Low Medium High
Sources: Frey and Osborne 2017; SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The years are 2019 for Guatemala and Honduras; 2018 for Panama; and 2014 for Nicaragua. The risk of automation is calculated according to the
task-based approach in Arntz, Gregory, and Zierahn (2016) and Egana del Sol et al. (2022) and then adjusted following Weller et al (2019). Low risk is a
probability of automation that is 30 percent or less; medium is above 30 but below 70 percent; and high is 70 percent or above. The adjustment assigns a risk
of automation equal to zero to workers in low-productivity sectors defined as self-employed workers with less than college education, wage employees and
employers in small firms, domestic workers, and workers who do not receive a labor income. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Artificial Intelligence, Mobile Robotics, and CADR Workers in the United States
CADR migrants in the United States are at greater risk of automatability than nonmigrants and
migrants from other country groups, but at much lower risk from the latest advances in artificial
intelligence. The average probability of automation is around 70 percent for migrants from CADR
countries, the highest among all migrant groups and substantially higher than the risk of US nonmigrant
workers (figure 3.18).30 This suggests that CADR migrants will likely continue experiencing labor market
disruptions of the kind they have experienced in the last several decades in the United States. However,
CADR migrants are unlikely to be as affected by the latest advances in AI. The exposure of CADR migrants
to GPTs like ChatGPT is lower than that of other migrants and of nonmigrants. Unlike automatability in
general, the share of the median CADR worker’s tasks that are exposed to GPTs is relatively low—16
percent for CADR migrants versus 44 percent for US workers—as is the share of workers in occupations
with at least 10 or 50 percent of their tasks exposed to GPTs (figure 3.19). The low-skill bias of CADR
migrants is likely to mean more labor market disruptions from certain AI and mobile robotics, but relatively
less from GPTs.
29 See appendix C for a detailed description of the methodology to calculate this alternative measure of the probability of automation.
30 The probability of automation calculated here is the occupation-based measure proposed in Frey and Osborne (2017). The task-based
methodology for estimating automation probabilities cannot be used because of the small sample sizes of migrants in the Program for the
International Assessment of Adult Competencies (PIAAC) data that is typically used for this analysis.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 37
FIGURE 3.18: Average Probability of FIGURE 3.19: Share of Workers with at Least
Automation, 2021 10 and 50 Percent of Tasks Exposed to
GPTs, 2021
Percentage average probability Percentage
80% 100%
90%
70%
80%
60% 70%
50% 60%
50%
40%
40%
30% 30%
20% 20%
10%
10% 0%
HICs USA MICs LICs CADR
0%
CADR LICs MICs USA HICs 10% of tasks 50% of tasks
Sources: ACS 2021; Frey and Osborne 2017. Sources: ACS 2021; Eloundou et al. 2023.
Note: CADR = Central America and the Dominican Republic; HICs = high- Note: CADR = Central America and the Dominican Republic; HICs = high-
income countries; LICs = lower-income countries; MICs = middle-income income countries; LICs = lower-income countries; MICs = middle-income
countries; USA = United States of America. countries; USA = United States of America.
CHANGES IN WORKING ARRANGEMENTS
ICT is changing how firms organize workers. Workers within a firm are increasingly able to do their jobs
at a distance (remote work). At the same time, firms are increasingly able to look to workers outside of the
firm to do tasks at a distance (platform work).
Changes in Working Arrangements: Remote Work
Remote work creates both challenges and opportunities for labor markets. Studies show that remote
work can be associated with higher productivity (Bloom et al. 2015; Choudhury, Foroughi, and Larson
2020; Mas and Pallais 2020). However, productivity may decline due to loss of information and knowledge
spillovers from face-to-face interactions, increased interruptions, challenges accomplishing certain tasks
from home, and poor telecommunications (Atkin, Schoar, and Shinde 2023; Behrens, Kichko, and Thisse
2021; Gibbs, Mengel, and Siemroth 2023). For workers, one of the primary benefits of remote work is the
flexibility it offers. Indeed, workers tend to value the flexibility of working from home in particular, relative
to other forms of job flexibility (for example, in work schedules) (Mas and Pallais 2017). This flexibility also
seems to have cushioned many workers from the worst labor market impacts of the pandemic. Overall,
jobs with more potential to be done remotely were associated with less job and income loss (Garrote
Sanchez et al. 2021).31 On the other hand, less-educated and lower-income workers are less likely to work
in jobs that can be done remotely, meaning these benefits are generally available only to those who are
already better off (Garrote Sanchez et al. 2021; Mongey, Pilossoph, and Wingerg 2020).
The potential for remote work is low in CADR countries. Estimates of remote work potential can be
created by assessing whether a worker’s tasks can be done at home (Dingel and Neiman 2020). Tasks
like use of email increase the likelihood that a job can be done at home while tasks like working outdoors
decrease the likelihood.32 Remote work potential in CADR countries is calculated using task data from
Program for the International Assessment of Adult Competencies (PIAAC) surveys and the United States
as a benchmark.33 In all CADR countries, jobs are less amenable to working from home than the average
31 See also Adams-Prassl et al. (2020), Mongey, Pilossoph, and Wingerg (2020), and Montenovo et al. (2022) for the United States; Guven,
Sotirakopoulos, and Ulker (2020) for Australia; Hatayama, Viollaz, and Winkler (2023) for Chile.
32 Other approaches are also possible, such as surveying workers about the share of tasks that can be done at home. See, for instance, Adams-
Prassl et al. (2022).
33 See appendix D for a detailed description of the methodology to calculate remote work potential.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 38
job in the United States.34 While in the United States around 40 percent of work could potentially be done
from home, rates are less than 30 percent in all CADR countries except Panama and Costa Rica (figure
3.20). Amenability to working from home in the region generally decreases with income level consistent
with the international literature (Gottlieb, Grobovšek, and Poschke 2020). In all countries except El
Salvador, women have better chances of working from home than men.
FIGURE 3.20: Share of Workers in Jobs with High Amenability to Working from Home, 2021
Percentage
60%
50%
40%
30%
20%
10%
0%
USA PAN CRI DOM SLV HND NIC GTM
All Women Men
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The years are 2019 for Guatemala and Honduras, 2014 for Nicaragua, and 2018 for Panama. CRI = Costa Rica; DOM = Dominican Republic;
GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador; USA = United States of America.
Individuals with jobs that can be done remotely in CADR countries tend to be better off. Work-from-home
potential increases with education. High-skilled occupations—managers, professionals, and technicians—
have higher work-from-home potential than lower-skilled occupations such as skilled agricultural workers,
plant and machine operators and assemblers, and crafts and related trades workers. This is consistent
with existing literature showing that in both developed and developing countries, workers who can work
from home tend to be less economically vulnerable. Across countries, these workers tend to be more
highly educated and better paid and to be wage, formal, permanent, and urban workers.35
Prior to the COVID-19 pandemic, actual rates of working from home were very low in CADR countries
among wage workers but higher among self-employed workers. Data from household and labor force
surveys show that before the pandemic 7 percent of workers worked from home in Guatemala (2019),
8 percent in Costa Rica (2019), 9 percent in Panama (2019), and 15 percent in Nicaragua (2012), the four
countries where data is available on both wage and self-employed workers. The vast majority of these
were self-employed workers: rates of working from home among wage workers were 2 percent or less in
all three countries with data available (figure 3.21). Rates were 20 percent or more in all CADR countries
with data on self-employed workers (figure 3.22). This pattern is similar to that observed in the rest of the
world: the ILO estimates that before the pandemic 8 percent of the global workforce worked from home,
most in self-employment (ILO 2021a).36 Across all CADR countries and both employment types, women
are more likely than men to work remotely.
34 These results are similar to those of other like studies. See appendix D for a description of these results.
35 See also Dingel and Neiman (2020) and Mongey, Pilossoph, and Wingerg (2020) for the United States; Adams-Prassl et al. (2022) for the United
States and the United Kingdom; and Lekfuangfu et al. (2020) for Thailand. Regarding age and gender, the evidence is not conclusive and
depends on the country analyzed. Young workers in Mexico and India are less likely to work from home than adult workers (Garrote Sanchez
2021), but the opposite relationship has been reported for a set of 10 low- and middle-income countries (Hatayama, Viollaz, and Winkler 2023).
Women have more possibilities of working from home in the same set of 10 low- and middle-income counties (Gottlieb et al. 2021; Hatayama,
Viollaz, and Winkler 2023), but there is no gender difference for a set of 11 large Latin American cities (Berniell and Fernandez 2021).
36 In the United States, around 6 percent of workers worked primarily from home in 2019 (Mas and Pallais 2020; Oettinger 2011). Fifteen percent
of workers in the European Union reported working from home sometimes or usually in 2018 (Alipour, Falck, and Schüller 2020). Data from
five LAC countries show that between 5 and 8 percent of workers worked from home just before the pandemic (Maurizio 2021).
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 39
FIGURE 3.21: Share of Wage Workers FIGURE 3.22: Share of Self-Employed
Working from Home Prior to the Working from Home Prior to the
Pandemic, 2019 Pandemic, 2019
Percentage Percentage
4.5% 70%
4.0%
60%
3.5%
3.0% 50%
2.5% 40%
2.0%
30%
1.5%
20%
1.0%
0.5% 10%
0.0% 0%
All
Female
Male
All
Female
Male
All
Female
Male
All
Female
Male
All
Female
Male
All
Female
Male
All
Female
Male
All
Female
Male
All
Female
Male
All
Female
Male
CRI GTM NIC PAN CRI SLV GTM HND NIC PAN
Sources: SEDLAC (CEDLAS and The World Bank). Sources: SEDLAC (CEDLAS and The World Bank).
Note: The years are 2012 for Nicaragua and 2018 for Panama. Data is Note: The years are 2012 for Nicaragua and 2018 for Panama. Data is not
unavailable for the other CADR countries. CRI = Costa Rica; GTM = Guatemala; available for the Dominican Republic. CRI = Costa Rica; GTM = Guatemala;
NIC = Nicaragua; PAN = Panama. HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
The higher rates of remote work among self-employed workers are evidence of a less-developed
organization of work, rather than a shift enabled by improvements in ICT. Self-employed workers
working at home in less-developed countries tend to be artisans, self-employed business owners, or
industrial home-based workers like seamsters (ILO 2021a). The higher actual work-from-home rates of
self-employed workers in CADR relative to wage workers contrast with the potential measure of working
from home, which predicts higher rates among wage workers because of their tendency to do office work
that can be done from home when facilitated by ICT.
Household and labor force survey data show that the COVID-19 pandemic led to a spike in remote
work in CADR’s more developed countries. The ILO estimates that during the pandemic, working from
home increased from around 8 percent of the global workforce to 17 percent, ranging from 14 percent of
workers in low- and lower-middle-countries to 25 percent in high-income countries (ILO 2021a). A similar
increase is observable in Costa Rica: the percentage of workers working from home increased from 8
percent in 2019 to 13 percent in 2020 and 18 percent in 2021.37 The rate jumped from 2 percent in 2019 to
14 percent in 2021 for wage workers but only from 28 percent to 30 percent for self-employed workers.
The increases were highest among the most educated. Panama experienced a small increase in remote
work of 2 percentage points between 2019 and 2021. Notably, the increase was larger for self-employed
workers (2.2 percentage points) than for wage workers (0.4 percentage points), though workers with
higher levels of education experienced all of the increase. In El Salvador, the only other CADR country
with prepandemic and postpandemic data, work-from-home rates did not change during the pandemic,
though data is only available for self-employed workers.
COVID-19-related surveys provide evidence of a regionwide increase in remote work during the
pandemic, including among wage workers. The World Bank and United Nations Development
Programme’s High-Frequency Phone Surveys show higher rates of working from home during the
pandemic relative to the rates observed from other data sources prior to the pandemic (figure 3.23). The
share of workers working at least half of their total hours from home between May and July of 2021 ranged
37 Gottlieb et al. (2021) estimates that 10.8 percent of urban workers worked from home in Costa Rica in the second quarter of 2020. Maurizio
(2021) reports an increase from 8 to 22 percent between the pre-pandemic level and the peak in mid-2020.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 40
from 11 to 21 percent versus a range of 7 to 15 percent prior to the pandemic.38 In contrast to what was
found prior to the pandemic, work-from-home rates were higher for wage workers than self-employed
workers in Costa Rica, the Dominican Republic, Guatemala, Honduras, and Panama. This provides some
evidence that the pandemic resulted in an increase in remote work among wage workers, though this
is speculative without comparable data from before the pandemic. Working from home arrangements
during the pandemic were more common for women.
FIGURE 3.23: Share of Workers Working FIGURE 3.24: Share of Workers Working
from Home During the Pandemic, 2021 from Home During the Pandemic, 2020
Percentage Percentage
25% 16%
14%
20%
12%
15% 10%
8%
10% 6%
4%
5%
2%
0% 0%
NIC SLV CRI DOM HND GTM PAN SLV GTM HND NIC
All Female Male Wage employement Self-employed June 2020 November 2020
Source: High-Frequency Phone Surveys 2021. Source: COVID-19 Business Pulse Surveys 2020.
Note: CRI = Costa Rica; DOM = Dominican Republic; NIC = Nicaragua; Note: GTM = Guatemala; HND = Honduras; NIC = Nicaragua; SLV = El
GTM = Guatemala; HND = Honduras; PAN = Panama; SLV = El Salvador. Salvador.
The persistence of the COVID-19–related spike in remote work is uncertain but preliminary evidence
suggests it was temporary. It is not yet possible to examine the persistence of these trends for long after
the pandemic. In developed countries, there is some evidence of an immediate decline in remote work
after the pandemic followed by stabilization at a higher level than prepandemic. In the United States,
for example, remote work rates increased from 8 percent prior to the pandemic to a high of around 60
percent during the pandemic before stabilizing at around 30 percent of paid full days in 2022 (Barrero,
Bloom, and Davis 2021; Dalton and Groen 2022). In CADR countries, there are signs that the spike will not
persist. Firm-level data from the World Bank’s COVID-19 Business Pulse Surveys show declines in work-
from-home rates among employees between June and November of 2020 during the pandemic (figure
3.24). In the Dominican Republic, a question about telework added to the labor force survey in 2020
shows that rates declined from 5.0 percent of workers in 2020 to 2.4 percent in 2021 and 1.7 percent in
2022.
Changes in Working Arrangements: Platform Work
Similar to remote work, platform work involves a mix of opportunities and challenges (table 3.1).
Workers can combine platform work with other paid and unpaid work. People with restricted mobility can
access online jobs and people living in areas with limited economic opportunities can access a larger,
even global, market for jobs (Datta and Chen 2023). Barriers to entry can be low with only limited assets
and skills required for certain types of platform jobs. At the same time, workers spend a significant amount
of (unpaid) time looking for work, taking qualification tests, and writing reviews, and lack the certainty in
38 The High-Frequency Phone Surveys asks how many of a worker’s hours were worked “remotely or virtually” while the labor force and
household surveys ask about the location of work or where a job is generally done.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 41
earnings that workers often value (Mas and Pallais 2021). Because many platforms operate internationally,
competition for jobs is strong and platform workers can be exposed to international shocks. Even the
limited skills requirements can be prohibitive for some people: most platform jobs require a minimum
level of digital skills, and many require knowledge of English. The legal status of platform work is often
disputed, and platform workers generally lack access to social protection, workplace protections, and
training. At the same time, firms see significant advantages in the ability to organize tasks flexibly, scale
their workforces up and down more easily in response to shocks, and avoid the costs associated with
permanent workers. However, they also face challenges in balancing flexibility and the benefits of
collaboration-based innovation that is harder to generate when tasks are atomized.
TABLE 3.1: Advantages and Disadvantages of Platform Work for Workers and Firms
Advantages Disadvantages
Workers Flexibility in time and location of work Insufficient and unpaid work
• Workers can combine platform work with • Looking for tasks, taking qualification tests,
unpaid and other paid work. researching clients, writing reviews can take
Barriers to entry may be lower one-third of working time.
• Formal qualifications and certifications may Volatile and low earnings
not be needed. • Monopsony and powerful intermediaries in
two-sided market
Lack of benefits of standard employment
• Social protection, workplace protections,
training limited
Algorithmic management
• Workers subject to client approval and
reputation scores with limited recourse
Firms Flexibility in task organization Challenges developing firm-specific human
• Can organize workers by project or task capital
Rapid shock response • Fewer opportunities for the development of
• Ability to scale workforce up and down firm-specific skills
• Ability to manage uncertainty
Cost savings
• Avoid costs of permanent workers
Sources: ILO 2021b; Oyer 2020; Wood 2019.
The same improvements in ICT that are facilitating remote work are also contributing to the rise of
platform work. Platform work connects workers and consumers via online platforms for work performed
either online and remotely (for example, image tagging through services like Amazon Turk or freelancing
through Upwork) or offline at a physical location (for example, transportation or delivery services through
Uber or Deliveroo). The work is typically paid on a piece-rate basis. Online work is less established in
LAC than location-based work, though it has grown more prevalent in recent years including during the
COVID-19 pandemic when downloads of online platform work apps increased in many countries (IDB
2021b). The number of platforms globally increased from the 10s in the 2000s to nearly 800 in 2020 with
at least 15 based in LAC (ILO 2021b).
Location-based platforms have become common in certain sectors in CADR, though the number of
location-based platform workers is uncertain. Globally, location-based platform workers typically work
across five sectors: transportation, delivery, home services, domestic work, and care services (ILO 2021b).
Only two of these sectors, transportation and delivery, are common in CADR (table 3.2). Platforms in
these sectors include both global ones like Uber (United States) and locally founded ones like Urban
(Guatemala), Delivery RD (Dominican Republic, later acquired by PedidosYa), and Sampopo (Honduras).
The industry in the region is characterized by frequent entry and exit of new domestic and international
firms. Successful domestic firms are also often acquired by international ones. The large international
platforms like Uber and Didi tend to dominate but are typically only present in major cities. Estimates of
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 42
the number of location-based platform workers in CADR countries, as in most countries, are generally
unavailable. One estimate for the Dominican Republic suggests that 0.2 percent of the labor force works
as a location-based gig worker (Garcia and Javier 2020).
TABLE 3.2: Primary Location-Based Platforms in CADR Countries
Country Location-based taxi Location-based food delivery
CRI Uber, Didi, InDrive Uber, UberEats, PedidosYa, DidiFood, Rappi
DOM Uber, Didi, InDrive Uber Eats, Indriver, PedidosYa, Delivery RD (PedidosYa), Komida!
GTM Uber, InDrive Urban, Picap PedidosYa, UberEats
HND InDrive, Uber, Bolt PedidosYa, Sampopo
NIC Picap UberEats, PedidosYa
PAN Uber, Didi, InDrive, Wageen UberEats, PedidosYa
SLV Uber, InDrive, Bolt PedidosYa, UberEats
Sources: Garcia and Javier 2020; Sensor Tower 2023.
The demand for online gig work has grown globally. Employers increasingly look abroad for workers
because of challenges finding skills locally, high labor costs, or lack of room and equipment (Agrawal et
al. 2015). Much of the demand for gig workers comes from developed countries and is concentrated on
several large platforms. Infrastructure is in place to support firms outsourcing tasks rather than jobs. For
instance, ModSquad manages contract workers around the world who provide digital engagement services
to firms worldwide.39 However, local and regional platforms have emerged in developing countries typically
in response to demand for workers with specific linguistic or cultural knowledge (Datta and Chen 2023). The
Online Labor Index (OLI) compiled by Oxford University and the ILO measures tasks posted on the largest
online labor platforms. The OLI shows that demand, proxied by task postings, increased about 50 percent
between mid-2016 and mid-2023 (figure 3.25). The United States represents one-third of global demand;
India, the United Kingdom, and the United States represent half. Notably, online gig work is sourced globally,
meaning workers in CADR are able to access this growing demand. While demand in CADR is generally much
lower as a share of the global market, local platforms do exist, including El Salvador–based SoyFreelancer.40
FIGURE 3.25: Global Demand for Online Gig Work, 2016–23
OLI, 2016 = 100
200
180
160
140
120
100
80
60
40
20
0
01/06/16
01/08/16
01/10/16
01/12/16
01/02/17
01/04/17
01/06/17
01/08/17
01/10/17
01/12/17
01/02/18
01/04/18
01/06/18
01/08/18
01/10/18
01/12/18
01/02/19
01/04/19
01/06/19
01/08/19
01/10/19
01/12/19
01/02/20
01/04/20
01/06/20
01/08/20
01/10/20
01/12/20
01/02/21
01/04/21
01/06/21
01/08/21
01/10/21
01/12/21
01/02/22
01/04/22
01/06/22
01/08/22
01/10/22
01/12/22
01/02/23
01/04/23
01/06/23
Source: Online Labor Index 2023.
39 See “How Technology is Redrawing the Boundaries of the Firm” in The Economist, January 8, 2023.
40 The OLI shows that demand for online gig work is low in CADR: Panama had the largest share of the global market at .08 percent, followed
by the Dominican Republic and Costa Rica at .05 percent each.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 43
Workers from all CADR countries participate in online gig work, though they represent a small share of
the global online gig workforce. A survey undertaken for a recent World Bank report on online gig work
shows that around one-quarter of the labor force in Costa Rica participates in online gig work (figure 3.26).
Between 10 and 20 percent of the labor force does in the Dominican Republic, Guatemala, and Panama.
Six percent or less does in El Salvador, Honduras, and Nicaragua. Workers in different CADR countries
specialize in different types of online gig work (table 3.3). Based on a measure of the workers active on
the largest online labor platforms, the OLI shows that Costa Rica again stands out. Online gig workers in
Costa Rica are significant providers of online sales and marketing tasks, representing 1.5 percent of global
workers, making it the seventh-largest supplier globally.
FIGURE 3.26: Online Gig Workers, 2022 TABLE 3.3: Most Common Type of Online
Percentage of labor force that participates in gig work Gig Tasks in CADR Countries, 2017–23
activities
30%
Country Occupation
25%
CRI Sales and marketing support
20% DOM Creative and multimedia
15% GTM Sales and marketing support
HND Creative and multimedia
10%
NIC Creative and multimedia
5%
PAN Software development and technology
0%
CRI ARG CHL COL GTM MEX BRA PAN DOM SLV NIC HND SLV Software development and technology
Source: Datta and Chen 2023. Source: Online Labor Index 2023.
Note: ARG = Argentina; BRA = Brazil; CHL = Chile; CRI = Costa Rica; COL = Colombia; Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; MEX = Mexico; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Location- and online-based platform workers tend to be younger, male, and more highly educated. Platform
workers are a heterogenous group, but global research and research on LAC including in several CADR
countries shows several commonalities (Azuara, González, and Keller 2019; ECLAC and ILO 2021b; Fernandez
and Benavides 2020; Garcia and Javier 2020; ILO 2021b; Madariaga et al. 2019). Platform workers skew young
and male, particularly in developing countries. They also tend to live in urban areas and to be more highly
educated, again particularly in developing countries. Many platform workers undertake gig work as a second
job. Online-based platform workers tend to be more highly skilled than location-based platform workers.
However, there are differences in these characteristics across CADR countries. A recent World Bank
survey of Workana freelancers generally supports these findings but highlights differences across
countries. For example, women represent two-thirds of freelancers in the Dominican Republic, the
highest share of any country surveyed, and more than half in Nicaragua (figure 3.27a). The majority
are younger than 40 in all countries surveyed, but younger workers between the ages of 15 and 29 are
more common in the Dominican Republic and Guatemala (figure 3.27b). Finally, online-based platform
work is much less common outside capital cities in the four CADR countries surveyed (figure 3.27c).
Both online- and location-based platform workers tend to lack access to social protection and other
workplace protections. A recent World Bank survey of freelancers on El Salvador’s SoyFreelancer found
that only one-third contributed to a government retirement savings scheme. Security is a particular
concern for location-based women platform workers (and women customers). Some local firms such as
Urban from Guatemala offer a separate platform available only to women drivers and riders to address
this concern.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 44
FIGURE 3.27: Characteristics of Workana Freelancers, 2022
Percentage of freelancers
a. Percentage female b. Percentage ages 15 to 29 c. Percentage outside capital city
70% 45% 90%
40% 80%
60%
35% 70%
50%
30% 60%
40% 25% 50%
30% 20% 40%
15% 30%
20%
10% 20%
10% 5% 10%
0% 0% 0%
DOM
NIC
URY
ARG
COL
PAN
ECU
VEN
PER
GTM
BRA
MEX
CHL
BOL
DOM
BRA
COL
GTM
ECU
VEN
URY
CHL
PER
MEX
ARG
PAN
NIC
BOL
BOL
MEX
ARG
VEN
ECU
BRA
COL
NIC
URY
CHL
PER
DOM
GTM
PAN
Source: Datta and Chen 2023.
Note: ARG = Argentina; BOL = Bolivia; BRA is Brazil; CHL = Chile; COL = Colombia; DOM = Dominican Republic; ECU = Ecuador; GTM = Guatemala; MEX is
Mexico; NIC = Nicaragua; PAN = Panama; PER = Peru; URY = Uruguay; VEN = Venezuela.
Changing Working Arrangements and CADR Workers in the United States
Platform work is attractive to migrants. Low entry barriers and limited job alternatives can make platform
work attractive to migrants, who may lack the necessary work documentation and start-up capital for self-
employment activities. Recent ILO research shows that 17 percent of online-based platform workers are
migrants with a higher proportion in developed countries, while 15 percent of app-based delivery workers
and 1 percent of app-based taxi workers are migrants (ILO 2021b).
Migrants from LAC, including most CADR countries, make up about 10 percent of platform workers
in the United States. The May 2017 supplement of the United States Current Population Survey includes
questions about platform work that provide some insight into the prevalence of platform work among
CADR migrants. About one-quarter of the 1.6 million platform workers in the United States in 2017 were
migrants. Nearly 150,000 (10 percent) were from LAC. The number of observations for workers from CADR
is very small, so any inferences must be drawn with substantial care, but the survey suggests that around
10 percent of all migrant platform workers were from CADR, lower than CADR’s share of total migrants in
the United States.
Chapter 3 Technological Progress and Labor Market Transformation in CADR Countries 45
Chapter 4
Barriers to and Enablers of the
Future of Work in CADR Countries
Chapter 3 revealed a puzzle: the technology to automate tasks and to enable remote and platform
work exists, but CADR’s transition to a future of work defined by nonroutine tasks and altered working
arrangements is only partial. To try to solve this puzzle, this chapter considers the factors beyond
technological potential that underlie employment trends in the region. Technology use is found to explain
only a small portion of the changes in tasks in CADR countries. Employment structure and skills play a
much more important role. The prevalence of services sector employment across CADR countries implies
less vulnerability to automation, but at the expense of the dominance of low-productivity employment.
Education levels have improved, but the lack of tertiary-educated workers and problems of educational
quality interrupt the potential complementarities between workers and technology. CADR’s transition to
the future of work is also being affected by technological progress abroad. The rise of global platforms
creates an opportunity to connect CADR services employment to global markets, but robot adoption
abroad could reduce employment opportunities for workers in CADR countries by reducing offshoring
and for CADR workers in other countries by reducing the demand for migrants.
SETTING THE STAGE: LOOKING BEYOND TECHNOLOGICAL POTENTIAL
The transition of CADR countries to the future of work has been only partial. Despite the existence of
technology to automate many tasks and to enable remote and platform work, CADR countries have not
fully transitioned to the so-called “future of work.” Employment is evolving towards nonroutine tasks, but
routine ones still dominate. Remote work surged during the pandemic but now seems to be diminishing.
Platform work is present but makes up a small share of total jobs in some countries. This highlights a
deficiency of research on the future of work, which tends to focus on the “technological potential” for a
task or an occupation to be automated or performed remotely and to set aside other essential factors
influencing which types of workers are hired and which types of capital investments are made.
Employment structure, skills, adoption and diffusion of technology, and globalization mediate the
impact of technology on jobs. Chapter 3 showed that there is variation across and within CADR countries
in how intensive employment is in routine, automatable tasks. This variation can be decomposed into the
different factors underlying employment change to understand which factors are most important. Factors
that are key to understanding changes in what work is being done and how work is being done include the
adoption and diffusion of technology (Do firms and workers use new technologies?), but also employment
structure (Do sectors that use technology dominate?), globalization (Is the economy open to technological
influences?), and the supply of skills (Do workers have skills that are complementary to technology?).41
41 For a discussion of the role of these factors, see Caunedo, Keller, and Shin (2021); Lewandowsi et al. (2022); Oesch (2013); Lo Bello, Puerta,
and Winkler (2019); and Martins-Neto et al. (2021). Other factors not discussed here include labor market institutions and the availability of
reliable and high-quality infrastructure (Breemersch, Damijan, and Konings 2017; Cirera, Comin, and Cruz 2022; Oesch 2013).
46
Understanding each of these factors can help show why technology is not having the kinds of impacts in
CADR countries that it is in advanced economies.
Changes in the nature of work in CADR countries are being influenced not only by developments
within the region but also by technological progress abroad. New digital working arrangements are
expanding opportunities for cross-border provision of services in the context of employment structures
in CADR countries that skew towards low-productivity services. Better and cheaper ICT could enable
the provision of services from CADR countries, facilitating delivery across borders. This could create
opportunities to scale services operations beyond local markets to reach global consumers with benefits
for productivity, competitiveness, and living standards (World Bank and WTO 2023). On the other hand,
automation abroad could lead to a reversal of the offshoring that has created jobs in CADR countries.
Automation abroad could drive a process of deoffshoring42 as automation-induced labor savings at
home and higher labor costs abroad change firms’ calculations about where to produce goods.43 New
technologies may also weaken the demand for CADR migrants outside of CADR countries, particularly in
the United States, though increased demand is also possible if automation leads to expansions in output
in jobs that favor migrants. This chapter discusses each of the factors underlying employment change
in CADR countries, including and beyond the adoption and diffusion of technology, and highlights how
technological progress abroad could play a role in shaping the future of work in the region.
THE CHANNELS OF EMPLOYMENT CHANGE
Skills, employment structure, adoption and diffusion of technology, and globalization all play a role in
explaining the evolution of employment in CADR countries. Examining the relationship between these
different factors and how intensive a job is in the routine tasks that are associated with jobs of the past shows
that each is important in explaining what workers do at work in CADR countries.44 Across CADR countries:
• More education is associated with employment that is less intensive in routine tasks.
• Less agricultural work is associated with employment that is less intensive in routine tasks.
• Greater technology use is associated with employment that is less intensive in routine tasks.
• Greater global value chain participation is associated with employment that is more intensive in
routine tasks.45
Technology is not a primary factor explaining differences in the tasks that workers do in CADR
countries. Decomposing the variation in the routine intensity of employment shows that employment
structure—the sectoral distribution of employment—explains about half of the cross-country variation in
routine intensity among CADR countries and is the most important factor for within-country differences
in nearly every CADR country (figure 4.1). Worker characteristics, primarily skills,46 explain another third
of the cross-country variation and are the second most important factor explaining within-country
differences in nearly every CADR country. Notably, technology only explains 8 percent of the variation
across countries. Globalization is a more important factor than technology in all CADR countries except
Panama. This result contrasts with results for high-income countries where technology is found to be the
main contributor to cross-country variation (Lewandowski et al. 2022). The following sections delve into
each of these factors.
42 Reshoring means moving production back to the original country of production. “Deoffshoring” is a broader concept, which captures
production that would have been but is not offshored and production that is moved to a third country as well as reshoring.
43 See, for example, Artuc, Bastos, and Rijkers (2023) and Maloney and Molina (2019).
44 See appendix E for full results and an explanation of the methodology for estimating the determinants of routine task intensity across
countries.
45 As in Lewandowski, Madoń, and Winkler (2023), this finding is specific to forward linkages—that is, production and shipment of goods that
are reexported. Results for backward linkages—that is, importation of inputs that are used to produce goods that are then exported, are not
statistically significant.
46 The decomposition includes three worker characteristics: gender, age, and education. Education is the dominant factor across countries.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 47
FIGURE 4.1: Factors Contributing to Cross-Country Variation in Routine Task Intensity,
2010–2021
Percentage
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
All CRI DOM GTM HND NIC PAN SLV
Worker skills Sector composition Technology Globalization Year fixed e ects
Sources: Borin, Mancini, and Taglioni 2021; SEDLAC (CEDLAS and The World Bank).
Note: Results in the figure show the decomposition of the R-squared from an ordinary least squares (OLS) regression where the routine task intensity (RTI)
index is regressed on worker characteristics (gender, age, and educational level), structural change (sector fixed effects), technology (share of ICT-intensive
occupations), and globalization (global value chain-related trade). The results use all available years (2011–21 for Costa Rica, 2010–21 for the Dominican
Republic and El Salvador, 2010–19 for Guatemala and Honduras, 2010–14 for Nicaragua, and 2011–18 for Panama) and include year fixed effects. CRI = Costa
Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Employment Structure
Jobs in CADR’s less advanced economies are less vulnerable to technological change in part because
of the continued prevalence of agricultural employment. The bias of the economic structure of several
CADR countries towards agriculture makes them less vulnerable to automation of the routine tasks that
are concentrated in manufacturing and (certain) service sector jobs (Martins-Neto et al. 2021). Cross-
country evidence suggests that structural transformation—the shift of jobs out of agriculture and into
industry and services—raises exposure to automation-linked routinization (Das and Hilgenstock 2022).47
For example, a recent study of European Union countries shows that later structural transformation—
higher initial shares of employment in agriculture—in Central and Eastern European countries contributed
to an increase in employment in routine-intensive jobs in contrast to the decline experienced in Western
Europe (Hardy, Keister, and Lewandowski 2018).48 This implies that the slow structural transformation
of several CADR countries—particularly Guatemala, Honduras, and Nicaragua—is linked to lower
exposure to automation. Unfortunately, this lower exposure comes at the expense of higher-productivity
(nonagricultural) employment.
However, technological advances can spur productivity increases in agriculture in CADR, including
among smallholder farmers. Digital technologies can improve the information available to farmers and
expand their access to input and output markets and financial products, in turn improving their efficiency
and productivity (Schroeder, Lampietti, and Elabed 2021; Morris et al. 2020; FAO 2022). Smartphone apps
and messaging services can link farmers to expert advice on farming techniques, provide tools in a range of
information-related areas (for example, pest detection, quality control, and grading), and facilitate the use of
shared asset or machinery hire services (“Uber for tractors”) that connect equipment owners with potential
users. Digital histories linked to phone usage can make financial products available to farmers lacking
credit histories. E-commerce platforms improve price discovery and facilitate matching of buyers and
sellers. More sophisticated technologies like distributed ledger technology such as blockchain can improve
quality control and traceability, while geo-enabled technologies like remote sensors and drones combined
47 This has also been found in the case of Central and Eastern European countries (Hardy, Keister, and Lewandowski 2018).
48 See also Bárány and Siegel (2018) examining the role of structural change in employment and wage polarization in the United States.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 48
with AI-enabled analysis can facilitate precision agriculture that can improve how fields and animals are
managed. Robots and automated equipment operation can reduce the need for manual labor for some
tasks. Table 4.1 highlights examples of digital tools currently deployed in CADR countries. Despite these
opportunities, deployment of digital technologies in agriculture faces challenges including lack of digital
literacy among smallholder farmers, a preference for using cash and relying on personal relationships,
limited connectivity, and expensive data services and equipment (Loukos and Arathoon 2021).
TABLE 4.1: Examples of Digital Interventions Targeted to Smallholder Farmers in CADR
Country Tool
CRI • Agromensajes provides advice on market prices for fresh produce.
• Coffee Cloud provides alerts on pests and diseases and weather for coffee.
• CR Orgánico provides e-commerce services for organic fresh produce.
• IICA/PROCAGICA uses smart farming IoT sensors for coffee.
• Celotor uses smart farming IoT sensors for cattle.
DOM • IICA/PROCAGICA uses smart farming IoT sensors for coffee.
GTM • Precios del Café provides advice on market prices for coffee.
• APP MAGA provides advice on market prices for fresh produce.
• Cacao Móvil provides advice on best practices for coffee.
• Coffee Cloud provides alerts on pests and diseases and weather for coffee.
• Digitagro provides e-commerce services for fresh produce.
• IICA/PROCAGICA uses smart farming IoT sensors for coffee.
HND • Cacao Móvil provides advice on best practices for coffee.
• Coffee Cloud provides alerts on pests and diseases and weather for coffee.
• IICA/PROCAGICA uses smart farming IoT sensors for coffee.
NIC • Cacao Móvil provides advice on best practices for coffee.
• Clima y Café and Cafenica Pronósticos is a localized weather and alert system for coffee.
• IICA/PROCAGICA uses smart farming IoT sensors for coffee.
PAN • Mercadito provides e-commerce services for fresh produce.
• IICA/PROCAGICA uses smart farming IoT sensors for coffee.
SLV • Cacao Móvil provides advice on best practices for coffee.
• Coffee Cloud provides alerts on pests and diseases and weather for coffee.
• Smart Agro 4.0 uses smart farming IoT sensors for coffee, potatoes, cotton, fresh produce.
• IICA/PROCAGICA uses smart farming IoT sensors for coffee.
Source: Loukos and Arathoon 2021.
Note: IoT is Internet of Things and refers to a network of devices equipped with sensors that exchange data; IICA = Inter-American Institute for Cooperation on
Agriculture. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Even as relatively high levels of agricultural employment persist in several CADR countries, the region
is also experiencing premature deindustrialization. The canonical path of economic development
involves employment first shifting from agriculture to (higher-productivity) industry and then from industry
to (lower-productivity) services as industrial productivity increases. Recent evidence suggests that in
developing countries and in Latin America, the shift of employment from industry to services is happening
at lower levels of development and at lower levels of peak manufacturing employment (Beylis et al. 2020;
Rodrik 2016).49 The same pattern of “premature deindustrialization” is apparent for CADR countries (figure
4.2). Indeed, the industrial share of employment contracted in all CADR countries except Panama and
Honduras between 1991 and 2019 (figure 4.3). Factors contributing to this early deindustrialization include
barriers to the movement of resources into manufacturing such as labor regulations and mismatched skills
(Sinha 2022).
49 The same is observed for sector value-added, though the phenomenon is more apparent for employment (Beylis et al. 2020). See also Felipe,
Mehta, and Rhee (2019).
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 49
Premature deindustrialization may mean that CADR countries do not follow the same path towards
increased exposure to routinization as the United States and other more advanced economies. As structural
transformation proceeds in CADR countries, the risks of routinization could increase. However, the transition
of employment from agriculture directly to services suggests that these risks will only increase moderately
via the structural transformation channel. Low-paid services are harder to automate because they require
more dexterous interactions, because the cost of automation technology is often high relative to labor cost,
and because human interaction is at least at times an important component of service provision (Autor 2022).
FIGURE 4.2: Deindustrialization in CADR and FIGURE 4.3: Change in the Industrial Share
High-Income Countries, 1991–2021 of Employment, 1991–2019
Percentage, log GDP per capita Percentage points
60 4%
50 2%
Share of manufacturing jobs
0%
40
-2%
30
-4%
20
-6%
10
-8%
8 9 10 11 12
Log of per capita GDP (PPP 2017 USD)
-10%
CADR countries High−income countries DOM CRI GTM SLV NIC HND PAN
Source: World Development Indicators data, World Bank. Source: World Development Indicators data, World Bank.
Note: Each data point corresponds to one country and one year. The years Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
are between 1991 and 2021. HND = Honduras; NIC = Nicaragua; PAN – Panama; SLV = El Salvador.
Though deindustrialization implies a lower risk of automation, it raises concerns about future growth in
CADR countries because manufacturing has historically been an accelerator of economic development.
Several characteristics thought to be unique to the manufacturing sector explain its links to growth. These
include tradability and scalability, scope to innovate, and spillovers combined with a significant capacity to
absorb low-skilled workers and make them more productive (Hallward-Driemeier and Nayyar 2018; Nayyar,
Hallward-Driemeier, and Davies 2021). The early shrinkage of the industrial sector and expansion of services
in CADR countries thus raises alarms about future sources of productivity growth as services do not typically
have these attributes. There are high-productivity, tradable services but employment in these subsectors
tends to be limited in number and confined to higher-skilled jobs in information, finance and insurance,
professional, and educational services (Blinder and Krueger 2013; Dingel and Nieman 2020; Jensen and
Kletzer 2010). In contrast, the employment-intensive services that increasingly dominate employment in the
region tend to lack the productivity growth and job creation that are associated with manufacturing jobs.
However, developments in ICT could enhance the benefits that services jobs have for development
within the region. Technological advancements are giving the services sector some of the same positive
productivity benefits that have historically been associated with manufacturing (Nayyar and Cruz 2018;
Nayyar, Cruz, and Zhu 2018; Nayyar, Hallward-Driemeier, and Davies 2021). ICT is increasingly enabling
remote delivery of services (for example, via platforms), which in turn increases the potential for scalability
including across borders. Estimates find that about a third of jobs in the United States can be performed
remotely, which suggests the potential to provide them overseas (Dingel and Nieman 2020).50 This amounts
50 Dingel and Nieman (2020) focuses specifically on whether a job can be done from home, while estimates of offshorability contemplate
whether a job can be done overseas (Blinder 2006; Blinder 2009; Blinder and Krueger 2013; Jensen and Kletzer 2010). This results in
differences particularly in the manufacturing sector where few jobs can be done remotely but many jobs can be done overseas.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 50
to tens of millions of jobs in the United States alone. In sum, improvements in ICT could help expand the
reach of lower-skilled, high-employment sectors like retail and personal services, facilitate the access of
lower-skilled workers to higher-skilled, low-employment sectors like ICT, and increase opportunities in
higher-skilled social services like education and health (Nayyar, Hallward-Driemeier, and Davies 2021).
Trade in services is becoming more important in CADR. In Central America, services exports grew faster
than goods between 1991 and 2017 (Ulku and Zaourak 2021). Cross-border services trade boosted GDP
per capita in Costa Rica, the Dominican Republic, Guatemala, and Panama between 2000 and 2014
(WTO 2019). Services exports can be important to employment. In Costa Rica, for instance, cross-border
services exports are responsible for more than 10 percent of jobs (World Bank and WTO 2023). While
traditional industries like tourism and travel still dominate services exports in the region, improvements
in ICT increasingly make proximity between producers and consumers less important and cross-border
provision of services more viable, which is reflected in the increasing share of modern services like finance,
ICT, and professional and business services in the services export basket of CADR countries. In several
CADR countries, the share of digitally deliverable services exports—exports that could, given adequate
infrastructure and financial resources, be delivered digitally—has increased substantially during the last
two decades (figure 4.4). Data on trade in services that are actually facilitated by digital technologies is
only available for Costa Rica, where around 40 percent of services exports were enabled by ICT in 2017
(BCCR 2019).
FIGURE 4.4: Digitally Deliverable Services Exports, 2005–21
Percentage points (2005 = 0)
60
50
40
30
20
10
0
-10 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
CRI DOM SLV GTM HND NIC PAN
Source: UNCTAD.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
The emergence of digital platforms has enabled the growth of cross-border services delivery. As
chapter 3 describes, platforms can enable workers in CADR to access a global market for jobs and
tasks. Platforms can also create opportunities for cross-border services provision. Platforms allow firms,
including SMEs in developing countries, to undertake a multitude of business functions from marketing
to e-commerce, financing, and payments. Platforms can lower entry barriers, create network effects, and
enable SMEs to access additional consumers and sourcing options in regional and global markets (ADB
2021; OECD 2021). Digital platforms use became common including among small firms during the COVID-
19 pandemic (figure 4.5a). Still, sales made via digital platforms are low (figure 4.5b).
Digital technologies, particularly digital platforms, are opening new opportunities for small businesses
in CADR. The Future of Business Survey is a survey of small- and medium-sized enterprises with an active
Facebook Business page conducted by Meta with the World Bank and the OECD. The survey provides
insight into how small businesses engage with digital tools in general and digital platforms in particular.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 51
SMEs across CADR report using digital tools to facilitate online sales and purchases: more than 40 percent
say they use such technologies. Most report that less than one-quarter of their purchases and sales are
online. However, when they do use the internet for purchases and sales, firms in CADR are more likely
to connect with suppliers or buyers outside their home country (figure 4.6a–b). This is consistent with
improved ICT expanding markets for SMEs. Digital platforms in particular seem capable of transformative
impacts. Large shares of firms across CADR report that digital platforms had a very or extremely important
impact on their business. Larger shares of firms in CADR than in Korea or the United States report that
digital platforms increased their sales and their customer or supplier base (figure 4.6c–d). SMEs most
commonly report using digital platforms for advertising and communication, but other uses are also
common including payments (Costa Rica and Guatemala), product or service development (Dominican
Republic and El Salvador), and sale or purchase of goods or services (Honduras and Nicaragua).
FIGURE 4.5: Use of Digital Platforms, 2020
Percentage
a. Share of firms starting or increasing the use of digital platforms
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Mid-2020 End-2020 Mid-2020 End-2020 Mid-2020 End-2020 Mid-2020 End-2020
GTM HND NIC SLV
b. Share of monthly sales using digital platforms during the last 30 days
35%
30%
25%
20%
15%
10%
5%
0%
Mid-2020 End-2020 Mid-2020 End-2020 Mid-2020 End-2020 Mid-2020 End-2020
GTM HND NIC SLV
Manufacturing Retail Other Services Small (5–19) Medium (20–99)v Large (100+) All
Source: World Bank Business Pulse Survey 2020.
Note: GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama.
In sum, the prevailing employment structure in CADR countries implies both less vulnerability to
automation in the near term and a more challenging services-driven growth model. The continued
prevalence of agricultural employment in several CADR countries on one hand, and the prevalence of
services employment in all CADR countries on the other, mean relatively fewer workers do the tasks that
have been automated away in advanced economies. But this leaves the challenge of substantial shares of
workers employed in less-productive economic sectors. Technological advances in agriculture and digital
trade in services offer potential new avenues for growth.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 52
FIGURE 4.6: SME Engagement with Digital Tools, 2020 and 2022
Percentage
a. SMEs reporting more than half of online b. SMEs reporting more than of online orders
orders placed within home country, 2022 received from home country, 2022
90% 70%
80% 60%
70%
50%
60%
50% 40%
40% 30%
30%
20%
20%
10%
10%
0% 0%
USA KOR HND CRI NIC GTM SLV USA DOM HND CRI KOR NIC SLV GTM
c. SMEs reporting an increase in sales d. SMEs reporting more suppliers or
from digital platforms, 2020 customers from digital platforms, 2020
50% 40%
45% 35%
40%
30%
35%
30% 25%
25% 20%
20%
15%
15%
10%
10%
5% 5%
0% 0%
PAN SLV NIC DOM CRI HND GTM USA KOR SLV GTM NIC CRI DOM PAN HND KOR USA
Source: Future of Business Surveys 2020, 2022.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; KOR = Korea; NIC = Nicaragua; PAN = Panama; SME = small and
medium-sized enterprise; SLV = El Salvador, USA = United States of America.
Supply of Skills
A larger supply of less-skilled workers favors more routine-intensive employment. The supply of
skills is a key factor in technology adoption (Caunedo, Keller, and Shin 2023; Martins-Neto et al. 2021).
Human capital is one of the most important determinants of how quickly a country adopts technologies
(Benhabib and Spiegel 1994; Comin and Hobijn 2004). Firms in countries with more abundant less-skilled
labor select technologies complementary to less-skilled labor, which are different from the technologies
selected by firms in countries with more abundant high-skilled labor (Caselli and Colemann 2006; Eden
and Gaggl 2020). For example, research on less-developed Central and Eastern European countries
finds that upskilling, particularly increasing tertiary education attainment, contributed to the growth in
nonroutine cognitive tasks in these countries, even as structural change implied more routine-intensive
jobs (Hardy, Keister, and Lewandowski 2018).
The low skill levels in the CADR region contribute to the prevalence of routine-intensive employment.
Education levels have improved throughout the CADR region, driven by a decline in workers who have
primary education or less and an increase in those who have secondary education (figure 4.7). This is
consistent with the recent trends away from routine employment identified in chapter 3. However, low skill
levels predominate in most CADR countries. Education levels in the higher-income CADR countries are
low relative to the most developed economies globally, and those in the lower-income CADR countries
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 53
are low relative to the higher-income CADR countries. In Costa Rica, the Dominican Republic, and
Panama, around a quarter of workers are tertiary educated versus more than half of workers in Korea and
the United States. Very few workers in Korea and the United States have less than secondary education
while at least a quarter do in Costa Rica, the Dominican Republic, and Panama. In the remaining CADR
countries, 14 percent or less of workers have tertiary education. Seventy percent of workers in Guatemala
have primary education or less.
FIGURE 4.7: Education of the Employed Population in CADR Countries, 2000–21
Percentage
a. Costa Rica b. Dominican Republic c. El Salvador
70% 70% 70%
60% 60% 60%
50% 50% 50%
40% 40% 40%
30% 30% 30%
20% 20% 20%
10% 10% 10%
0% 0% 0%
2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
2020
2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
2020
2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
2020
d. Guatemala e. Honduras f. Nicaragua
90% 100% 80%
80%
70% 80%
60%
60%
60%
50%
40%
40%
40%
30%
20% 20%
20%
10%
0% 0% 0%
2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
2020
2000
2003
2006
2009
2012
2015
2018
2021
2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
2020
g. Panama
50%
45%
40%
35%
30%
25% Low
20% Medium
15%
High
10%
5%
0%
2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
2020
Source: SEDLAC (CEDLAS and the World Bank).
Note: Low education is less than 9 years of school, medium is between 9 and 13 years, and high is 14 or more.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 54
Low educational quality could also help explain the prevalence of routine-intensive employment.
Indicators of learning and education system performance in CADR suggest that students may not
be acquiring the skills that would enable them to take on more skill-intensive jobs. For the subset of
CADR countries that participate in the Program for International Student Assessment, an internationally
comparable student assessment administered to 15-year-olds, the results are stark (figure 4.8). Even in the
best performing country in the region, Costa Rica, less than 60 percent of 15-year-olds meet the minimum
proficiency standard in reading and just 40 percent in math. In the poorest performing CADR countries,
less than 30 percent and 20 percent do, respectively. This compares with at least three-quarters of
students in both subjects in the OECD.
FIGURE 4.8: Share of Students Meeting Minimum Proficiency in Reading and Math, 2022
Percentage of 15-year-olds
80%
70%
60%
50%
40%
30%
20%
10%
0%
OECD CRI PAN GTM SLV DOM
Mathematics Reading
Source: OECD 2022a.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; OECD = Organisation for Economic Co-operation and Development; PAN = Panama;
SLV = El Salvdaor.
Workers in CADR countries also seem to lag in the kinds of skills needed to accomplish the new
tasks associated with new technologies. Globally, technological progress is translating into stronger
demand for critical thinking and other higher-order cognitive skills, socioemotional skills, digital skills, and
combinations of these skills (World Bank 2016, 2019). There is growing complementarity between social
and cognitive skills (Deming and Kahn 2018; Weinberger 2014). The availability of ICT skills in particular is
associated with the adoption of digital technologies (Nicoletti, von Rueden, and Andrews 2020). Though
data on skills is scarce in CADR, the kinds of skills associated with technological change seem to be
lacking. For example, an analysis of job vacancies in the Dominican Republic finds that teamwork, service
orientation, responsibility, and big data are the most demanded skills (Guataquí 2021). However, firms
report difficulty filling vacancies with the most important limitation being the lack of adequate technical
skills among jobseekers (MEPyD 2022). Inadequate digital skills are also cited as a limitation when trying
to fill vacancies. In El Salvador, digital startups report low quality training and education as barriers to
hiring talent (World Bank 2022c). Data on technology skills taken from LinkedIn profiles in Costa Rica and
Panama show a similar lag in technology skills in these countries (box 4.1).
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 55
BOX 4.1: The Availability of Technology Skills in Costa Rica and Panama
Data from LinkedIn provide insight into the extent to which different types of skills are utilized
in Costa Rica and Panama. The global job networking site LinkedIn has detailed data on its users,
including on the skills they have learned during employment. In CADR, Costa Rica and Panama
have sufficient LinkedIn users to make valid insights with this data. The user profile data on skills
is used to calculate skill penetration—that is, the share of an industry’s top 50 skills that come
from one of 249 different skills categories that group the thousands of skills of LinkedIn users.
Argentina is included in the analysis as a comparator for Costa Rica and Panama.
Costa Rica and Panama lack the technology and disruptive technology skills most closely
associated with new technologies. The skills penetration metric shows that Costa Rica and
Panama lag substantially relative to the global average in the penetration of technology skills like
digital literacy, graphic design, and mobile app and web development and in disruptive technology
skills like AI, data science, and robotics (figure B4.1.1). Panama does perform above the global
average on technology skills in several industries—hospitals and health care, accommodation and
food services, and education—but is well below the global average in others such as technology,
information, and media. Costa Rica performs below the global average across all sectors. The
performance of both countries is worse in the disruptive technology skills that are most closely
associated with the most innovative technologies. Only Costa Rica’s real estate and equipment
rental sector performs above the global average.
FIGURE B4.1.1: Penetration of Technology and Disruptive Technology Skills, 2022
Relative skills penetration (1 = global average)
a. Technology b. Disruptive technology
Accommodation Accommodation
Administrative
Administrative
Construction
Construction
Consumer Services
Education Consumer Services
Entertainment
Education
Farming
Financial Services
Financial Services
Government Government
Health Care Health Care
Manufacturing
Manufacturing
Professional Services
Professional Services
Real Estate
Retail Real Estate
Technology & Media
Retail
Transportation
Technology & Media
Utilities
Wholesale Transportation
0.00 0.50 1.00 1.50 0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60
Global average PAN CRI ARG
Source: LinkedIn 2022.
Note: ARG = Argentina; CRI = Costa Rica; PAN = Panama.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 56
Lack of management skills may also be hindering adoption of technologies in CADR. Good management
is increasingly viewed as a prerequisite to technology adoption. For instance, low management quality is
associated with lower digital technology adoption in Europe (Nicoletti, von Rueden, and Andrews 2020).
Evidence from the World Bank’s Firm-Level Adoption of Technology Surveys show that “firm capabilities”—
management capacity, education, and ability to learn—are necessary for technology to function effectively
(Cirera, Comin, and Cruz 2022). The surveys, which focus on developing countries, find that lack of these
capabilities is the second-most important obstacle to adoption among firms of all sizes after lack of
demand and uncertainty. These findings are consistent with the evidence that management practices are
a key component of firm success (Bloom, Sadun, and Van Reenen 2017). Data on management practices in
CADR are very limited. However, the World Management Survey, which measures management practices
across countries, was undertaken in Nicaragua. The average management score in Nicaragua is 2.4, the
seventh lowest among the 35 countries surveyed (Bloom, Sadun, and Van Reenen 2017). This is consistent
with findings of a “management gap” between the quality of management in Latin America and that in
more developed economies (Lederman et al. 2014).
Adoption and Diffusion of Technology
Employment may be shifting more slowly away from routine-intensive employment in CADR simply
because technology use is less common. There is substantial variation across countries globally in
technological sophistication. Recent efforts to better understand the role of technology in developing
country settings finds that innovation is generally at the margins and firms are not close to the technological
frontier (Cirera, Comin, and Cruz 2022; Cirera and Maloney 2017).
CADR countries rank low on summary measures of technological progress. The World Bank’s Digital
Adoption Index (DAI) measures digital technology adoption by the people, governments, and businesses
of most countries in the world. The DAI shows that out of 180 countries, the Dominican Republic, El
Salvador, Guatemala, Honduras, and Nicaragua score in the bottom half of countries globally (figure 4.9).
Costa Rica and Panama rank better but are still below most OECD countries. A similar picture emerges
from the Frontier Technologies Readiness Index compiled by the United Nations Conference on Trade
and Development (UNCTAD) (figure 4.10). The index shows the same 5 countries ranking at or below the
average of 166 countries on the specific ICT readiness sub-measure.
FIGURE 4.9: The Digital Adoption FIGURE 4.10: The Frontier Technology
Index, 2016 Readiness Index: ICT, 2022
Rank Rank
0 HND NIC GTM SLV DOM PAN CRI HND NIC GTM SLV DOM PAN CRI
0
47
75 66 63
90 76
97 95
110 103 100
116 106
118
180 166
Source: World Bank 2016. Source: UNCTAD 2023.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador. HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 57
Slow technological progress is reflected in the relatively small contribution of ICT to economic growth
in CADR countries. The Conference Board’s Total Economy Database includes data on the contribution
of capital services provided by ICT and non-ICT assets to real GDP growth for Costa Rica, the Dominican
Republic, and Guatemala (figure 4.11a). These data provides an indication of the intensity of ICT use in
production (Eden and Gaggl 2020; Spiezia 2012). The contribution of ICT assets to GDP growth has been
substantially lower than the contribution of non-ICT assets in all CADR countries since 2000. In the United
States, in contrast, ICT assets have made a larger contribution than non-ICT assets. Data compiled by the
Inter-American Development Bank and the University of Santiago, Chile generally show similar trends for
El Salvador and Honduras using the contribution of ICT and non-ICT assets to value-added growth (figure
4.11b).
FIGURE 4.11: The Contribution of ICT and Non-ICT Assets to Economic Growth, 2000s–10s
Percentage points
a. Difference in ICT and non-ICT assets’ b. Difference in in ICT and non-ICT assets’
contribution to real GDP growth contribution to value added growth
1.0 0.5
0.0
0.0
-1.0
-2.0 -0.5
-3.0
-1.0
-4.0
-5.0 -1.5
-6.0
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
-2.0
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
USA CRI DOM
GTM Mature economies SLV HND
Source: The Conference Board. Source: LAKLEMS.
Note: A negative value indicates non-ICT assets contributed more to Note: A negative value indicates non-ICT assets contributed more to growth
growth than ICT assets. CRI = Costa Rica; DOM = Dominican Republic; than ICT assets. HND = Honduras; SLV = El Salvador.
GTM = Guatemala; USA = United States of America.
CADR countries have generally been slower to adopt new technologies, though these lags have
shortened with each significant technological advancement. On average, the most advanced
economies globally have adopted technologies 42 years earlier than developing countries, though
these gaps have narrowed over time (Comin and Hobijn 2010; Comin and Mestieri 2018). CADR countries
lagged about 70 years behind the United States in the adoption of the telephone (figure 4.12). Adoption of
the computer lagged the United States by between 13 years in Costa Rica and 20 in Honduras. Lags were
also experienced for the internet and for cell phones. However, with each new technology, the lags have
become shorter. The mean lag in CADR declined from 70 years for the telephone to 25 for the computer,
just over 7 for the internet, and just under 7 for the cell phone.
However, diffusion is limited even when technology is available. Intensity of use is a critical factor
that can mediate the impact of technology on the labor market. Even in a country where a technology is
adopted, take-up and use might be limited. For example, cell phones are prevalent in CADR countries,
highlighting the importance and availability of this technology (figure 4.13a). However, use of the internet
varies across countries. Costa Rica and the Dominican Republic have rates of internet use that approach
those of Korea and the United States (figure 4.13b), but in all other CADR countries, at least one-third of
the population did not use the Internet in 2021. Notably, unlike lags in technology adoption, lags between
advanced and developing countries in how intensively technologies are used have widened over time
(Comin and Mestieri 2018).
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 58
FIGURE 4.12: Lags in Technology Adoption in CADR Countries
Years
HND
NIC
Telephone
GTM
SLV
DOM
CRI
PAN
HND
GTM
Computer
PAN
NIC
SLV
CRI
NIC
HND
GTM
Internet
SLV
PAN
DOM
CRI
NIC
HND
Cellphone
GTM
CRI
SLV
PAN
DOM
0 10 20 30 40 50 60 70 80
Source: Comin, Hobijn, and Rovito 2008.
Note: Lags are calculated as the years between a benchmark year and the year in which the United States had the same adoption as the CADR country
in the benchmark year. Data are not available for computers for the Dominican Republic. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
FIGURE 4.13: Diffusion of the Internet and Cell Phones, 2021
a. Mobile cellular subscriptions (per 100 people) b. Individuals using the Internet (% of population)
200 100%
180 90%
160 80%
140 70%
120 60%
100 50%
80 40%
60 30%
40 20%
20 10%
0 0%
SLV CRI KOR PAN GTM USA NIC DOM HND KOR USA DOM CRI PAN SLV NIC GTM HND
Source: World Development Indicators data, World Bank.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; KOR = Korea; NIC = Nicaragua; PAN = Panama; SLV = El Salvador;
USA = United States of America.
The evidence of low diffusion of technology is evident in the small share of CADR workers using
computers and the internet. Ten percent or less of workers in the Dominican Republic, El Salvador,
Guatemala, Honduras, and Nicaragua work in jobs that are intensive in the use of computers or the internet
(figure 4.14a–b). The share in Costa Rica and Panama is slightly higher at 12 to 13 percent. However, these
shares are significantly lower than that in the United States, which is 27 percent in the case of computers
and 25 percent in the case of the internet.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 59
FIGURE 4.14: Use of Computers and the Internet at Work, 2021
% of workers in computer- and internet-intensive occupations
a. Computer use b. Internet use
30% 30%
25% 25%
20% 20%
15% 15%
10% 10%
5% 5%
0% 0%
USA PAN CRI DOM SLV NIC HND GTM USA PAN CRI DOM NIC SLV HND GTM
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The years are 2019 for Guatemala and Honduras, 2018 for Panama, and 2014 for Nicaragua. Computer- and internet-intensive occupations are defined
as occupations in the top 25 percent of computer and internet use at work as defined using PIAAC data from comparator countries. CRI = Costa Rica;
DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador; USA = United States of America.
Technology use by CADR firms is also limited. Data on technology use by CADR firms is outdated, but
surveys undertaken during the COVID-19 pandemic and trade data on robots provide insight into the
prevalence of deployment of new technologies by firms.51 The World Bank’s Business Pulse Survey fielded
during the COVID-19 pandemic shows that half or more of firms in El Salvador, Guatemala, Honduras, and
Nicaragua reported starting to use or increasing their use of digital platforms at the end of 2020 (figure
4.15). Despite this high prevalence, the share of monthly sales using digital platforms is relatively low at
15 percent or less in all four countries. In Chile, in contrast, the share was 40 percent in mid-2021. Data
FIGURE 4.15: Use of Digital Platforms, 2021 TABLE 4.1: Imports of Industrial Robots, 2021
Percentage US$
80% Country Imports
70% United States 1,082,146,087
60%
South Korea 270,898,096
50%
Costa Rica 7,726,490
40%
30% Dominican Republic 2,096,294
20% Honduras 841,254
10% El Salvador 501,163
0%
HND SLV NIC GTM NIC SLV GTM HND Panama 122,704
Firms starting or increasing Monthly sales via
use of digital platform digital platform Guatemala 48,693
Mid-2020 End-2020
Nicaragua 1,850
Source: World Bank Business Pulse Surveys 2021. Source: Observatory of Economic Complexity.
Note: GTM = Guatemala; HND = Honduras; NIC = Nicaragua; SLV = El Note: Industrial robot imports are defined using Harmonized System (HS)
Salvador. Code 8479.50.
51 Data on how firms use technology is available for Enterprise Surveys for all CADR countries. However, this data is old. El Salvador is the only
country where data was collected in the last five years. These older data show that e-mail use by firms is widespread, websites are common
but far from universal, and most firms do not use technology licensed from a foreign-owned company. Many firms introduce new products
and processes, but spending on research and development is not common.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 60
on firms’ use of robots is not available for CADR countries, but trade data on imports of industrial robots
provide suggestive evidence that robotic technologies are not widespread. Costa Rica and the Dominican
Republic lead the way with imports in 2021 of US$7.7 million and US$2.1 million respectively (table 4.1).
At tens of thousands of dollars per robot, these imports could represent around 300 robots imported by
Costa Rica or six for every 100,000 people and 80 by the Dominican Republic or less than one robot for
every 100,000 people. These imports are very low compared to the US$1 billion of robots imported by
the United States (13 for every 100,000 people) and the US$270 million imported by South Korea (21 for
every 100,000 people). However, they are also much larger than other countries in CADR whose imports
were less than US$850,000 (equivalent to 35 robots or less than half a robot for every 100,000 people).
The persistence of agricultural employment in several CADR countries is further evidence of the lack
of diffusion of technologies in the region. The green revolution brought new technologies to farmers in
LAC in the 1960s and 1970s, shifting labor out of agriculture and into more productive economic activities
(FAO 2022; Fuglie et al. 2020; Schlogl and Sumner 2018). However, technological progress in agriculture
has not occurred evenly across the region (Elverdin, Piñeiro, and Robles 2018). Data on agricultural
mechanization is very limited and, where available, outdated, but estimates based on projections
compiled by the United States Department of Agriculture suggest that farm machinery measured in total
horsepower per hectare is around 10 percent of the level in the United States in Guatemala, Honduras,
and Nicaragua versus 50 percent and 66 percent in Costa Rica and Panama (USDA 2023). This lack of
mechanization helps explain the persistence of agricultural employment in the former 3 countries where
about 30 percent of employment remains in agriculture. The lack of mechanization also helps explain
why several CADR countries have high rates of automatability: more than 70 percent of jobs at high risk
of automation are in agriculture in every CADR country (figure 4.16). The figure is as high as 94 percent
in Honduras. The persistence of agricultural employment highlights the puzzle at the beginning of the
chapter: many jobs have been automatable for a long time but have not been automated (Schlogl and
Sumner 2018).
FIGURE 4.16: Share of Jobs at High Risk of Automation by Sector, 2021
Percentage
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
HND NIC SLV PAN GTM DOM
Agriculture Industry Services
Source: Frey and Osborne 2017; SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The years are 2019 for Guatemala and Honduras, 2018 for Panama, and 2014 for Nicaragua. DOM = Dominican Republic; GTM = Guatemala;
HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Trade data provide evidence that imports of technology from abroad are limited. Imports of ICT goods
are low, suggesting that CADR countries are not relying on technologies developed outside of the region
(figure 4.17). In other developing economies, technology transfer has been shown to shift demand towards
more highly skilled workers, a phenomenon referred to as skill-enhancing trade or skill-enhancing
technology import (Araújo, Bogliacine, and Vivarelli 2011; Conte and Vivarelli 2011; Meschi, Taymaz, and
Vivarelli 2011, 2016). Imports of ICT services are also limited.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 61
FIGURE 4.17: Imports of ICT Goods and ICT Services as Share of Total Trade, 2021
Percentage of total trade in merchandise and services
18%
16%
14%
12%
10%
8%
6%
4%
2%
0%
KOR USA PAN CRI GTM HND SLV DOM NIC GTM CRI HND USA KOR DOM NIC SLV PAN
Goods Services
Source: UNCTAD.
Note: CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; KOR = Korea; NIC = Nicaragua; PAN = Panama; SLV = El Salvador;
USA = United States of America.
The cost of new technologies may play a different role in CADR than in advanced countries. The decline
in the price of computer capital is a key driver of the trend away from occupations that are intensive in
routine tasks (Autor and Dorn 2013; Autor, Levy, and Murnane 2003). However, there is evidence that this
decline is primarily a developed country phenomenon. For instance, the relative price of investment has
declined substantially in advanced economies, driven by computers and equipment, but only mildly in
emerging ones (Dao, Das, and Koczan 2019). Consistent with the evidence in chapter 3, this decline has
been linked to the greater polarization experienced in developed economies (Das and Hilgenstock 2022).
Additionally, cost pressures may not arise to incentivize the adoption of labor-saving technologies in the
way they do in developed countries, given the availability of relatively cheap labor. Even in the United
States, cost is one of the top two reasons firms cite for not adopting automation technologies (Acemoglu
et al. 2022).
The relatively high cost of ICT services in most CADR countries provides an example of how cost
may inhibit technological adoption and diffusion. The International Telecommunications Union (ITU),
the United Nation’s ICT agency, collects comparable international price data on several types of ICT
services. These services are expensive in all CADR countries relative to high-income countries (figure
4.18). The region’s more developed countries—Costa Rica, the Dominican Republic, and Panama—all face
prices that are well above those of high-income countries globally. Prices in the region’s less-developed
countries—El Salvador, Guatemala, Honduras, and Nicaragua—are higher still, and generally higher than
those in low- and middle-income countries. Prices in Honduras and Nicaragua in particular are well above
CADR peers, as well as other comparators.
Beyond cost, the challenges of organizational change can also inhibit technology adoption. Recent
literature on technology adoption, particularly of AI, highlights the importance of firm structures in
determining where new technologies are applied (Agrawal, Gans, and Goldfarb 2023; Bresnahan 2021;
Brynjolfsson and Mitchell 2017). Adoption requires rethinking organizational design, business models, and
production methods, a process that can take years and significant expertise to accomplish (Feigenbaum
and Gross 2023). For example, many producers of soccer balls in Pakistan did not take up a new waste-
reducing technology offered to them, likely because of misaligned incentives within the firm—a payment
that adjusted these incentives’ increased uptake (Atkin et al. 2017). The informal firms prevalent in CADR
countries are unlikely to be well suited to the kind of organizational redesign required to repurpose
business models for new technologies, especially the most sophisticated ones.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 62
FIGURE 4.18: Prices of ICT Service Baskets, 2022
Percentage of GNI
a. Fixed broadband b. Data-only mobile broadband c. Mobile data and voice
(high consumption)
14% 9% 10%
8% 9%
12%
7% 8%
10% 7%
6%
8% 6%
5%
5%
6% 4%
4%
3% 3%
4%
2% 2%
2% 1% 1%
0% 0% 0%
HND
NIC
SLV
GTM
LMIC
PAN
DOM
CRI
HIC
HND
NIC
GTM
SLV
LMIC
PAN
DOM
CRI
HIC
HND
NIC
LMIC
DOM
GTM
SLV
PAN
CRI
HIC
d. Mobile data and voice e. Mobile-cellular and low-usage
(low consumption)
9% 7%
8% 6%
7%
5%
6%
5% 4%
4% 3%
3%
2%
2%
1% 1%
0% 0%
HND
NIC
GTM
SLV
LMIC
PAN
DOM
CRI
HIC
HND
NIC
GTM
SLV
LMIC
DOM
PAN
CRI
HIC
Source: ITU 2022.
Note: CRI = Costa Rica; DOM = Dominican Republic; GNI = gross national income; GTM = Guatemala; HIC = high-income countries; HND = Honduras; LMIC = low-
and middle-income countries; NIC = Nicaragua; PAN = Panama; SLV – El Salvador..
Globalization
Globalization tends to favor more employment in routine-intensive jobs in developing countries.
The emergence of global value chains (GVCs) and offshoring is a countervailing force to technology’s
tendency to favor employment intensive in nonroutine tasks. Developed countries tend to offshore jobs
that are intensive in routine tasks to developing countries where labor costs are cheaper (Caunedo,
Keller, and Shin 2023; OECD 2017b).52 Indeed, cross-country research shows that the offshoring of jobs
intensive in routine tasks led to increased demand for routine-intensive jobs in emerging China and
Poland (Reijnders and de Vries 2018).53 Other cross-country evidence finds that participation in GVCs
works alongside structural transformation to increase routine-intensive employment (Das and Hilgenstock
2022). This is found to be true only for occupations that can be offshored (Lewandowski, Madoń, and
Winkler 2023).
52 In developed countries, in contrast, offshoring is associated with wage and employment losses for workers in routine-intensive occupations
and wage and employment gains for those in nonroutine-intensive ones (Hummels, Munch, and Xiang 2018). However, routine-biased
technological change is found to be a more important factor in explaining job polarization in developed countries (Goos, Manning, and
Salomons 2014).
53 Technological change had a larger impact, meaning that the overall trend in these countries was towards employment intensive in nonroutine
tasks.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 63
Connections to the global economy have become increasingly important in CADR in recent decades.
Foreign direct investment (FDI) inflows grew from a regionwide average of 1.0 percent of GDP between
1970 and 1989 to 4.5 percent between 2010 and 2022 (figure 4.19). This is in part the result of the entry
of CADR countries in manufacturing GVCs, particularly in textile and garments (Dominican Republic, El
Salvador, Guatemala, and Nicaragua), automotive (El Salvador, Honduras, and Nicaragua), and medical
devices (Costa Rica and the Dominican Republic) (Fernandes, Nievas, and Winkler 2022; Ulku and
Zaourak 2021). Though the sophistication of exports is low, several CADR countries are fairly integrated
in GVCs. GVC-related output makes up more than 15 percent of total output in Honduras, Nicaragua, and
Panama (figure 4.20). This is substantially less than Viet Nam’s 36 percent, but similar to that of Mexico
and more than that of Colombia and India. In every CADR country, the manufacturing sector is the most
important contributor to GVC participation.
FIGURE 4.19: Net Inflows of Foreign Direct FIGURE 4.20: Share of Output in GVCs by
Investment, 1970–2022 Sector, 2021
Percentage of GDP Percentage of output
8.0% 40%
7.0% 35%
6.0% 30%
5.0% 25%
4.0% 20%
3.0% 15%
2.0% 10%
1.0% 5%
0.0% 0%
CRI DOM SLV GTM HND NIC PAN VNM PAN NIC MEX HND CRI SLV IND DOM GTM COL
1970–1989 1990–2009 2010–2022 Agriculture Manufacturing Services
Source: World Development Indicators data, World Bank. Source: Borin, Mancini, and Taglioni 2021.
Note : CRI = Costa Rica; DOM = Dominican Republic; GDP = gross Note: COL = Colombia; CRI = Costa Rica; DOM = Dominican Republic;
domestic product; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; GTM = Guatemala; HND = Honduras; IND = India; MEX = Mexico;
PAN = Panama; SLV = El Salvador. NIC = Nicaragua; PAN = Panama; SLV = El Salvador; VNM = Viet Nam.
Consistent with the global findings, the association between the RTI index and participation in GVCs
is positive in CADR countries. Positive correlations between the RTI at the country, sector, and year
level and the corresponding value of GVC measures indicate that a higher value in GVC activities favors
routine-intensive employment. Notably, the positive correlation primarily arises for the forward GVC
measure than when using the backward GVC measure, meaning that routine-intensive employment is
more common when value added is generated within domestic supply chains without imported inputs
(figure 4.21a–b).54
54 This is consistent with global cross-country comparisons (Lewandowski, Madoń, and Winkler 2023).
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 64
FIGURE 4.21: Relationship Between Routine Task Intensity Index and GVCs, 2010–2021
a. Pure backward GVC-related trade b. Pure forward GVC-related trade
1.5 1.5
1.0 1.0
RTI index
RTI index
0.5 0.5
0 0
-0.5 -0.5
0 1000 2000 3000 0 500 1000 1500
Pure backward GVC related−trade (US$ of 2019) Pure forward GVC related−trade (US$ of 2019)
Sources: Borin, Mancini, and Taglioni 2021; SEDLAC (CEDLAS and The World Bank); WITS; PIAAC 2017.
Note: The years are 2011–2021 for Costa Rica, 2010–2021 for the Dominican Republic and El Salvador, 2010–2019 for Guatemala and Honduras, 2010–2014
for Nicaragua, and 2011–2018 for Panama. GVC = global value chain; RTI = routine task intensity.
Though the industrial sector has shrunk in every CADR country except Honduras and Panama,
international trade has contributed positively to manufacturing and so likely to the growth of routine-
intensive employment. As previously described, the industrial sector of every CADR country except
Honduras and Panama shrank between 1991 and 2019. However, trade has counteracted this trend.
Recent research on Central America decomposes the factors leading to deindustrialization and finds that
international trade had a mitigating influence, meaning that it promoted industrial expansion (Sinha 2022).
Looking at each country separately, international trade contributed positively to industrial employment in
Costa Rica, Guatemala, Honduras, and Panama but negatively in El Salvador and Nicaragua, though the
magnitude of the effect was small in these two countries.
Robot Adoption and Offshoring
Recent evidence suggests that technological progress abroad could reduce the reliance of more
developed countries on (routine-intensive) production in developing countries. There is some evidence
that offshoring of goods production has reached a peak and that reshoring has increased, though this
is uncertain due to measurement challenges (Baldwin 2022; Feenstra 2017; Krenz and Strulik 2021).
Automation in more developed countries is a possible mechanism driving this change: automation can
create labor savings that incentivize companies to bring jobs back for domestic production or create
new jobs domestically instead of abroad. Indeed, cross-country research shows that robots increase
reshoring and decrease offshoring (Carbonero, Ernst, and Ekkehard 2020; De Backer et al. 2018; Krenz,
Prettner, and Strulik 2021). This phenomenon has been documented for several LAC countries. Adoption
of robots in the United States, for instance, has been found to have negative employment effects in Brazil,
Colombia, and Mexico (Artuc, Christiaensen, and Winkler 2019; Faber 2020; Kugler et al. 2020; Stemmler
2019).
Robot adoption in the United States has a negative effect on labor markets in most CADR countries.
Following the previous literature in this area, we examine the impact that adoption of robots in the United
States has on employment in CADR countries.55 The United States is the main export partner for all CADR
countries except Panama, so any shift in offshoring patterns in the United States would have a substantial
effect on these countries. We are able to undertake the analysis for Costa Rica, the Dominican Republic, El
55 See appendix F for full results and a detailed description of the methodology.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 65
Salvador, and Honduras. The period of analysis is 2010 to 2019. Overall, adoption of robots in the United
States has a negative impact on employment in the region, likely as a result of reduced opportunities for
offshoring. In the Dominican Republic and El Salvador, the average yearly change in CADR countries’
exposure measure to robot adoption in the United States is associated with a decline of 0.4 percentage
points in labor force participation and employment for workers with medium education. In Costa Rica, the
effects of robot adoption are channeled through an increase of 0.2 percentage points in unemployment
for low- and medium-educated workers. Honduras is an exception. Robot adoption led to an increase of
0.2 percentage points in the labor force participation of medium-educated workers there. This may be
linked to the composition of Honduras’s export basket to the United States, which is dominated by raw
materials (42 percent of total exports in 2021) and could allow the country to benefit where robot adoption
leads to expansions in demand.
Still, factors beyond automation will have a significant influence on the participation of CADR
countries in the global economy. Despite evidence of reshoring, the phenomenon remains limited.
Indeed, offshoring continues with evidence of growth from the United States through the 2010s (Krenz,
Prettner, and Strulik 2021). Global instability related to trade tensions between the United States and
China and to armed conflict, increasing labor costs, and trade tariffs in China, as well as a growing desire
to limit supply chain vulnerability after COVID-19 pandemic-related disruptions are shifting decisions
about where to invest to countries closer to the FDI source (IFC 2023). The CADR region’s proximity to the
United States will likely continue to make its countries good candidates for nearshoring, which increased
from 15 percent of total United States FDI between 2003 and 2006 to 18 percent between 2016 and 2019
(OECD 2020b). As was discussed earlier in the chapter, the growing importance of trade in services will
create new opportunities for offshoring. Finally, there is evidence that robots themselves can have a
positive relationship to FDI. Robot density in high-income economies increases FDI stock in developing
countries, though this positive impact declines once a certain level of robot density is reached (Hallward-
Driemeier and Nayyar 2019).56
Robot Adoption and Migration
Technological advances could reduce the demand for migrant workers from CADR countries.
CADR migrant workers might experience a decline in demand if they are employed in industries where
automation is occurring. This may even lead to wage declines if these previously employed migrants
seek out less-exposed industries and increase labor supply. Automation may also lead to a smaller inflow
of migrants. For instance, the end of the bracero agreements that had allowed agricultural workers from
Mexico to work in the United States did not increase domestic employment or wages in part because
employers changed production technology or production levels (Clemens, Lewis, and Postel 2018).57
Instances of increased immigration enforcement that have reduced the supply of migrant workers have
also been shown to lead to increases in automation (Charlton 2023; Ifft and Jodlowski 2022; Konstandini,
Mykerezi, and Escalante 2014).58
On the other hand, technological advances may alter the type of migrants demanded without reducing
the level of migrant flows. As described in chapter 2, automation can increase employment in some
cases; for example, when automation-linked labor savings result in increased demand. Migrants may
be beneficiaries of this increased employment if increases are concentrated in sectors demanding the
skills in which migrants tend to specialize. For instance, migrants in the United States tend to specialize
in manual-intensive jobs that are in relatively high demand as routine tasks are automated. Evidence for
this is presented in chapter 3, which shows how migration from CADR countries has increased in recent
decades and become more concentrated in jobs intensive in nonroutine manual (and interpersonal) tasks.
Evidence from Germany and the United States shows that the spread of personal computers, robots, and
56 See also Artuc, Bastos, and Rijkers (2023) and Maloney and Melina (2019).
57 See also Lewis (2011).
58 Studying the relationship between robots and internal migrants in China, Giuntella (2019) finds that the population share of internal migrants
declined in provinces with more industrial robots.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 66
AI either does not decrease immigration inflows or actually lead to more immigration, in part as a result
of migrants specializing in certain (manual-intensive) tasks (Basso, Peri, and Rahman 2020; Giesing and
Rude 2022; Mandelman and Zlate 2022).
Robot adoption in the United States had no observable impact on overall migration flows from CADR
countries to the United States in the 2000s and 2010s. Following the previous literature in this area,
we examine the impact that adoption of robots in the United States has on inflows of CADR migrants and
on the employment outcomes of these migrants.59 We use data on robot adoption at the sector level
from the International Federation of Robots and focus on impacts on local markets known as commuting
zones during two periods, the second of which (2010 to 2019) saw significantly more robot adoption than
the first (2000 to 2010).60 The adoption of robots does not affect the overall number of migrants from
CADR countries to the United States. However, robot adoption does have an effect on the distribution of
migrants by their level of education in the second period and on labor market indicators in both periods.
Robot adoption in the United States in the early 2000s increased the demand for high-skilled CADR
migrants but in low-skilled, nonroutine occupations. During the first period between 2000 and 2010, an
increase of one robot per thousand industrial workers was associated with a higher employment rate and
a lower unemployment rate for high-educated CADR migrants. This could be the result of complementarity
between high-educated CADR migrant workers and robots or of an increase in aggregate demand due
to a surge in productivity. The evidence indicates that nearly half of highly educated CADR migrants
work in occupations that only require low- to mid-skilled qualifications, compared to about 20 percent
of nonmigrant workers. We speculate that the increase in their employment rate is related to a switch to
lower-skilled, nonroutine, manual (less automatable) jobs, particularly in food preparation and building
and grounds cleaning and maintenance, where a demand expansion linked to automation created more
employment.
Robot adoption in the United States between 2010 and 2019 led to less demand for high-educated
CADR migrants. In the second period between 2010 and 2019, the results go in the opposite direction.
The employment rate of high-educated CADR migrants declined 0.4 percentage points for each additional
robot, while the total number of high-educated CADR migrants fell by 53 migrants for each additional robot
adopted. The negative employment impact could be a direct effect of robot adoption if high-educated
CADR migrants were employed in those industries more exposed to automation during this period, or of
a weakening of their response to the demand expansion effect that seems to have been at play in the first
period. The evidence shows that the participation of high-educated CADR migrants in jobs requiring lower
levels of education stopped increasing and even declined during this period. Additionally, the change
in the industrial distribution of employment of high-educated CADR migrants and the change in robot
adoption has a correlation close to zero, suggesting that labor outcomes of high-educated CADR migrants
continued to be driven mainly by the aggregate demand implications of automation.
59 See appendix F for a detailed description of the methodology.
60 Around 170,000 robots were adopted in the United States between 2010 and 2019, versus around 100,000 between 2000 and 2010.
Chapter 4 Barriers to and Enablers of the Future of Work in CADR Countries 67
Chapter 5
Policy Recommendations
This chapter sets forth policy priorities to help facilitate technological change while mitigating the
negative effects of disruptions that occur as a result of this progress. Supporting technology adoption
and diffusion will be key to CADR countries’ continued development. This will need to be done taking
into account the unique circumstances in CADR countries, particularly the persistence of agricultural
employment in several and the rapid shift to services employment across the region. Business advisory
and related services can promote technology uptake, improve the management skills that are often a
prerequisite of technology adoption, and improve linkages between small businesses and the digital
platforms that can open new markets. From a labor market perspective, two priorities emerge for CADR
countries to take advantage of the benefits of technological progress while mitigating the downsides.
First, pathways for developing skills complementary to new technologies will be essential. Second, social
protection and labor market policies will need to be adapted to new working arrangements on one hand
and to the disruptions associated with technological progress on the other. Within these adaptations lies
a short-term opportunity to expand access to social protection by developing partnerships among digital
platforms, governments, and service providers.
SETTING THE STAGE: FACILITATING TECHNOLOGICAL PROGRESS FROM WHICH
WORKERS CAN BENEFIT
CADR countries have not yet experienced significant disruptions from technological progress, nor
have they taken full advantage of its potential benefits. Chapter 3 and chapter 4 provided empirical
evidence of the relatively limited viability of automation in CADR countries. Labor market disruptions
associated with technological change have been limited thus far in CADR. While many workers in the
region seem to be at risk of losing their jobs to machines, a number of factors related to employment
structure, skills, the use of technology, and globalization mean that this risk has not materialized and is
unlikely to do so in the near term. However, this also means that the region is not yet benefiting from the
potential gains associated with technological progress.
The challenge for policymakers in CADR is to balance policies that enable technological progress and
the benefits it brings with policies that mitigate the effects of any ensuing disruptions. Technological
progress is a key component of development, allowing for increased productivity, new products and
services, and access to new markets. Chapter 2 provided evidence that in many cases, technological
progress goes hand in hand with better employment outcomes. Given the low adoption and diffusion
of technology in CADR countries described in chapter 4, supporting adoption and diffusion will be key
to facilitating CADR countries’ continued development. From a labor market perspective, two priorities
emerge for CADR countries to take advantage of the benefits of technological progress while mitigating
the downsides. First, pathways for developing skills complementary to new technologies will be essential.
Second, social protection and labor market policies will need to be adapted to new working arrangements
on one hand, and to the disruptions associated with technological progress on the other. The region’s
68
TABLE 5.1: Policies to Facilitate Technological Change and Mitigate the Negative Effects of
Resulting Disruptions
Objective 1: Promote the adoption and diffusion of technology by building firm capabilities
Building strong • Ensure quality infrastructure (e.g., electricity, internet service, mobile networks)
foundations • Promote competition, ensure regulations enable technology adoption and
diffusion, and expand access to finance
All CADR countries can work to CADR’s less developed countries CADR’s more developed
can focus more on countries can focus more on
Strengthen business advisory and Utilizing extension services to Targeting services through
technology extension services, and increase technology uptake in the assessments of market failures,
technology centers to promote agricultural sector strength of demand, and risk of
technology uptake among firms and overcrowding the market
improve management capabilities
Promote SME use of digital Piloting initiatives that develop digital Developing initiatives that help
platforms by increasing digital skills, skills among SMEs connect SMEs to overseas
awareness, and technology uptake markets
Objective 2: Strengthen pathways for skills development and deployment
Building strong • Invest in early childhood education and strengthen basic literacy and numeracy
foundations for school-age children
All CADR countries can work to CADR’s less developed countries CADR’s more developed
can focus more on countries can focus more on
Develop labor market insight tools Introducing or strengthening Deploying vacancy, skills profiling,
to collect, analyze, and disseminate labor force surveys and utilizing and other specialized surveys and
information about the labor market administrative data exploring novel sources of labor
market information (for example,
online job postings)
Build foundations-driven, demand- Piloting remedial skills and demand- Identifying areas of growing
oriented education and training driven training programs that improve demand in real time, developing
systems that are designed to be literacy and numeracy and basic training programs in response,
lifelong and targeted to workers digital skills and that fill labor market and incorporating other support
at greater risk of labor market demand in strategic areas services into these training
disruptions programs
Design digitally enabled, fit-for- Building a public employment Expanding the public employment
purpose intermediation programs services system that is a reliable services system to provide labor
that focus on overcoming information source market intelligence, career and
geographic disparities and Exploring global skills partnerships skills guidance, job matching, and
information problems (GSPs) to create safe and inclusive referral services
migration pathways
Objective 3: Adapt social protection and labor market policies to new forms of work
Building strong • Move away from reliance on traditional employer-employee relationships for
foundations financing and providing social protection
All CADR countries can work to CADR’s less developed countries CADR’s more developed
can focus more on countries can focus more on
Exploit the potential of platform Avoid regulations that lead to further In the short term, explore models
work labor market segmentation for expanding access to social
protection to platform workers
Monitor potential anti-competitive Develop models of the businesses, Develop more sophisticated
practices characteristics, and anti-competitive analyses of the particular
potential of digital platforms anticompetitive practices of
platforms, especially issues
related to data
Chapter 5 Policy Recommendations 69
more developed countries—Costa Rica, Panama, and, in many respects, the Dominican Republic—are
generally at a more advanced stage of structural change and technology adoption. This means that
different CADR countries have somewhat different priorities (table 5.1). The region’s less advanced
countries need to focus more on building foundational structures and piloting new initiative while the
region’s more advanced countries can work to improve the sophistication of existing systems.
PROMOTE THE ADOPTION AND DIFFUSION OF TECHNOLOGY BY BUILDING FIRM
CAPABILITIES
CADR countries lag in the adoption and diffusion of technology. Chapter 4 describes the deficits in
technology use by CADR workers and firms. Overall measures of adoption of digital technologies as
well as measures related to adoption of frontier technologies show that all countries in the region are
substantially behind advanced economies.
A range of barriers can interrupt a firm’s adoption of new technology. Good infrastructure—electricity,
internet service, and mobile networks—is key to enabling technological progress. Beyond these basic
enabling factors, firms face several challenges. Recent research by the World Bank organizes the drivers
of technology adoption at the firm level into factors internal and external to the firm (table 5.2). External
factors include market conditions and regulations, access to finance, and the supply of knowledge and
human capital. Internal factors are firm capabilities that include skills, management and organizational
practices, and the informational and behavioral biases of entrepreneurs and managers. The assessment
makes clear that, far from just pushing certain types of technology, policies to support technology
adoption must focus on improving the availability of information and the abilities of firms to manage and
learn, especially from external sources (Caselli and Colemann 2006).
TABLE 5.2: Drivers of Technology Adoption Among Firms
Factors Examples
External Competition, demand, • Competition creates incentives to adopt technology.
regulations
Access to finance • Access to finance permits firms to finance investment in technology.
Access to knowledge • Knowledge transfer from other firms, within the firm, from consultant
and human capital firms, or from exporting can create technology demand.
Internal Information and • Biases like reference group neglect can lead entrepreneurs to believe
behavioral biases they are already investing in more sophisticated technology.
• Lack of information about returns to technology, difficulty evaluating
uncertainty, or lack of knowledge about use of technology can lead a
firm not to adopt new technology.
Management and • Low management quality leads to less technology adoption.
organization
Know-how and skills • Human capital in a firm is necessary for the adoption of more
capabilities sophisticated technologies.
Source: Cirera, Comin, and Cruz 2022.
Governments can support technology uptake among firms. The point of departure for policymakers
is ensuring that infrastructure of sufficient quality is in place to allow firms to adopt more sophisticated
technologies and that regulations do not create challenges to adoption (Cirera, Comin, and Cruz 2022;
Dutz, Almeida, and Packard 2018). Beyond this, a range of policies can be used to address the internal
and external drivers of technology adoption. This section focuses on policies that can support the skills
and knowledge needed inside and outside of the firm to facilitate uptake, though policies that increase
Chapter 5 Policy Recommendations 70
access to finance, subsidize technology, or increase competition are also crucial (box 5.1). In any case, an
assessment of the market failures justifying government intervention and a diagnostic of the causes of
underutilization of technology is a critical first step (Cirera, Comin, and Cruz 2022).
BOX 5.1: Promoting Competition, Rightsizing Regulations, and Expanding Access
to Finance to Facilitate Technology Adoption
Promoting competition, rightsizing regulations, and expanding access to finance are key
components of promoting technology adoption and of benefiting from the opportunities that
adoption creates. Competition helps spur technology adoption, particularly where there are
low levels of competition. This is evident in the tendency for greater global linkages to increase
adoption and technological sophistication. Regulation can both facilitate and hinder technology
adoption. Regulations in areas including e-payments, consumer protection, and cybersecurity can
address new issues raised by digital commerce, facilitate remote and cross-border transactions,
and promote trust. At the same time, regulations can restrict digital markets from flourishing when
there are strict restrictions on the goods or services available in digital marketplaces or when
restrictions make data transfer challenging. In these cases, regulations can hinder the expansion
of markets that increase opportunities for small and medium-sized enterprises. Regulations that
impose taxes and tariffs on ICT can hinder technology adoption with knock-on effects on GDP
growth. Finally, insufficient access to finance can impede technology adoption if firms need
to undertake costly technology investments on their own. Overall, to address these issues an
adaptable regulatory framework is needed with policies that promote competition and access to
finance, ranging from careful consideration of any taxes and tariffs on ICT to efforts to increase
the financial sector’s knowledge of and comfort with financing for technology-related projects.
Sources: Cirera, Comin, and Cruz 2022; Dutz, Almeida, and Packard 2018; Jaller, Gaillard, and Molinuevo 2020.
Note: GDP = gross domestic product; ICT = information and communication technology.
Business advisory services, technology extension services, and technology centers can promote
technology uptake.61 Business advisory services provide advice on a range of general business functions
ranging from human resources to accounting and marketing and advertising. The services are typically
targeted to SMEs, which often lack skills and knowledge across these areas. More effective models begin
with a diagnostic phase followed by the development of an action plan. Digital upgrading programs
are a business advisory service devoted to facilitating the uptake of digital technologies. A subset of
these programs focuses on linking firms to digital platforms by providing support in areas like customer
orientation, pricing, and reputation maintenance. Technology extension services are another option that
tends to focus on the demand for and the skills needed to utilize more sophisticated, sector-specific
technologies. They typically function via field offices and extension staff. Finally, technology centers
tend to be sector-specific and help develop new technologies or adapt existing ones and provide the
skills needed to adopt these technologies. Technology centers generally focus more on the diffusion
of technologies in developing countries than on research and development. Evidence on the impacts
of each of these three types of services is rare, though suggestive of positive effects on upgrading
technology. Important considerations for policy are factors that can limit effectiveness, which include the
availability of expertise, the strength of demand among the firms most in need of support, and the risk of
overcrowding the market for business support services. The importance of taking these limiting factors
into account makes these types of services better suited for CADR’s more advanced countries.
61 This paragraph is based on Cirera, Comin, and Cruz (2022).
Chapter 5 Policy Recommendations 71
Special focus is needed in CADR’s less advanced economies on how technology can help modernize
agriculture. Agricultural sectors in CADR remain large as a share of employment relative to more
advanced economies, particularly in Guatemala, Honduras, and Nicaragua. A focus on uptake of digital
technologies and automation in agriculture via extension services could help spur growth (Yusuf 2017).
Low adoption of technology in CADR’s agricultural sectors is likely undermining agricultural productivity,
given the strong link between the two (Fuglie et al. 2020). For example, in El Salvador, technological
change and technical efficiency (a proxy for managerial performance) have been detrimental to agricultural
productivity (Bravo-Ureta et al. 2022). This points to the potential utility of extension services that can
enhance the adoption and diffusion of technologies and improve managerial abilities. Provision of these
services is made more important by the negative impacts of climate change on agricultural productivity
and the need for mitigation and adaptation strategies that themselves require new technologies (box 5.2)
(Bravo-Ureta et al. 2022; Lachaud, Bravo-Ureta, and Ludena 2017). Although evidence for the impacts
of business support services on technology uptake in general is limited, positive results are common
for agricultural extension services (Cirera, Comin, and Cruz 2022). Chapter 4 provided examples of how
digital technologies ranging from SMS messages to remote sensors can help increase the productivity
of smallholder farmers.
BOX 5.2: Green Jobs in CADR
Shifting to greener growth strategies will involve both the development of new technologies
and the disruption of labor markets. Climate change is causing substantial economic damage
and reducing productivity in CADR countries as the frequency and intensity of extreme weather-
related events increase and slow onset effects like temperature increase take hold (World Bank
2022a). Despite these challenges, CADR countries have opportunities for more sustainable,
low-carbon development (World Bank 2023a; 2023b). This transition will require adopting
cleaner, lower-carbon technologies. As with any technology-related transition, however, there is
potential for labor market disruption. The extent to which a shift towards greener growth could
affect labor markets in CADR countries can be evaluated by looking at the green task intensity
(GTI) of jobs in the region—that is, how many jobs involve green tasks currently (narrow GTI)
and how many could be green if greener technologies were adopted (broad GTI) (Granata and
Posadas 2022).62
Green jobs make up a small share of employment in CADR countries currently, but greener
technologies could expand the scope of green employment substantially, particularly in
agriculture. Between 3 and 6 percent of employment in Costa Rica, the Dominican Republic, El
Salvador, Honduras, and Panama is in jobs that currently involve environmentally friendly tasks
(figure B5.2.1). This share rises significantly when jobs with tasks that could be environmentally
friendly, given greener technologies, are considered. This compares with estimates of between
2 and 15 percent of employment in Indonesia and 4 and 41 percent in Viet Nam using the same
methodology, and between 2 and 3 percent in the United States using an alternative methodology
(Doan et al. 2023; Granata and Posadas 2022; Vona, Marin, and Consoli 2019). The large increase
under the broad measure is driven by agriculture: with more environmentally friendly technologies,
employment in agriculture, which accounts for a substantial share of employment in the CADR
countries considered, would be a source of many green jobs (figure B5.2.2). Notably, men are
more likely to work in green jobs than women, and the introduction of green technologies would
primarily expand the employment of men in green jobs rather than women.
62 Green tasks reduce a consumer’s and/or firm’s environmental impact (Granata and Posadas 2022).
Chapter 5 Policy Recommendations 72
BOX 5.2: Green Jobs in CADR (continued)
FIGURE B5.2.1: Share of Green Jobs in CADR Countries Overall and by Gender, 2021
Percentage of employment
70%
60%
50%
40%
30%
20%
10%
0%
All
Women
Men
All
Women
Men
All
Women
Men
All
Women
Men
All
Women
Men
All
Women
Men
All
Women
Men
All
Women
Men
All
Women
Men
All
Women
Men
CRI DOM HND PAN SLV CRI DOM HND PAN SLV
Narrow Broad
Sources: Granata and Posadas 2022; SEDLAC (CEDLAS and The World Bank).
Note: Data are 2019 for Honduras. CRI = Costa Rica; DOM = Dominican Republic; HND = Honduras; PAN = Panama; SLV = El Salvador.
FIGURE B5.2.2: Share of green jobs in CADR countries by sector, 2021
Percentage of employment
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
DOM HND PAN SLV DOM HND PAN SLV
Narrow Broad
Sources: Granata and Posadas 2022; SEDLAC (CEDLAS and The World Bank).
Note: Data are 2019 for Honduras. Data is not available by sector for Costa Rica.
Data from LinkedIn suggests that green jobs are prevalent in some sectors in Costa Rica.
The professional networking site LinkedIn calculates a measure of green talent as the share of
LinkedIn members in an industry working in a green job or who have at least one green skill. Data
is only available for Costa Rica, which performs favorably compared to Argentina and the United
States on green talent. An average of 15 percent of LinkedIn members in Costa Rica have green
talent versus 13 percent in Argentina and 14 percent in the United States. The industries with the
most green talent are farming, construction, and oil and gas—the same as in Argentina and the
United States—while those with the least are retail, wholesale, and finance (figure B5.2.3).
Chapter 5 Policy Recommendations 73
BOX 5.2: Green Jobs in CADR (continued)
FIGURE B5.2.3: Green Talent in Costa Rica and Comparator Countries, 2023
Percentage of employment
Accommodation & Food Services
Administrative & Support Services
Construction
Consumer Services
Education
Entertainment Providers
Farming, Ranching, & Forestry
Financial Services
Government Administration
Holding Companies
Hospitals & Health Care
Manufacturing
Oil, Gas, & Mining
Professional Services
Real Estate & Equipment Rental
Retail
Technology, Information, & Media
Transportation and Logistics
Utilities
Wholesale
0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50%
ARG CRI USA
Source: LinkedIn.
At the more sophisticated end of the skills spectrum, management skills are a key part of the ecosystem
needed for technology adoption. As described in chapter 4, good management is increasingly viewed as
a prerequisite for technology adoption and a key determinant of firm performance. Improved capabilities
were an important part of the substantial growth experienced by countries in East Asia (Cirera and
Maloney 2017). Business trainings that seek to improve management practices generally have positive
impacts, though given their typically short duration, impacts are expected to be fairly small (McKenzie
2021). New approaches to such trainings could improve effectiveness and scale the programs to larger
numbers of small and medium-sized firms. These alternatives include adapting programs to local contexts,
targeting programs to women, and taking advantage of technology to expand the audiences reached.
Policy can also help small and medium-sized businesses take advantage of opportunities in digital
trade in services by increasing their use of platforms. Platforms create an opportunity for small
businesses to expand their customer and supplier bases but accessing and utilizing them can be
challenging for inexperienced firms. SMEs in all CADR countries with data available cite the need for more
technical skills and knowledge as one of the top three challenges when using or trying to adopt digital
platforms (table 5.2). Though a range of policy instruments is available to facilitate adoption and use of
platforms, any public program must be clear about its motivation and the market failure being addressed.
Platforms themselves have motivations for making themselves attractive to users and even to provide
training. The most common policy objective for programs promoting SME use of platforms is to increase
digital skills and awareness and uptake of technologies (OECD 2021). Several OECD countries have
developed initiatives that help SMEs evaluate how they could benefit from digital technologies, provide
information on platforms, and instruct SMEs on how to utilize platforms to their advantage. These include
Australia’s Guide to Digital Transformation, Denmark’s E-Commerce Center, and the United Kingdom’s
Chapter 5 Policy Recommendations 74
E-Exporting Programme that helps support businesses to sell overseas online. There is some evidence
that programs of this type are effective. A recent impact evaluation of a large-scale training program for
new e-commerce sellers finds that sellers receiving training in practical skills in online business operations
earned higher revenue and attracted more consumers (Jin and Sun 2021).
TABLE 5.3: Main Challenges Cited by SMEs When Using or Trying to Adopt Digital
Platforms, 2022
Percentage of firms citing challenge as one of top 3 challenges
Country Top 3 challenges
Costa Rica 1. Need for more technical skills and knowledge (34%)
2. Extreme or unfair competition (32%)
3. Fees for accessing digital platforms (23%)
Dominican Republic 1. Need for more technical skills and knowledge (35%)
2. Need for more resources or infrastructure (for example, broadband) (23%)
3. No challenges (22%)
El Salvador 1. Need for more technical skills and knowledge (40%)
2. No challenges (24%)
3. Extreme or unfair competition (23%)
Guatemala 1. Extreme or unfair competition (33%)
2. Fees for accessing digital platforms (31%)
3. Need for more technical skills and knowledge (28%)
Honduras 1. Need for more technical skills and knowledge (37%)
2. Fees for accessing digital platforms (28%)
3. Extreme or unfair competition (26%)
Nicaragua 1. Need for more technical skills and knowledge (32%)
2. No challenges (22%)
3. Extreme or unfair competition (21%)
Source: Future of Business Surveys.
Note: Data are not available for Panama. Adoption is for the sale or purchase of goods and services.
STRENGTHEN PATHWAYS FOR SKILLS DEVELOPMENT AND DEPLOYMENT
Mitigating the disruptions and taking advantage of the opportunities of technological progress
will require a special focus on investments in human capital. Chapter 4 described the human capital
quantity and quality challenges facing CADR countries that are contributing to low technology adoption,
a phenomenon that has long been identified as a challenge in LAC (Maloney 2002). Ensuring that
current and future workers are equipped with the right skills to complement new technologies is an
essential component of taking advantage of the growth made possible by technological advancement.
Strengthening skills development pathways in basic, secondary, and tertiary education for children and
youth as well as in technical and vocational education and training (TVET) and upskilling and reskilling
programs for adults, will be critical (Ferreyra et al. 2021). To the extent that the demand for tasks and skills
is changing in CADR countries, demand-driven TVET for future workers and demand-driven upskilling and
reskilling for current workers will be important. At the same time, incorporating the socioemotional and
digital literacy skills that are becoming essential across occupations throughout the learning cycle from
early childhood development to training programs will be essential.
Building human capital is necessary but insufficient, as people also need to be able to deploy their
investments in human capital effectively. Skills building programs generally have a positive impact on
employment and earnings, but designing them with labor market insertion in mind is critical to ensuring
Chapter 5 Policy Recommendations 75
that these impacts are meaningful and cost-effective. The impact of training programs on employment and
earnings in developing countries, including LAC, is positive but modest (Escudero et al. 2019; McKenzie
2017). More promising programs tend to target sectors with growing labor market demand, combine
training with complementary services including labor market intermediation and financial support,
take into account the challenges faced by particular groups, and address mismatches in skills across
geographies and sectors (Carranza and McKenzie 2023; Ferreyra et al. 2021; Katz et al. 2022; Kluve et al.
2019; McKenzie 2017; Stöterau 2019). For labor market intermediation programs in particular, encouraging
job search in different places, updating a candidate’s beliefs, and improving a candidate’s ability to signal
their skills are linked to greater effectiveness (Carranza and McKenzie 2023). These efforts need to be
adapted to the unique challenges facing different groups.
Improving skills development pipelines requires strengthening three aspects of employment support
services. First, labor market insight tools collect, analyze, and disseminate information about the labor
market to help design employment programs that are able to target disadvantaged groups, funnel
beneficiaries to sectors in growing demand, and update beliefs and correct biases about the labor market
that may hold individual jobseekers back. Second, foundations-driven, demand-oriented education and
training systems provide the foundational abilities and in-demand skills that current and future workers
need to complement technological progress. Finally, digitally enabled, fit-for-purpose intermediation
programs help deploy human capital effectively by complementing private sector offerings, utilizing
technology to expand the scope and efficiency of service delivery, and focusing on service gaps and
underserved groups.
Deploy Labor Market Insight Tools
Strengthening skills pathways begins with identifying the skills that are needed to facilitate
technological progress and the skills that are needed as technology advances. This means enhancing
efforts to anticipate and respond to needs for more specialized skills.
Developing tools that can rapidly identify private sector demand is more important than identifying
specific skills in advance. Technological change means that skill needs change rapidly. For example,
cross-country evidence suggests that vocational education, which tends to provide training in specific
skills, assists with entry into jobs but can then reduce a worker’s adaptability to technological change
(Hanushek et al. 2017).63 Tools to identify in-demand skills can help policymakers inform training and
educational institutions about the kinds of skills that are currently in demand in the labor market to help
create skills development pathways that are adaptable and that complement technological progress (Isik-
Dikmelik et al. 2022). In Switzerland, for example, updated education curricula are shown to hasten the
arrival of new technologies in firms (Schultheiss and Backes‑Gellner 2022b). Crucially, these tools can help
target limited training resources to areas where demand is strong and growing. A variety of instruments
are available to undertake such identification. These vary in sophistication and can be deployed
in a sequential manner over time to create a fuller picture as labor market information capacities are
developed (table 5.3). In the case of digital skills in particular, the World Bank has developed a guidebook
for developing digital skills country action plans that includes guidance on how to assess demand for and
supply of digital skills in environments where data is limited (World Bank 2021a; 2021b).
63 Similarly, graduates in science, technology, engineering, and math majors in the United States enjoy high wages upon entering the labor
market but over time technologies replace tasks using these skills leading to lower returns (Deming and Noray 2018).
Chapter 5 Policy Recommendations 76
TABLE 5.4: Tools for Identifying In-Demand Skills
Tool Objective Examples
Critical occupations list • Combine quantitative and qualitative data sources • Malaysia
to identify occupations that are in shortage • Indonesia
Occupational employment • Firm survey collecting data on demand at the • Indonesia
and vacancy survey detailed occupational level • United States
Survey of detailed skills and • Worker surveys collecting data on skills and tasks • European Union
tasks • Indonesia
• United States
• Vietnam
Online vacancy data • Collect online job postings that have information on • Australia
skills, education, and experience requirements and • Indonesia
can also be used to assess demand • United States
Source: Based on Posadas and Testaverde 2022.
Build Foundations-driven, Demand-oriented Education and Training Systems
Increasing the skill level of the workforce is necessary to complement technological change. For
future workers, this begins with enhancing the quality and increasing the availability of early childhood
education (Vegas and Santibáñez 2009), improving the quality of basic education so that students
master foundational skills in the early years of schooling (World Bank 2022e), and increasing the share
of young people who obtain secondary and post-secondary education and the quality of their learning in
these years (OECD 2023). For current workers, this means adopting training approaches that build both
foundational and specialized skills that are demanded in the labor market.
Foundational skills are essential for both current and future workers. Tasks that involve foundational
skills like literacy, numeracy, and socioemotional skills are key to unlocking the benefits of technological
progress both because they form the building blocks of more advanced, specialized skills and because
they complement other skills (Levin et al. 2023). Without strong foundations, it is difficult for students and
workers to benefit from more advanced training. In Mexico, for example, scholarship programs to encourage
youth to stay in school failed to raise learning levels because the young people lacked the foundational
skills to succeed in secondary education (de Hoyos, Attanasio, and Meghir 2019). Evidence indicates that
foundational skills are important complements for new technologies (Dalvit et al. 2023; Cunningham et al
2022; World Bank 2018). In the United States, for example, jobs requiring high levels of social interaction
grew substantially between 1980 and 2012, while there is evidence of growing complementarity between
social and cognitive skills, which are correlated with pay and firm performance, perhaps because of
differences in how firms utilize technology (Deming 2017; Deming and Kahn 2018; Weinberger 2014).
Digital skills are a linchpin of efforts to benefit from technological progress. Digital skills are the skills
needed to accomplish tasks with ICT. These can be broken down into a set of basic skills—digital literacy—
used to perform rudimentary tasks (for example, e-mailing, using digital devices, and storing and accessing
digital information); intermediate ones for using professional software for analysis, creation, management,
and design; and advanced ones for undertaking specialized tasks like data science, cybersecurity, or
programming (Cunningham et al. 2022; UNESCO 2017b; IFC 2019). Digital occupations like software
developers, programmers and engineers, and data scientists and engineers are growing quickly (OECD
2022b). At the same time, digital skills are growing in importance across occupation types (Feijao et al.
2021; Cunningham et al. 2022; World Bank 2021a). For example, digital literacy is key in agriculture to enable
small-scale farmers to access and benefit from digital support services like market and weather services
(FAO 2022; Morris, Sebastian, and Perego 2020; Fuglie et al. 2020). Regardless of occupation, platform
workers need basic digital skills to manage the apps through which they access jobs and receive payment.
Chapter 5 Policy Recommendations 77
Education and training can help prepare the workforce to benefit from technological progress while
also assisting workers displaced by automation. Several principles can help ensure that education and
training is impactful in the context of technological progress.
First, education and training should focus on the provision of transferable, particularly foundational,
skills. As previously described, there is increasing evidence of the importance of foundational skills like
basic literacy and numeracy, soft or socioemotional skills like critical thinking and teamwork, and digital
skills. Improving these foundational skills for future workers requires significant effort across the education
system, beginning with early childhood education (box 5.3) and continuing through education for school-
aged children and young people. Improving basic literacy and numeracy can be achieved by improving
student readiness to learn through early childhood interventions; focusing curriculum, assessment, and
instruction on essential learning and foundational reading and math; improving teaching quality and
effectiveness; and aligning school management to focus on foundations (World Bank, 2022h).
BOX 5.3: The Labor Market Benefits of Early Childhood Education
Quality childcare can have long-run positive impacts on labor market outcomes. Quality
childcare is important to enable women to participate fully in the workforce, particularly in CADR
where women’s labor force participation is low and motherhood pushes many women into part-
time and informal jobs (Díaz and Rodríguez-Chamussy 2016). In addition to the short-term benefits
for women’s labor force participation, quality childcare can have positive long-run effects on labor
market outcomes by helping young children develop the cognitive and socioemotional skills that
enable them to reach their full potential as adults (Devercelli and Beaton-Day 2020).
Quality childcare that includes a focus on early childhood education (ECE) can help prepare
young children for school. Children’s brains develop rapidly in early childhood. Deficits in early
years can have lasting impacts. The early learning provided in childcare settings can help children
build the cognitive and language skills needed to succeed when they enter school, as well as the
socioemotional skills that are increasingly critical later in life (Devercelli and Beaton-Day 2020).
These benefits can last into employment in adulthood. A recent study comparing outcomes of
children who benefited from the initial years of the Head Start Early Childhood Education (ECE)
program in the mid-1960s in the United States to those who just missed the introduction of the
program found significant effects on human capital formation and long-run economic outcomes
(Bailey, Sun, and Timpe 2021). Head Start children received 0.65 more years of education and were
12 percentage points more likely to complete college than their peers who did not benefit. Adults
who benefited from the program as children were more successful in the labor market, working
an average of two more weeks per year than nonbeneficiaries. Quality ECE also has significant
positive effects on later life outcomes in less developed settings. A study of 12 low- and middle-
income countries found that adults who participated in ECE programs as children were more likely
to be employed in jobs requiring higher skill levels (Shafiq, Devercelli, and Valerio 2018).
Vocational and short-course training can play an important role in reinforcing foundational skills
for those already in the workforce. For example, in Cambodia, pilots of a skills bridging program to
provide foundational skills to out-of-school youth led 60 percent of the participants to enroll in certificate
programs in technology (World Bank, UNESCO, and ILO 2023). In the Dominican Republic, vocational and
soft skills training provided by the Programa Juventud y Empleo increased employment among women
(Acevedo et al. 2020). Notably, the participants who only received the soft skills training benefited as
much as participants who received the technical training. In Colombia, vocational training participants
Chapter 5 Policy Recommendations 78
who received training emphasizing social skills did not experience the same type of erosion of training
benefits that recipients who received training emphasizing technical skills did (Barrera-Osorio, Kugler, and
Silliman 2023). Digital skills training can teach basic digital literacy or provide more advanced, but still not
firm-specific, technical skills like programming or cybersecurity (box 5.4). A recent impact evaluation of a
coding bootcamp program for women in Argentina and Colombia found that the bootcamps increased
coding skills and the probability of getting a job in the technology sector (Aramburu and Goicoechea
2021). Training of this type may be effective because it addresses a market failure in skills provision—
employers may underinvest in transferable skills that workers could take to other firms—and because soft
and basic digital skills are more durable than technical skills, the demand for which evolves quickly over
time (Carranza and McKenzie 2023; Schultheiss and Backes‑Gellner 2022a).
BOX 5.4: Incorporating Digital Skills Training into Technical and Vocational
Education and Training
TVET systems globally have responded slowly to the growing demand for digital skills. A World
Bank, UNESCO, and ILO (2023) report on TVET highlights digital skills training as a challenge
for TVET systems globally. Many learners enter TVET programs lacking foundation skills. Many
instructors lack proficiency to adequately instruct learners and to ensure that the digital skills
curricula developed in TVET systems respond to labor market demands (Banga and te Velde
2019). The low level of digital skills among both learners and instructors was made clear during the
COVID-19 pandemic when TVET systems struggled to transition to remote learning.
TVET programs can mainstream digital skills into training offerings. TVET programs can integrate
training in foundational skills with training in key digital, socioemotional, and cognitive skills demanded
by the labor market. This can be done via standalone courses that provide instruction in basic
foundational skills first and then layer practical skills training in separate courses, or via courses that
integrate foundational and transversal skills acquisition into the practical course content (World Bank,
UNESCO, and ILO 2023). Programs aiming to support the development of advanced, job-specific digital
skills can integrate those alongside other industry capabilities and soft skills or utilize “boot camp”
models narrowly focused on specific digital skills. For example, an Integrated Skills Development
Scheme by the Ministry of Textiles in India covered both the digital (graphic design and robotics
hardware and software) and soft skills (business and merchandising) needed to enter this sector.
Second, technical training should be directed towards skills provision in areas where skills are in
demand. Beyond training in foundational skills, training is more effective when directed to areas of
demand. For example, sector-targeted training programs in the United States focus on connecting less-
skilled workers with better jobs in industries with increased demand. The programs show positive impacts
on earnings, which is likely related to training in transferable and certifiable skills, combining technical
training with soft skills and career-readiness training, and wraparound support services (Katz et al. 2022).
The tools to identify the in-demand skills previously described can help target industries with strong labor
market demand that are suitable for such sector-focused training.
Third, training should be viewed as a lifelong activity. The constant changes in skills associated with
technological progress mean that skills need to be updated throughout working lives. Lifelong learning
systems recognize that workers will face different labor market landscapes throughout their lives, and
help individuals have more responsibility for their learning pathways. These learner-focused approaches
utilize e-learning to connect trainings to workers at scale; contemplate different financing approaches that
may provide individuals with vouchers to select their own trainings; and adapt training programs to the
needs of different types of learners: for example, recognizing that the pedagogy suitable for adult workers
Chapter 5 Policy Recommendations 79
is different from that suitable for young ones (Bendini, Levin, and Oral-Savonitto 2019). Notably, the kind
of retraining that is required because of technological change and automation may not require learning
entirely new fields, but instead adapting skills to new tasks that emerge within occupations. As noted in
chapter 2, much of the change in work that results from technological progress occurs within occupations
rather than across them. Research from Germany, for example, shows that occupations that evolved from
being routine to less routine experienced wage gains (Bachmann et al. 2022).
Finally, training will need to be targeted to particular types of workers who face greater challenges
recovering from displacement by technology. These groups vary from country to country, but as chapter
3 shows, in CADR, the most vulnerable to disruption are generally less-educated, male, younger, and
rural workers. This is important because there is evidence from OECD countries that workers who are
at higher risk of automation receive less job training (Nedelkoska and Quintini 2018). Still, identifying the
most vulnerable to disruption is insufficient, as different workers have different opportunities after losing
their jobs and when deciding to enter the labor market. For instance, in advanced economies, women
may have fewer options for labor market transitions after technology-related labor market disruptions,
though reskilling can narrow the gap (WEF 2018b). Research from China and Germany shows that
whereas younger people exposed to robots seek out training, older workers tend to drop out of the labor
market (Battisti, Dustmann, and Schönberg 2023; Giuntella, Lu, and Wang 2022). Understanding these
differences is critical to targeting training to those who need it most (box 5.5).
BOX 5.5: The Complex Interaction of Technology, Labor Markets, and Gender
Many factors will influence how technological progress in and outside of CADR countries
interacts with the employment of women in the region, with some factors reinforcing barriers
and others creating opportunities. One set of factors relates to the current distribution of human
capital by gender, including gender differences in education level (Ñopo 2012), field of study (for
example, gender gaps in science, technology, engineering, and mathematics education) (UNESCO
2017a), and digital skills (OECD 2020a; OECD 2018a). Another set relates to gender differences
in how human capital is deployed, including occupational segregation (for example, gender gaps
in management roles) and sectoral segregation (for example, gender gaps in clerical and factory
work and in health and care services) (World Bank 2012a; Sinha 2020); segregation of tasks done
within occupations (for example, gender differences in who does undesirable tasks not associated
with promotion) (Babcock et al. 2017; Chan and Anteby 2017); gender differences in returns to skills
associated with the future of work (Bustelo, Flabbi, and Viollaz 2020); and gender differences in
engagement with new working arrangements like work-from-home and platform work (Berniell et al.
2021; ECLAC and ILO 2021). A final set relates to how gender is mediated through technology itself,
including gender differences in access to and use of ICT (Quirós et al. 2022; Agüero, Bustelo, and
Viollaz 2020); the gender biases and gender neutrality of technology itself (UNESCO 2022); and
the potential for certain technologies to break down physical obstacles to women’s participation
in work that requires physical strength (Mealy, Rio-Chanona, and Farmer 2018). Evidence on each
of these different factors is not available for CADR countries, but some initial insights are possible.
Employed women in CADR are less susceptible to automation. Chapter 3 showed that women
are generally at lower risk than men of automation by computers, AI, and mobile robotics. Several
factors explain this lower risk. First, employed women have higher education levels than men in
CADR, which are correlated with working in less automatable jobs. Second, employment in CADR
countries is gender segregated. Women tend to work in services and sales jobs, which are harder
to automate, while men tend to work as plant and machine operators and assemblers and in crafts
jobs, which are easier to automate. Third, this gender segregation extends beyond paid work to
unpaid work at home. Women are much more likely than men in CADR countries to participate in
Chapter 5 Policy Recommendations 80
BOX 5.5: The Complex Interaction of Technology, Labor Markets, and Gender
(continued)
unpaid work at home. This work is excluded from estimates of automatability, though technologies
like domestic appliances may, in fact, automate these tasks in the same way that tasks done in
paid jobs are automated, which may free women’s time to take on paid work.
The ultimate impact of automation on women’s labor market outcomes in CADR depends not
just on susceptibility to automation, but on how well female workers displaced from their jobs
are able to recover and on how well female workers can benefit from the expansions in demand
that result from automation. Current female workers in CADR seem well placed to take advantage
of technological progress. The higher education levels of employed women mean that their skills
are more likely to be complementary to new technologies. In fact, across all CADR countries,
employed women are already more likely to work in jobs intensive in computers and the internet
(figure B5.5.1). Women are also overrepresented in the kinds of service jobs that may expand as
automation creates spillover benefits. Still, while less susceptible to automation, these service jobs
tend to be low productivity and low wage. Where women are displaced, evidence from advanced
economies suggests that providing retraining can be essential because women may have fewer
options for job transitions after technology-related labor market disruptions (WEF 2018b).
FIGURE B5.5.1: Use of Computers and the Internet at Work by Gender, 2021
Percentage of workers in computer- and internet-intensive occupations
a. Computer use b. Internet use
35% 30%
30% 25%
25%
20%
20%
15%
15%
10%
10%
5% 5%
0% 0%
DOM GTM HND NIC PAN SLV CRI USA DOM GTM HND NIC PAN SLV CRI USA
Women Men
Sources: SEDLAC (CEDLAS and The World Bank); PIAAC 2017.
Note: The years are 2019 for Guatemala and Honduras, 2014 for Nicaragua, and 2018 for Panama. Computer- and internet-intensive occupations
are defined as occupations in the top 25 percent of computer and internet use at work as defined using PIAAC data from comparator countries.
CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador;
USA = United States of America.
On the other hand, women face challenges to benefiting fully from automation. Despite higher
enrolment rates in tertiary education, women in CADR countries are less likely to study in the
science, technology, engineering, and mathematics (STEM) fields that are complementary to many
technological advancements (figure B5.5.2). In the Dominican Republic, for example, women make
up 67 percent of students in tertiary education programs but just under 40 percent of students in ICT-
and engineering-focused tertiary degree programs. Returns to STEM and ICT skills are also higher for
men (Bustelo, Flabbi, and Viollaz 2020). Social norms related to women’s work will play a critical role in
how women benefit, or do not, from technological progress. Research on Costa Rica, El Salvador, and
Panama shows that lower obstacles to human capital accumulation for women led to improvements in
female labor market participation and increased women’s participation in high-income jobs between
Chapter 5 Policy Recommendations 81
BOX 5.5: The Complex Interaction of Technology, Labor Markets, and Gender
(continued)
1995 and 2015 (Sinha 2020). However, increases in labor market discrimination in Costa Rica and
Panama muted this transformation. Norms could thus inhibit the kinds of labor market transitions
necessary for women in CADR to benefit from technological progress.
FIGURE B5.5.2: Female Enrolment in Tertiary Programs and in ICT- and Engineering-
Related Tertiary Degree Programs, 2021
Percentage of students who are female
80%
70%
60%
50%
40%
30%
20%
10%
0%
CRI DOM SLV GTM HND NIC PAN
Tertiary ICT Engineering
Source: UNESCO Institute for Statistics for tertiary enrolment; UNESCO (2020) for ICT and engineer enrolment.
Note: The years are 2019 for Costa Rica and Honduras and 2020 for El Salvador and Guatemala. Data is not available for ICT and Engineering for
Nicaragua. CRI = Costa Rica; DOM = Dominican Republic; GTM = Guatemala; HND = Honduras; NIC = Nicaragua; PAN = Panama; SLV = El Salvador.
Changing working arrangements are likely to have different, and perhaps more substantial,
impacts on women than on men in CADR countries. Chapter 3 showed that a higher share of
women than men in CADR countries works in jobs that can be done from home. Expansions in the
adoption or diffusion of ICT technologies are then likely to disproportionately enable remote work
for women. Chapter 3 also showed that women are frequent participants in online-based platform
work, particularly in the Dominican Republic where they make up nearly two-thirds of workers
on Workana. This means that in some countries in the region, women seem to be benefiting as
platform work becomes more common.
The flexibility associated with new types of working arrangements could have benefits for women
who are balancing paid and unpaid work responsibilities but could also entrench restrictive
social norms related to women’s roles inside and outside of the household. The flexibility that
characterizes remote work and platform jobs may complement the working patterns of women in
CADR. For instance, motherhood increases women’s employment in more flexible jobs like parttime
employment and self-employment (Berniell et al. 2023). The rise of remote and platform work could
thus draw some women (back) into the workforce. Expansions in access to markets enabled by
platforms could also benefit self-employed women disproportionally, given their overrepresentation
in service jobs. At the same time, however, these new developments in working arrangements
may reinforce normative expectations about women’s role inside and outside of the household.
Intergenerational norms tend to pressure women in CADR, especially married women, to avoid
working outside of the house (Chioda 2016). Increased opportunities for home-based work could
reinforce these norms by making home-based work more attractive. Finally, location-based platform
work exhibits the same kind of gender segregation as work outside the gig economy: women tend
to concentrate in sectors traditionally viewed as “female” like personal care, cleaning, and beauty
(Deshpande, Singh, and Murthy 2022).
Chapter 5 Policy Recommendations 82
BOX 5.5: The Complex Interaction of Technology, Labor Markets, and Gender
(continued)
Beyond the changes in tasks and working arrangements that are the main focus of this
report, technologies including but also beyond computers, AI, and mobile robotics could shift
incentives and preferences for women to participate in the labor market. “Home technologies”
like domestic appliances (for example, washers, dryers, and freezers) and access to infrastructure
(for example, electricity and water) and “health technologies” like contraception may reduce time
spent on (unpaid) domestic work, changing household’s incentives and preferences and potentially
leading more women to enter the paid labor market. For instance, Almeida and Viollaz (2022) find
a positive relationship between household ownership of a washing machine and female labor
force participation in Guatemala. Cubas (2016) finds that falling relative prices for appliances and
increased access to electricity were associated with increased female labor force participation
in Brazil and Mexico.64 Gasparini and Marchionni (2015) find a positive relationship between
contraception use and female labor force participation in LAC.65 “Home-work technologies” like
access to the Internet may increase opportunities to engage in (more flexible) platform work.66
This is particularly important because women face obstacles to accessing technology in CADR.
Data from Meta’s Survey on Gender Equality at Home show that women in all countries in the
region are less likely than men to have access to smartphones and computers.
Mitigating the disruptions and taking advantage of the opportunities of technological
progress will require special focus on adapting responses to the particular obstacles—and
opportunities—faced by women. This means addressing the preexisting labor market challenges
faced by women, many of which create barriers to deploying their human capital in the labor
market, and thus, to taking advantage of the opportunities created by technological progress.
Policies to encourage women’s participation in STEM fields, to expand access to care, and to
improve regulatory policies and enforcement related to nondiscrimination in the workplace stand
out as priorities for governments in the region to improve the development and deployment of
women’s human capital as technological progress advances.
Design Digitally Enabled, Fit-for-purpose Intermediation Programs
Beyond efforts to develop human capital, deploying skills in the labor market is an additional challenge.
Workers may lack information about jobs and wages. Employers may be uncertain about the skills of
job candidates. Workers may be in locations where job creation is weak. Labor market intermediation
tools can help overcome these barriers. These include the provision of labor market information, career
guidance and job applications workshops that improve workers’ expectations about the labor market,
skills certificate programs that assist with skills signaling, and subsidies that help jobseekers search for or
travel to jobs in other locations.
Intermediation programs are more successful when they focus on overcoming barriers that prevent
jobseekers from matching with jobs where there is labor market demand. Existing intermediation efforts
may struggle to have substantial impacts if they do not do enough to help workers find jobs in the long
run or if they replicate services already provided (more effectively) by the private sector. In their review
of the effectiveness of labor market intermediation services, Carranza and McKenzie (2023) highlight
the potential for transport subsidies and skill signaling interventions to help connect jobseekers with
64 See also Dinkelman (2011) and Koolwal and van de Walle (2013). See Cavalcanti and Tavares (2008), Greenwood, Seshadri, and Yorukoglu
(2005), and Coen-Pirani, León, and Lugauer (2010) for research on developed countries.
65 See also, for example, Bailey (2006) and Albanesi and Olivetti (2016) for the United States.
66 See, for example, Dettling (2017).
Chapter 5 Policy Recommendations 83
employment. The interventions seem to be more successful because they help jobseekers overcome
geographic disparities in jobs in the first case, and information problems in the second. Because many
jobseekers in CADR countries search for jobs via personal networks, young, informal, and less-skilled
workers may be more affected by information problems as they may not have access to networks with
accurate information on good jobs. An obvious, though crucial, success factor for these programs is labor
market demand: without jobs, interventions to link workers to employment cannot succeed. This again
emphasizes the importance of using labor market information to identify areas of strong demand.
Technology is enabling improvements in the delivery of intermediation services. Technology is being
used to increase jobseekers’ access to labor market information that is relevant to their job search. In
Peru, text messages have been used to inform jobseekers about opportunities matching their profiles.
These messages had a positive impact on employment (Dammert, Galdo, and Galdo 2015). In the United
Kingdom, an online tool that provided jobseekers with relevant occupations and jobs expanded the
options they considered and increased their job interviews (Belot, Kircher, and Muller 2019). New tools
have also been developed that assist with the identification and classification of jobseeker skills and the
presentation of these skills to employers (box 5.6). These tools use simple questions to elicit information
about jobseekers’ experiences beyond employment that may be useful in the labor market, which can
be particularly relevant for young people or women who may lack traditional labor market experience.
Finally, governments are experimenting with incorporating platform jobs onto traditional online job search
portals, in part to help disadvantaged workers build experience and work histories that are tracked and
verified in the platform (box 5.7). A potential area to explore is how CADR migrants returning from abroad
could be engaged in platform work. Less-skilled migrants returning from the United States may have the
English skills that can be a barrier to work on platforms for less-skilled workers. Call centers have become
more common in the region, taking advantage of migrants returning or deported from the United States
with good English skills.
BOX 5.6: Using SkillCraft to Connect Disadvantaged Young People to the Labor
Market
SkillCraft produces assessments of twenty-first century skills to help connect disadvantaged
young people with the labor market. SkillCraft is a free online skills assessment and career
guidance tool developed by South Africa’s Youth Employment Service (YES) and Harambee Youth
Employment Accelerator in partnership with the World Bank. The application uses standardized
assessments delivered in the form of interactive games and quizzes to assess 23 twenty-first
century skills and traits. The application is accessible via computer, tablet, or mobile device. Based
on user results, the program produces two outputs: a printable, formal certificate that can be shared
with potential employers; and an interactive feedback report meant to be used by the jobseekers
to help them identify and select jobs or entrepreneurship opportunities that align with their skills
and traits. A recent evaluation of the program conducted by the World Bank and YES found that
the tool produced reliable assessments of twenty-first century skills among disadvantaged youth,
providing an equitable and unbiased assessment of skills for participants across education levels
and genders. In addition, the evaluation found that the assessment provided useful feedback for
jobseekers, leading to improvements in job search behaviors. A comparison between a group
who only received the skills certificate and those who received both the interactive feedback tool
and the skills certificate found that the interactive feedback tool increased the likelihood of the
jobseekers sharing the skills certificate with potential employers, perhaps because it helped them
understand the relevance of twenty-first century skills to jobs they were seeking.
Source: World Bank 2022d.
Chapter 5 Policy Recommendations 84
BOX 5.7: Public Gig Work Platforms
Some governments are undertaking initiatives to connect workers with platform jobs . In
Malaysia, for example, the eRezeki platform helps connect potential workers with platform jobs
in the private sector. Workers can visit in-person centers to receive support in setting up profiles
and selecting platforms and jobs, whether online- or location-based, that match their skills.
In the United StateIthe City of Long Beach, California has taken a more advanced approach,
creating the WorkLB program to connect workers with flexible job opportunities, including in the
provision of city services. Launched in 2018 and expanded during the COVID-19 pandemic, the
program originally offered opportunities such as backup school support staff, event staff, and
community health teams. Since then, the program has expanded to include both public and private
employment opportunities ranging from childcare to food service. The platform allows workers
to acquire badges that certify their capabilities and to search based on specific available working
hours and job locations. The program also has an emphasis on worker protections, ensuring that
all positions posted are with formal employers (Briggs and Rowan 2023).
Intermediation programs can also help create safe and inclusive migration pathways. Migrants
from CADR tend to work in low-skilled occupations, primarily in the United States. Informal migration
is common. Though thus far fairly limited in their application, Global Skills Partnerships (GSPs) have
significant potential as a model to facilitate safe, regular migration that is mutually beneficial for origin
and destination countries, including for more skilled occupations ranging from tourism to care. Under
GSPs, destination countries provide resources to train people in the origin country in skills that are in
demand in both. Trainees then decide whether to migrate—via a legal pathway—or to remain in the home
country. GSPs require the involvement of public employment and migration agencies as well as the private
sector and depend on the labor market insights and demand-driven training described in the previous
two sections. One such GSP, the Australia Pacific Training Coalition, provides training in 14 Pacific Island
countries and has enrolled more than 18,000 students since 2007 (Chand, Clemens, and Dempster 2021).
Digital job platforms can potentially facilitate matches between workers and employers utilizing GSPs
based on skills, experience, and preferences.
ADAPT SOCIAL PROTECTION AND LABOR MARKET POLICIES TO NEW FORMS OF WORK
The shifts in work that are occurring in CADR will create additional challenges for social protection and
labor market policies that already confront large informal workforces. Even moderate technological
progress will create disruptions in labor markets in CADR countries. Where the deployment of technology
and the development and deployment of human capital improve, economic growth is likely to ensue,
but so too will more significant labor market disruptions. At the same time, the rise of platform work
complicates labor market regulations. Platforms raise questions about how to ensure that platform
workers have equal access to benefits and workplace protections. On one hand, the rise of this new form
of employment creates an opportunity for formalization: earnings via platform work are observable and
so can be incorporated into social protection schemes, as has been done in Indonesia, Malaysia, and
Uruguay (ILO 2021b). On the other hand, there is a risk that a poorly coordinated regulatory response
leads to further labor market segmentation by creating new special classifications for platform workers.
Reform of labor market regulations in response to the emergence of platform work should seek to
avoid further labor market segmentation. Platform workers frequently lack access to social protection
and other workplace protections like unemployment, sickness, and disability insurance that are frequently
tied to traditional employment (Datta and Chen 2023; ILO 2021b). This is often because platform and
other “nonstandard” workers do not fit into definitions of employment prescribed in labor codes and other
regulations. Approaches to fill this gap include incorporating platform and other nonstandard workers
into existing regulatory frameworks or creating a new category of workers to extend existing protections
Chapter 5 Policy Recommendations 85
to these groups, or defining a minimum level of protections that apply to all workers, regardless of type
(Apella, Moroz, and Zunino 2023). While the best approach will vary from country to country, the risk of
generating additional labor market segmentation through additional regulations is substantial and could
further exacerbate existing challenges for CADR’s large informal labor markets.
Basing the design of labor regulations on an analysis of market failures is key . Simply expanding
existing regulations may not be appropriate because of existing challenges with enforcement and
differences in the market failures justifying regulations67 (Moroz and Santos 2022). Indeed, certain issues
that go beyond labor market regulations may be more salient in the case of platform work. For example,
monopsony is a significant concern in some cases: studies of workers on Amazon Mechanical Turk and
of Uber drivers find low residual labor supply elasticities, indicating strong employer buying power (Chen
et al. 2019; Dube et al. 2020; Kingsley, Gray, and Suri 2015). This is also true in the case of platforms that
facilitate digital provision of services. These platforms have characteristics—returns to scale, network
externalities, the accumulation and use of personal data—that make market concentration more likely
(Cirera, Comin, and Cruz 2022). This makes competition and antitrust regulations important. This can
begin with developing heuristics of the business models, characteristics, and anti-competitive potential
of different digital platforms and evolve into more sophisticated analyses of the particular anticompetitive
practices of platforms, especially issues related to data (Nyman and Barajas 2021). Efforts to increase
competition in general are also important because they can increase incentives for technology adoption
and output expansion, critical factors to compensate for the labor market disruptions of technological
progress (Dutz, Almeida, and Packard 2018; Vivarelli 2014).
In the long term, moving away from reliance on the traditional employer-employee relationship for the
financing and provision of protections and benefits is necessary. Technological progress will continue
creating disruptions in CADR’s labor markets. These disruptions will make policies protecting people’s
wellbeing even more important. At the same time, labor market disruptions will also make access to jobs
with social protection and other workplace benefits challenging, particularly in CADR countries where
coverage of social protection is low. This makes a general rethinking of social protection financing and
provision important. In general, reform should be biased towards creating uniform protections designed
to apply to all forms of work and to models that move away from reliance on the traditional employer-
employee relationship for the financing and provision of protections and benefits (Packard et al. 2019;
Beylis et al. 2020).
In the short term, the emergence of platform work creates an opportunity to expand access to social
protection. Platform work has a few special characteristics that make expanding access to its workers
easier than expanding access to other types of informal workers. These include the observability of
income, which is paid online, and the existence of a few large formal employers (rather than many small
employers) that governments can feasibly work with directly (Moroz and Santos 2022). Gig workers seem
to want to pay for greater protection. In Malaysia, a recent study found substantial unmet demand for social
insurance and a high willingness to pay for it (Ghorpade, Rahman, and Jasmin 2023). Governments can
partner with platforms to identify informal workers and incorporate them into social registries; introduce
novel design and behavioral interventions into existing social protection schemes to facilitate enrolment
and payment of contributions; and link with third-party providers of financial services like insurance and
savings accounts targeted to the needs of gig workers (Anand and Murthy 2023; Datta and Chen 2023;
Tapia 2023). For example, deductions for work accident and death benefits are automatically deducted
from the e-wallets of workers for ride-hailing apps GoJek and Grab in Indonesia. In Malaysia, Grab offers a
matching contribution to workers who register with the government’s retirement savings scheme.
67 For example, the externalities from worker dismissals may be lower in the case of platform work because platform work tends not to be a
worker’s primary employment and the costs of switching jobs is lower.
Chapter 5 Policy Recommendations 86
References
Abdullaev, Umidjon and Marcello Estevão. 2013. “Growth and Employment in the Dominican Republic:
Options for a Job-Rich Growth.” Working Paper 13/40, IMF, Washington, DC.
Abramovsky, Laura, and Rachel Griffith. 2006. “Outsourcing and Offshoring of Business Services: How
Important Is ICT?” Journal of the European Economic Association 42 (2–3):594–601.
Acemoglu, Daron, and David Autor. 2011. “Skills, Tasks and Technologies: Implications for Employment
and Earnings.” In Handbook of Labor Economics, Volume 4B, edited by Orley Ashenfelter and David
Card, 1043–171. Amsterdam: North Holland.
Acemoglu, Daron, and Pascual Restrepo. 2018. “The Race between Man and Machine: Implications
of Technology for Growth, Factor Shares, and Employment.” American Economic Review 108 (6):
1488–1542.
Acemoglu, Daron, and Pascual Restrepo. 2019. “Automation and New Tasks: How Technology Displaces
and Reinstates Labor.” Journal of Economic Perspectives 33 (2): 3–30.
Acemoglu, Daron, and Pascual Restrepo. 2020. “Robots and Jobs: Evidence from US Labor Markets.”
Journal of Political Economy 128 (6): 2188–2244.
Acemoglu, Daron, and Pascual Restrepo. 2022. “Tasks, Automation, and the Rise in U.S. Wage Inequality.”
Econometrica 90 (5): 1973–2016.
Acemoglu, Daron, Gary Anderson, David Beede, Catherine Buffington, Eric Childress, Emin Dinlersoz,
Lucia Foster, Nathan Goldschlag, John Haltiwanger, Zachary Kroff, Pascual Restrepo, and Nikolas
Zolas. 2022. “Automation and the Workforce: A Firm-Level View from the 2019 Annual Business
Survey.” Working Paper 30659, National Bureau of Economic Research, Cambridge.
Acevedo, Paloma, Guillermo Cruces, Paul Gertler, and Sebastian Martinez. 2020. “How Vocational
Education Made Women Better Off but Left Men Behind.” Labour Economics 65: 101824.
Adams-Prassl, Abi, Teodora Boneva, Marta Golin, and Christopher Rauh. 2022. “Work That Can Be Done
from Home: Evidence on Variation within and across Occupations and Industries.” Labour Economics
74: 102083.
ADB (Asian Development Bank). 2021. Global Value Chain Development Report 2021: Beyond Production.
Manila: ADB.
Agrawal, Ajay, Joshua S. Gans, and Avi Goldfarb. 2019. “Artificial Intelligence: The Ambiguous Labor
Market Impact of Automating Prediction.” Journal of Economic Perspectives 33 (2): 31–50.
Agrawal, Ajay, Joshua S. Gans, and Avi Goldfarb. 2023. “Artificial Intelligence Adoption and System-wide
Change?” Journal of Economics & Management Strategy (November): 1–11.
Agrawal, Ajay, John Horton, Nicola Lacetera, and Elizabeth Lyons. 2015. “Digitization and the Contract Labor
Market: A Research Agenda.” In Economic Analysis of the Digital Economy, edited by Avi Goldfarb,
Shane M. Greenstein, and Catherine E. Tucker, 219–250. Chicago: University of Chicago Press.
References 87
Agüero, Aileen, Monserrat Bustelo, and Mariana Viollaz. 2020. “¿Desigualdades en el mundo digital?
Brechas de género en el use de las TIC.” Technical Note 01879, IDB, Washington, DC.
Akerman, Ander, Ingvil Gaarder, and Magne Mogstad. 2015. “The Skill Complementarity of Broadband
Internet.” The Quarterly Journal of Economics 130 (4): 1781–824.
Albanesi, Stefania, and Claudia Olivetti. 2016. “Gender Roles and Medical Progress.” Journal of Political
Economy 124(3):650-95.
Anand, Malika, and Gayatri Murthy. 2023. “Financial Services for Gig Workers: Lessons on Getting Design
and Delivery Right.” Brief, Consultative Group to Assist the Poor, Washington, DC.
Antonczyk, Dirk, Thomas DeLeire, and Bernd Fitzenberger. 2018. “Polarization and Rising Wage Inequality:
Comparing the U.S. and Germany.” Econometrics 6 (2).
Albanesi, Stefania, and Claudia Olivetti. 2016. “Gender Roles and Medical Progress.” Journal of Political
Economy 124 (3): 650–95.
Albinowski, Maciej, and Piotr Lewandowski. 2022. “The Impact of ICT and Robots on Labour Market
Outcomes of Demographic Groups in Europe.” Discussion Paper 15752, Institute of Labor Economics,
Bonn.
Alipour, Jean-Victor, Oliver Falck, and Simone Schüller. 2020. “Germany’s Capacities to Work from Home.”
Discussion Paper Series 13152, Institute of Labor Economics, Bonn.
Almeida, Rita K., and Mariana Viollaz. 2022. “Women in Paid Employment: A Role for Public Policies and
Social Norms in Guatemala.” Policy Research Working Paper 9919, World Bank, Washington, DC.
Althoff, Lukas, Fabian Eckert, Sharat Ganapati, and Conor Walsh. 2022. “The Geography of Remote Work.”
Regional Science and Urban Economics 93: 103770.
Apella, Ignacio, and Gonzalo Zunino. 2022. “El cambio tecnológico y las tendencias del mercado laboral
en América Latina y el Caribe: un análisis basado en las tareas.” Revista de la CEPAL 136: 65–88.
Apella, Ignacio, Harry Moroz, and Gonzalo Zunino. 2023. Regulating Nonstandard Forms of Employment in
Uruguay: Insights from International Experience and Options for the Future. Washington, DC.: World Bank.
Aramburu, Julian, and Ana Goicoechea. 2021. Coding Bootcamps for Female Digital Employment:
Evidence from a Randomized Control Trial in Argentina and Colombia. Washington, DC.: World Bank.
Araújo, Bruno César, Francesco Bogliacino, and Marco Vivarelli. 2011. “Technology, Trade, and Skills in
Brazil: Evidence from Microdata.” CEPAL Review 105: 157–71.
Arntz, Melanie, Terry Gregory, and Ulrich Zierahn. 2016. “The Risk of Automation for Jobs in OECD
Countries.” OECD Social, Employment and Migration Working Papers 189, Organisation for Economic
Co-operation and Development, Paris.
Artuc, Erhan, Paulo Bastos, and Bob Rijkers. 2023. “Robots, Tasks, and Trade.” Journal of International
Economics 145: 103828.
Artuc, Erhan, Luc Christiaensen, and Hernand Winkler. 2019. “Does Automation in Rich Countries Hurt
Developing Ones: Evidence from the U.S. and Mexico.” Policy Research Working Paper 8741, World
Bank, Washington, DC.
References 88
Atalay, Enghin, Phai Phongthiengtham, Sebastian Sotelo, and Daniel Tannenbaum. 2018. “New
Technologies and the Labor Market.” Journal of Monetary Economics 97: 48–67.
Atalay, Enghin, Phai Phongthiengtham, Sebastian Sotelo, and Daniel Tannenbaum. 2020. “The Evolution
of Work in the United States.” America Economic Journal: Applied Economics 12 (2): 1–34.
Atkin, David, Antoinette Schoar, and Sumit Shinde. 2023. “Working from Home, Worker Sorting and
Development.” Working Paper 31515, National Bureau of Economic Research, Cambridge.
Autor, David. 2022. “The Labor Market Impacts of Technological Change: From Unbridled Enthusiasm
to Qualified Optimism to Vast Uncertainty.” Working Paper 30074, National Bureau of Economist
Research, Cambridge.
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 (5): 1553–97.
Autor, David H., Frank Levy, and Richard J. Murnane. 2003. “The Skill Content of Recent Technological
Change: An Empirical Exploration.” The Quarterly Journal of Economics 118 (4): 1279–333.
Autor, David H., Lawrence F. Katz, and Melissa S. Kearney. 2006. “The Polarization of the U.S. Labor
Market.” AEA Papers & Proceedings 96 (2): 189–94.
Autor, David H., Lawrence F. Katz, and Melissa S. Kearney. 2008. “Trends in U.S. Wage Inequality: Revising
the Revisionists.” The Review of Economics and Statistics 90 (2): 300–23.
Autor, David H., Caroline Chin, Anna M. Salomons, and Bryan Seegmiller. 2022. “New Frontiers: The
Origins and Content of New Work, 1940–2018.” Working Paper 30389, National Bureau of Economic
Research, Cambridge, MA.
Azuara, Oliver, Stephanie González, and Lukas Keller. 2019. “Who Drives on Ride-Hailing Platforms in Latin
America: A Profile of Uber Drivers in Brazil, Chile, Colombia and Mexico.” Technical Note 1779, Inter-
American Development Bank, Washington, DC.
Babcock, Linda, Maria P. Recalde, Lise Vesterlund, and Laurie Weingart. 2017. “Gender Differences in
Accepting and Receiving Requests for Tasks with Low Promotability.” American Economic Review
107(3):714-47.
Bachmann, Ronald, Merve Cim, and Colin Green. 2019. “Long-Run Patterns of Labour market Polarization:
Evidence from German Micro Data.” British Journal of Industrial Relations 57 (2): 350–76.
Bailey, Martha J. 2006. “More Power to the Pill: The Impact of Contraceptive Freedom on Women’s Life
Cycle Labor Supply.” The Quarterly Journal of Economics 121(1):289-320.
Bailey, Martha J., Shuqiao Sun, and Brenden Timpe. 2021. “Prep School for Poor Kids: The Long-Run
Impacts of Head Start on Human Capital and Economic Self-Sufficiency.” American Economic Review
111 (12): 3963–4001.
Baldwin, Richard. 2022. “Globotics and Macroeconomics: Globalisation and Automation of the Service
Sector.” Working Paper Series 30317, National Bureau of Economic Research, Cambridge.
Banegas, Nancy, and Hernan Winkler. 2020. Jobs Diagnostic: El Salvador. Understanding Challenges for
More and Better Jobs in El Salvador: An Integrated Approach. Jobs Series. Washington, DC.: World
Bank.
References 89
Banga, Karishma, and Dirk Willem te Velde. 2019. “Preparing Developing countries for the Future of
Work: Understanding Skills-Ecosystem in a Digital Era.” Background Paper, Pathways for Prosperity
Commission, Oxford.
Bárány, Zsófia, and Christian Siegel. 2018. “Job Polarization and Structural Change.” America Economic
Journal: Macroeconomics 10 (1): 57–89.
Barrera-Osorio, Felipe, Adriana Kugler, and Mikko Silliman. 2023. “Hard and Soft Skills in Vocational
Training: Experimental Evidence from Colombia.” The World Bank Economic Review 37 (3): 409–36.
Barrero, Jose Maria, Nicholoas Bloom, and Steven J. Davis. 2021. “Why Working from Home Will Stick.”
Working Paper 28731, National Bureau of Economic Research, Cambridge.
Bashir, Sajitha, T.H. Gindling, and Ana Maria Oviedo. 2012. Better Jobs in Central America: The Role of
Human Capital.” Washington, DC.: World Bank.
Basso, Gaetano, Giovanni Peri, and Ahmed S. Rahman. 2020. “Computerization and Immigration: Theory
and Evidence from the United States.” Canadian Journal of Economics 53 (4): 1457–94.
Battisti, Michele, Christian Dustmann, and Uta Schönberg. 2023. “Technological and Organizational
Change and the Careers of Workers.” Journal of the European Economic Association 21 (4):1551–94.
BCCR (Banco Central de Costa Rica). 2019. Costa Rica: Exportaciones de Servicios Mediante Redes de
Tecnología, Información y Comunicación (TIC). Banco Central de Costa Rica, San José.
Behrens, Kristian, Sergey Kichko, and Jacques-François Thisse. 2021. “Working From Home: Too Much of
a Good Thing?” Working Paper 8831, CESifo, Munich.
Belafi, Carmen, Yue-Yi Hwa, and Michelle Kaffenberger. 2020. “Building on Solid Foundations: Prioritising
Universal, Early, Conceptual and Procedural Mastery of Foundational Skills.” RISE Insight Series.
Oxford; Research on Improving Systems of Education Programme.
Belot, Michèle, Philipp Kircher, and Paul Muller. 2019. “Providing Advice to Jobseekers at Low Cost: An
Experimental Study on Online Advice.” The Review of Economic Studies 86 (4):1411–47.
Bendini, Magdalena, Victoria Levin, and Isil Oral-Savonitto. 2019. “Stocktaking of Adult Literacy Programs:
Design Features for Impact.” Presentation at the World Bank, Washington, DC., June 10, 2019.
Benhabib, Jess, and Mark M. Spiegel. 1994. “The Role of Human Capital in Economic Development:
Evidence from Aggregate Cross-Country Data.” Journal of Monetary Economics 34:143–73.
Berg, Andrew, Edward F. Buffie, and Luis-Felipe Zanna. 2018. “Should We Fear the Robot Revolution? (The
Correct Answer Is Yes).” Working Paper 18/116, International Monetary Fund, Washington, DC.
Berger, Thor, and Carl Benedikt Frey. 2016. “Did the Computer Revolution Shift the Fortunes of U.S. Cities?
Technology Shocks and the Geography of New Jobs.” Regional Science and Urban Economics 57:
38–45.
Berniell, Inés, Lucila Berniell, Dolores de la Mata, María Edo, and Mariana Marchionni. 2023. “Motherhood
and Flexible Jobs: Evidence from Latin American Countries.” World Development 167 (July 1, 2023).
Berniell, Lucila, and Daniel Fernandez. 2021. “Jobs’ Amenability Is Not Enough: The Role of Household
Inputs for Safe Work under Social Distancing in Latin American Cities.” World Development 140 (April).
References 90
Berniell, Inés, Leonardo Gasparini, Mariana Marchionni, and Mariana Viollaz. 2021. “The Role of Work-
from-Home in the Gender Asymmetries of COVID-19: An Analysis for Latin America Based on High-
Frequency Surveys.” Policy Research Working Paper 9887, Washington, DC., World Bank.
Bessen, James, Erich Denk, and Chen Meng. 2022. “The Remainder Effect: How Automation Complements
Labor Quality.” Research Paper Series 22–3, Boston University School of Law, Boston.
Beylis, Guillermo, Roberto Fattal Jaef, Michael Morris, Ashwini Rekha Sebastian, and Rishabh Sinha.
2020. Going Viral: COVID-19 and the Accelerated Transformation of Jobs in Latin America and the
Caribbean. Washington, DC.: World Bank.
Blinder, Alan S. 2006. “Offshoring: The Next Industrial Revolution?” Foreign Affairs (2): 113–128.
Blinder, Alan S. 2009. “How Many US Jobs Might Be Offshorable?” World Economics 10 (2): 41–78.
Blinder, Alan S., and Alan B. Krueger. 2013. “Alternative Measures of Offshorability: A Survey Approach.”
Journal of Labor Economics 31 (2): S97–128.
Bloom, Nicholas, Raffaella Sadun, and John Van Reenen. 2017. “Management as Technology?” Working
Paper 22327, National Bureau of Economic Research, Cambridge.
Blyde, Juan S. 2020. “Heterogenous Labor Impacts of Migration across Skills Groups: The Case of Costa
Rica.” Working Paper 1145, Inter-American Development Bank, Washington, DC.
Borin, Alessandro, Michele Mancini, and Daria Taglioni. 2021. “Measuring Exposure to Risk in Global Value
Chains.” Policy Research Working Paper 9785, World Bank, Washington, DC.
Bravo-Ureta, Boris E., Eric Njuki, Ana Claudia Palacios, and Lina Salazar. 2022. “Agricultural Productivity in
El Salvador: A Preliminary Analysis.” Working Paper Series 01319, Inter-American Development Bank,
Washington, DC.
Breemersch, Koen, Jože P. Damijan, and Jozef Konings. 2017. “Labour Market Polarization in Advanced
Countries: Impact of Global Value Chains, Technology, Import Competition from China and Labour
Market Institutions.” OECD Social, Employment and Migration Working Paper 197, Organisation for
Economic Co-operation and Development, Paris.
Bresnahan, Timothy. 2021. “Artificial Intelligence Technologies and Aggregate Growth Prospects.” In
Prospects for Economic Growth in the United States, edited by John W. Diamond and George R.
Zodrow, 132–70, Cambridge: Cambridge University Press.
Bresnahan, Timothy. 1999. “Computerisation and Wage Dispersion: An Analytical Reinterpretation.” The
Economic Journal 109: F390-415.
Briggs, Xavier and Wingham Rowan. 2023. “To Enhance Community Services and Empower Workers,
Local Governments Are Building Their Own Gig Work Platforms.” Brookings Commentary, May 9,
2023. https://www.brookings.edu/articles/to-enhance-community-services-and-empower-workers-
local-governments-are-building-their-own-gig-work-platforms/.
Brynjolfsson, Erik, Danielle Li, and Lindsey Raymond. 2023. “Generative AI at Work.” Policy Research
Working Paper 31161, National Bureau of Economic Research, Cambridge.
References 91
Brynjolfsson, Erik, Thomas W. Malone, Vijay Gurbaxani, and Ajit Kambil. 1989. “Does Information
Technology Lead to Smaller Firms?” Center for Information Systems Research Working Paper 211,
Massachusetts Institute of Technology, Cambridge.
Brynjolfsson, Erik, and Tom Mitchell. 2017. “What Can Machine Learning Do? Workforce Implications.”
Science 358 (6370): 1530–4.
Brynjolfsson, Erik, Tom Mitchell, and Daniel Rock. 2018. “What Can Machines Learn and What Does It Mean
for Occupations and the Economy.” AEA Papers and Proceedings 108: 43–7.
Brynjolfsson, Erik, Tom Mitchell, and Daniel Rock. 2023. “Quantifying the Distribution of Machine
Learning’s Impact on Work.” Unpublished Working Paper.
Busso, Matías, and Diana Hincapié. 2017. “The Market for Skills: Beyond Supply and Demand.” In Learning
Better: Public Policy for Skills Development, edited by Matías Busso, Julián Cristia, Diana Hincapié,
Julián Messina, and Laura Ripani, 69–98. Washington, DC.: Inter-American Development Bank.
Bustelo, Monserrat, Luca Flabbi, and Mariana Viollaz. 2020. “The Gender Labor Market Gap in the Digital
Economy.” Working Paper Series 01056, Washington, DC., IDB.
Calì, Massimiliano and Giorgio Presidente. 2021. “Automation and Manufacturing Performance in a
Developing Country.” Policy Research Working Paper 9653, World Bank, Washington, DC.
Carbonero, Francesco, Ekkehard Ernst, and Enxo Weber. 2020. “Robots Worldwide: The Impact of
Automation on Employment and Trade.” IAB Discussion Paper 7/20, German Federal Employment
Agency, Nuremberg.
Carranza, Eliana, and David McKenzie. 2023. “Job Training and Job Search Assistance Policies in
Developing Countries.” Policy Research Working Paper 10576, World Bank, Washington, DC.
Caselli, Francesco, and Wilbur John Colemann II. 2006. “The World Technology Frontier.” The American
Economic Review 96 (3): 499–522.
Castro-Leal, Florencia T., and Janet Porras-Mendoza. 2020. Jobs Diagnostic: Nicaragua. Toward a Jobs
and Economic Transformation Strategy. Jobs Series. Washington, DC.: World Bank.
Caunedo, Julieta, Elisa Keller, and Yongseok Shin. 2023. “Technology and the Task Content of Jobs across
the Development Spectrum.” The World Bank Economic Review 37 (3): 479–93.
Cavalcanti, Tiago, and José Tavares. 2008. “Assessing the ‘Engines of Liberation’: Home Appliances and
Female Labor Force Participation.” The Review of Economics and Statistics 90 (1): 81–8.
Chan, Curtis K., and Michel Anteby. 2016. “Task Segregation as a Mechanism for Within-Job Inequality: Women
and Men of the Transportation Security Administration.” Administrative Science Quarterly 61(2):184-216.
Chand, Satish, Michael A. Clemens, and Helen Dempster. 2021. “A Pacific Skills Visa: Improving
Opportunities for Skilled Migration throughout the Pacific Region.” Policy Paper 231, Center for Global
Development, Washington, DC.
Charlton, Diane. 2023. “Closing the Agricultural Gap: Decent Migrant Work or Automation?” Policy Note,
Rome Dialogue II, Rome.
References 92
Chen, M. Keith, Judith A. Chevalier, Peter E. Rossi, and Emily Oehlsen. 2019. “The Value of Flexible Work:
Evidence from Uber Drivers.” Journal of Political Economy 127 (6): 2735–94.
Chioda, Laura. 2016. Work and Family: Latin American and Caribbean Women in Search of a New
Balance. Washington, DC.: World Bank.
Chiplunkar, Gaurav, and Pinelopi Koujianou Goldberg. 2022. “The Employment Effects of Mobile Internet
in Developing Countries.” Working Paper 30741, National Bureau of Economic Research, Cambridge.
Choudhury, Prithwiraj, Cirrus Foroughi, and Barbara Larson. 2020. “Work-from-Anywhere: The Productivity
Effects of Geographic Flexibility.” Strategic Management Journal 42: 655–83.
Cirera, Xavier, Diego Comin, and Marcio Cruz. 2022. Bridging the Technological Divide: Technology
Adoption by Firms in Developing Countries. Washington, DC.: World Bank.
Cirera, Xavier, Barbara Cunha, and Woori Lee. 2022. “Post Covid-19. Building a Resilient and More
Sustainable Recovery in El Salvador, Guatemala, and Honduras.” World Bank, Washington, DC.
Cirera, Xavier, and William F. Maloney. 2017. The Innovation Paradox: Developing-Country Capabilities
and the Unrealized Promise of Technological Catch-Up. Washington, DC.: World Bank.
Clemens, Michael A., Ethan G. Lewis, and Hannah M. Postel. 2018. “Immigration Restrictions as Active
Labor Market Policy: Evidence from the Mexican Bracero Exclusion.” American Economic Review 108
(6): 1468–87.
Coase, Ronald. 1937. “The Nature of the Firm.” Economica 4 (16): 386–405.
Coen-Pirani, Daniele, Alexis León, and Steven Lugauer. 2010. “The Effect of Household Appliances on
Female Labor Force Participation: Evidence from Microdata.” Labour Economics 17: 503–13.
Comin, Diego, and Bart Hobijn. 2004. “Cross-country Technology Adoption: Making the Theories Face the
Facts.” Journal of Monetary Economics 51: 39–83.
Comin, Diego, and Bart Hobijn. 2010. “An Exploration of Technology Diffusion.” American Economic
Review 100: 2031–59.
Comin, Diego, Bart Hobijn, and Emilie Rovito. 2008. “Technology Usage Lags.” Journal of Economic
Growth 13: 237–56.
Comin, Diego, and Martí Mestieri. 2018. “If Technology Has Arrived Everywhere, Why Has Income
Diverged?” American Economic Journal: Macroeconomics 10 (3): 137–78.
Conte, Andrea, and Marco Vivarelli. 2011. “Imported Skill-Biased Technological Change in Developing
Countries.” The Developing Economies 49 (1): 36–65.
Cortes, Guido Matias. 2016. “Where Have the Middle-Wage Workers Gone? A Study of Polarization Using
Panel Data.” Journal of Labour Economics 34 (1): 63–105.
Cortés, Patricia, Ying Feng, Nicolás Guida-Johnson, and Jessica Pan. 2023. “Automation and Gender:
Implications for Occupational Segregation and the Gender Skill Gap.” Working Paper 32030, National
Bureau of Economic Research, Cambridge.
CSF (Centre for Strategic Futures). 2015. Foresight. Singapore: CSF.
References 93
Cubas, German. 2016. “Distortions, Infrastructure, and Female Labor Supply in Developing Countries.”
European Economic Review 87: 194–215.
Cunningham, Wendy, Harry Moroz, Noël Muller, and Aivin Solatorio. 2022. “The Demand for Digital
and Complementary Skills in Southeast Asia.” Policy Research Working Paper 10070, World Bank,
Washington, DC.
Cusolito, Ana Paula, Daniel Lederman, and Jorge Peña. 2020. “The Effects of Digital-Technology Adoption
on Productivity and Factor Demand Firm-level Evidence from Developing Countries.” Policy Research
Working Paper 9333, World Bank, Washington, DC.
Dalton, Michael, and Jeffrey A. Groen. 2022. “Telework during the COVID-19 Pandemic: Estimates using the
2021 Business Response Survey.” Monthly Labor Review, Bureau of Labor Statistics, Washington, DC.
Dalvit, Nicolò, Rafael de Hoyos, Leonardo Iacovone, Ioanna Pantelaiou, Aleksandra Peeva, and Iván Torre.
2023. The Future of Work: Implications for Equity and Growth in Europe. Washington, DC.: World Bank.
Dammert, Ana C., Jose Galdo, and Virgilio Galdo. 2015. “Integrating Mobile Phone Technologies into
Labor-Market Intermediation: A Multi-Treatment Experimental Design.” IZA Journal of Labor &
Development 4 (11).
Dao, Mai Chi, Mitali Das, and Zsoka Koczan. 2019. “Why Is Labour Receiving a Smaller Share of Global
Income?” Economic Policy 34 (100): 723–59.
Das, Mitali, and Benjamin Hilgenstock. 2022. “The Exposure to Routinization: Labor Market Implications
for Developed and Developing Countries.” Structural Change and Economic Dynamics 60: 99–113.
Datta, Namita, and Rong Chen. 2023. Working without Borders: The Promise and Peril of Online Gig Work.
Washington, DC.: World Bank.
De Backer, Koen, Timothy DeStefano, Carlo Menon, and Jung Ran Suh. 2018. “Industrial Robotics and the
Global Organisation of Production.” Science, Technology and Industry Working Papers, Organisation
for Economic Co-operation and Development, Paris.
de Hoyos, Rafael E., Orazio Pietro Attanasio, and Costas Meghir. 2019. “Can Scholarships Increase High
School Graduation Rates ? Evidence from A Randomized Control Trial in Mexico.” Policy Research
Working Paper 8826, World Bank, Washington, DC.
de Vries, Gaaitzen J., Elisabetta Gentile, Sébastien Miroudot, and Konstantin M. Wacker. 2020. “The Rise
of Robots and the Fall of Routine Jobs.” Labour Economics 66.
Del Carmen, Giselle, and Liliana D. Sousa. 2018. “Human Capital Outflows: Selection into Migration from
the Northern Triangle.” Policy Research Working Paper 8334, World Bank, Washington, DC.
Delaporte, Isaure, and Werner Peña. 2020. “Working from Home Under COVID-19: Who Is Affected?
Evidence From Latin American and Caribbean Countries.” Discussion Paper 528. Global Labor
Organization, Essen.
Deloitte. 2014. Agiletown: the Relentless March of Technology and London’s Response. London: Deloitte.
Deming, David J. 2017. “The Growing Importance of Social Skills in the Labor Market.” The Quarterly
Journal of Economics 132 (4): 1593–1640.
References 94
Deming, David J., and Lisa B. Kahn. 2018. “Skill Requirements across Firms and Labor Markets: Evidence
from Job Postings for Professionals.” Journal of Labor Economics 31 (S1): S337–69.
Deming, David J., and Kadeem L. Noray. 2018. “STEM Careers and Technological Change.” Working Paper
25065, National Bureau of Economic Research, Cambridge.
Deshpande, Rani, Nisha Singh and Gayatri Murthy. 2022. “Women in the Platform Economy: Emerging
insights.” CGAP, Washington, DC.
Dettling, Lisa. 2017. “Broadband in the Labor Market: The Impact of Residential High-Speed Internet on
Married Women’s Labor Force Participation.” ILR Review 70 (2): 451–82.
Devercelli, Amanda E., and Frances Beaton-Day. 2020. Better Jobs and Brighter Futures: Investing in
Childcare to Build Human Capital. Washington, DC.: World Bank.
Díaz, Mercedes Mateo, and Lourdes Rodriguez-Chamussy. 2016. Cashing in on Education: Women,
Childcare, and Prosperity in Latin America and the Caribbean. Washington, DC.: World Bank Group
and Inter-American Development Bank.
Dicarlo, Emanuele, Salvatore Lo Bello, Sebastian Monroy-Taborda, Ana Maria Oviedo, Maria Laura
Sanchez-Puerta, and Indhira Santos. 2016. “The Skill Content of Occupations across Low and Middle
Income Countries: Evidence from Harmonized Data.” Discussion Paper 10224, Institute of Labor
Economics, Bonn.
Dingel, Jonathan I., and Brent Nieman. 2020. “How Many Jobs Can Be Done at Home?” Journal of Public
Economics 189.
Dinkelman, Taryn. 2011. “The Effects of Rural Electrification on Employment: New Evidence from South
Africa.” American Economic Review 101:3078-108.
Doan, Dung, Trang Luu, Nga Thi Nguyen, and Abla Safir. 2023. “Green Jobs: Upskilling and Reskilling
Vietnam’s Workforce for a Greener Economy.” Policy Note, World Bank, Washinton, DC.
Dube, Arindrajit, Jeff Jacobs, Suresh Naidu, and Siddharth Suri. 2020. “Monopsony in Online Labor
Markets.” American Economic Review: Insights 2 (1): 33–46.
Dustmann, Christian, Johannes Ludstock, and Uta Schönberg. 2009. “Revisiting the German Wage
Structure.” The Quarterly Journal of Economics 124 (2): 843–881.
Dutz, Mark A., Rita K. Almeida, and Truman G. Packard. 2018. The Jobs of Tomorrow: Technology,
Productivity, and Prosperity in Latin America and the Caribbean. Washington, DC.: World Bank.
Eberhard-Ruíz, Andreas. 2021. Jobs Diagnostic: Guatemala. Jobs Series. Washington, DC. World Bank.
ECLAC (Economic Commission for Latin America and the Caribbean). 2021. “The Economic Autonomy of
Women in a Sustainable Recovery with Equality.” Special Report, Santiago.
ECLAC (Economic Commission for Latin America and the Caribbean) and ILO (International Labor
Organization). 2021. Employment Situation in Latin America and the Caribbean: Decent Work for
Platform Workers in Latin America. ECLAC, Santiago.
Eden, Maya, and Paul Gaggl. 2020. “Do Poor Countries Really Need More IT?” The World Bank Economic
Review 34 (1): 48–62.
References 95
Egana-del Sol, Pablo, Monserrat Bustelo, Laura Ripani, Nicolas Soler, and Marina Viollaz. 2022. “Automation
in Latin America: Are Women at Higher Risk of Losing Their Jobs?” Technological Forecasting & Social
Change 175.
Eloundou, Tyna, Sam Manning, Pamela Mishkin, and Daniel Rock. 2023. “GPTs Are GPTs: An Early Look at
the Labor Market Impact Potential of Large Language Models.” Working Paper, arXiv.org.
Elverdin, Pablo, Valeria Piñeiro, and Miguel Robles. 2018. “La mecanización agrícola en América Latina.”
Discussion Paper 017410, International Food Policy Research Institute, Washington, DC.
Escudero, Verónica, Jochen Kluve, Elva López Mourelo, and Clemente Pignatti. 2019. “Active Labour
Market Programmes in Latin America and the Caribbean: Evidence from a Meta-Analysis.” The Journal
of Development Studies 55 (12): 2644–661.
Ezell, Stephen. 2024. “Assessing the Dominican Republic’s Readiness to Play a Greater Role in Global
Semiconductor and PCB Value Chains.” Information Technology & Innovation Foundation, Washington, DC.
Faber, Marius. 2020. “Robots and Reshoring: Evidence from Mexican Labor Markets.” Journal of
International Economics 127.
FAO (Food and Agricultural Organization of the United Nations). 2022. The State of Food and Agriculture:
Leveraging Automation for Transforming Agrifood Systems. Rome: Food and Agriculture Organization.
Feenstra, Robert C. 2017. “Statistics to Measure Offshoring and its Impact.” Working Paper 23067, National
Bureau of Economic Research, Cambridge.
Feigenbaum, James, and Daniel P. Gross. 2022. “Answering the Call of Automation: How the Labor Market
Adjusted to the Mechanization of Telephone Operation.” Working Paper 28061, National Bureau of
Economic Research, Cambridge.
Feigenbaum, James, and Daniel P. Gross. 2023. “Organizational and Economic Obstacles to Automation:
A Cautionary Tale from AT&T in the Twentieth Century.” Working Paper 29580, National Bureau of
Economic Research, Cambridge.
Feijao, Carolina, Isabel Flanagan, Christian van Stolk, and Salil Gunashekar. 2021. The Global Digital Skills
Gaps: Current Trends and Future Directions. Santa Monica: RAND Corporation.
Felipe, Jesus, Aashish Mehta, and Changyong Rhee. 2019. “Manufacturing Matters…But It’s the Jobs That
Count.” Cambridge Journal of Economics 43:139–68.
Felten, Edward W., Manav Raj, and Robert Seamans. 2018. “A Method to Link Advances in Artificial
Intelligence to Occupational Abilities.” AEA Papers & Proceedings 108: 54–57.
Felten, Edward W., Manav Raj, and Robert Seamans. 2021. “Occupational, Industry, and Geographic
Exposure to Artificial Intelligence: A Novel Dataset and Its Potential Uses.” Strategic Management
Journal 42: 2195–217.
Felten, Edward W., Manav Raj, and Robert Seamans. 2023. “How Will Language Modelers Like ChatGPT
Affect Occupations and Industries.” Research Paper, ResearchGate, Berlin.
Fernandes, Ana, Gaston Nievas, and Deborah Winkler. 2022. “Trade and GVC Integration in Latin America
and the Caribbbean.” In Deep Trade Agreements: Anchoring Global Value Chains in Latin America
and the Caribbean, edited by Nadia Rocha and Michele Ruta, 15–60. Washington, DC.: World Bank.
References 96
Fernandez, Cristina, and Juan Benavides. 2020. “Las plataformas digitales, la productividad y el empleo
en Colombia.” Article, Centro de Investigación Económica y Social, Bogotá.
Ferreyra, María Marta, Lelys Dinarte Díaz, Sergio Urzúa, and Marina Bassi. 2021. The Fast Track to New
Skills: Short-Cycle Higher Education Programs in Latin America and the Caribbean. Washington, DC.:
World Bank.
Frank, Morgan R., Lijun Sun, Manuel Cebrian, Hyejin Youn, and Iyad Rahwan. 2018. “Small Cities Face
Greater Impact from Automation.” Journal of the Royal Society Interface 15.
Frey, Carl Benedikt, and Michael A. Osborne. 2017. “The Future of Employment: How Susceptible Are Jobs
to Computerisation?” Technological Forecasting & Social Change 114: 254–80.
Fuglie, Keith, Madhur Gautam, Aparajita Goyal, and William F. Maloney. 2020. Harvesting Prosperity:
Technology and Productivity in Agriculture. Washington, DC.: World Bank.
Gaggl, Paul, and Greg C. Wright. 2017. “A Short-Run View of What Computers Can Do: Evidence from a UK
Tax Incentive.” American Economic Journal: Applied Economics 9 (3): 262–94.
Gagnon, Jason. 2011. “‘Stay with Us?’ The Impact of Emigration on Wages in Honduras.” Working Paper
300, Organisation for Economic Co-operation and Development, Paris.
Garcia, José Alexander, and Katherine Javier. 2020. “Los trabajadores de plataformas digitales en la
República Dominicana.” Documentos de Proyectos, Comisión Económica para América Latina,
Santiago.
Gasparini, Leonardo, Carlo Lombardo, Irene Brambilla, Andrés César, and Guillermo Falcone. 2021.
“Routinization and Employment: Evidence for Latin America.” Working Paper 276, Centro de Estudios
Distributivos, Laborales y Sociales, La Plata.
Gasparini, Leonardo, and Mariana Marchionni, eds. 2015. Bridging Gender Gas? The Rise and Decelaration
of Female Labor Force Participation in Latin America. Buenos Aires: Centro de Estudios Distributivos,
Laborales y Sociales.
Ghorpade, Yashodhan, Amanina Abdur Rahman, and Alyssa Jasmin. 2023. “Social Insurance for Gig
Workers: Insights from a Discrete Choice Experiment in Malaysia.” Policy Research Working Paper
10629, World Bank, Washington, DC.
Gibbs, Michael, Friederike Mengel, and Christoph Siemroth. 2023. “Work from Home and Productivity:
Evidence from Personnel and Analytics Data on Information Technology Professionals.” Journal of
Political Economy Microeconomics 1 (1): 7–41.
Giesing, Yvonne, and Britta Rude. 2022. “Robots, AI, and Immigration – A Race for Talent or of Displaced
Workers.” EconPol Forum 23 (5): 20–3.
Giuntella, Osea, Yi Lu, and Tianyi Wang. 2022. “How do Workers and Households Adjust to Robots?
Evidence from China.” Working Paper 30707, National Bureau of Economic Research, Cambridge.
Giuntella, Osea and Tianyi Wang. 2019. “Is an Army of Robots Marching on Chinese Jobs?” Discussion
Paper 12281, Institute of Labor Economics, Bonn.
Goldin, Claudia, and Lawrence F. Katz. 2002. “The Power of the Pill: Oral Contraceptives and Women’s
Career and Marriage Decisions.” Journal of Political Economy 110 (4): 730–70.
References 97
Goos, Maarten, and Alan Manning. 2007. “Explaining Job Polarization: Routine-Biased Technological
Change and Offshoring.” The American Economic Review. 104 (8): 2509–26.
Goos, Maarten, Alan Manning, and Anna Salomons. 2009. “Job Polarization in Europe.” American
Economic Review: Papers & Proceedings 99 (2): 58–63.
Goos, Maarten, Alan Manning, and Anna Salomons. 2014. “Explaining Job Polarization: Routine-Biased
Technological Change and Offshoring.” American Economic Review 104 (8): 2509–26.
Gottlieb, Charles, Jan Grobovšek, and Markus Poschke. 2020. “Working from Home across Countries.”
Covid Economics 8 (22): 71–91.
Gottlieb, Charles, Jan Grobovšek, Markus Poschke, and Fernando Saltiel. 2021. “Working from Home in
Developing Countries.” European Economic Review 133: 103679.
Grace, Katja, John Salvatier, Allan Dafoe, Baobao Zhang, and Owain Evans. 2018. “Viewpoint: When Will AI Exceed
Human Performance? Evidence from AI Experts.” Journal of Artificial Intelligence Research 62: 729–54.
Graetz, George, and Guy Michaels. 2018. “Robots at Work.” Review of Economics and Statistics 100 (5):
753–68.
Granata, Julia, and Josefina Posadas. 2022. Which Jobs Are Green? A Methodological Note on How to
Measure Green Jobs for Skills Policy and an Application to Indonesia. Washington, DC: World Bank.
Green, David A., and Benjamin M. Sand. 2015. “Has the Canadian Labour Market Polarized?” Canadian
Journal of Economics 48 (2): 612–46.
Greenwood, Jeremy, Ananth Seshadri, and Mehmet Yorukoglu. 2005. “Engines of Liberation.” The Review
of Economic Studies 72 (1): 109–33.
Gregory, Terry, Anna Salomons, and Zierahn. 2022. “Racing with or against the Machine? Evidence on the
Role of Trade in Europe.” Journal of the European Economic Association 20 (2): 869–906.
Gries, Thomas, and Wim Naudé. 2022. “Modelling Artificial Intelligence in Economics.” Journal of Labour
Market Research 56.
Guataquí, Juan Carlos Roa, and Diana Isabel Londoño. 2021. República Dominicana: Análisis de oferta
y demanda de empleo y trazabilidad de egresados Infotep, a partir de bases de datos. Geneva:
International Labour Organization.
Guven, Cahi, Panagiotis Sotirakopoulos, and Aydogan Ulker. 2020. “Short-term Labour Market Effects
of COVID-19 and the Associated National Lockdown in Australia: Evidence from Longitudinal Labour
Force Survey.” Working Paper 635, Global Labor Organization, Essen.
Hallward-Driemeier, Mary, and Gaurav Nayyar. 2018. Trouble in the Making? The Future of Manufacturing-
Led Development. Washington, DC. World Bank.
Hallward-Driemeier, Mary, and Gaurav Nayyar. 2019. “Have Robots Grounded the Flying Geese? Evidence
from Greenfield FDI in Manufacturing.” Policy Research Working Paper 9097, World Bank, Washington, DC.
Hanushek, Eric A., Guido Schwerdt, Ludger Woessmann, and Lei Zhang. 2017. “General Education,
Vocational Education, and Labor-Market Outcomes over the Lifecycle.” Journal of Human Resources
52 (1): 48–87.
References 98
Hardy, Wojciech, Roma Keister, and Piotr Lewandowski. 2018. “Education Upgrading, Structural Change
and the Task Composition of Jobs in Europe.” Economics of Transition 26 (2): 201–31.
Hatayama, Maho, Mariana Viollaz, and Hernan Winkler. 2023. “Jobs’ Amenability to Working from Home:
Evidence from Skills Surveys for 53 Countries.” Economía LACEA Journal 22 (1): 1–18.
Hausmann, Ricardo, Luis Espinoza, and Miguel Angel Santos. 2017. “Shifting Gears: A Growth Diagnostics
of Panama.” Working Paper 325, Center for International Development, Harvard University, Cambridge.
Hiller, Tatiana, and Marisol Rodríguez Chatruc. 2023. “South-South Migration and Female Labor Supply in
the Dominican Republic.” Journal of Economic Geography 23 (2): 419–48.
Hirvonen, Johannes, Aapo Stenhammer, and Joonas Tuhkuri. 2022. “New Evidence on the Effect of
Technology on Employment and Skill Demand.” Research Paper, ResearchGate, Berlin.
Hitt, Lorin M. 1999. “Information Technology and Firm Boundaries: Evidence from Panel Data.” Information
Systems Research 10 (2): 134–49.
Hjort, Jonas, and Jonas Poulsen. 2019. “The Arrival of Fast Internet and Employment in Africa.” American
Economic Review 109 (3): 1032–79.
Hötte, Kerstin, Melline Somers, and Angelos Theodorakopoulos. 2023. “Technology and Jobs: A
Systematic Literature Review.” Technological Forecasting & Social Change 194.
Hummels, David, Jakob R. Munch, and Chong Xiang. 2018. “Offshoring and Labor Markets.” Journal of
Economic Literature 56 (3): 981–1028.
IDB (Inter-American Development Bank). 2021b. The Future of Work in Latin America and the Caribbbean:
How Can Technology Facilitate Job Recovery after COVID-19. Washington, D.C.: IDB.IFC (International
Finance Corporation). 2019. Digital Skills in Sub-Saharan Africa: Spotlight on Ghana. Washington, DC:
IFC.
IFC (International Finance Corporation). 2023. Creating Markets in the Dominican Republic: Investment
Opportunities and Growth Pathways to Higher Value Addition, Resilience, and Inclusion. Washington,
DC.: IFC.
Ifft, Jennifer and Margaret Jodlowski. 2022. “Is ICE Freezing US Agriculture? Farm-level Adjustment to
Increased Local Immigration Enforcement.” Labour Economics 78.
ILO (International Labour Organization). 2021a. “From Potential to Practice: Preliminary Findings on the
Number of Workers Working from Home during the COVID-19 Pandemic.” Policy Brief, ILO, Geneva.
ILO (International Labour Organization). 2021b. World Employment and Social Outlook: The Role of Digital
Labour Platforms in Transforming the World of Work. Geneva: ILO.
IMF (International Monetary Fund). 2018. “Technology and the Future of Work.” Working Paper, IMF,
Washington, DC.
Isik-Dikmelik, Aylin, Ana Sofia Martinez, Ines Rodriguez, Nancy Banegas, Asta Zviniene, and Polly Jones.
2022. Stronger Social Protection and Labor Systems in Central America for a Resilient and Inclusive
Recovery. Washington, DC.: World Bank.
References 99
Jaller, Lillyana Daza, Simon Gaillard, and Martín Molinuevo. 2020. “The Regulation of Digital Trade: Key
Policies and International Trends.” Research Paper, World Bank, Washington, DC.
Jensen, J. Bradford, and Lori G. Kletzer. 2010. “Measuring Tradable Services and the Task Content of
Offshorable Services Jobs.” In Labor in the New Economy, edited by Katharine G. Abraham, James R.
Spletzer, and Michael Harper, 309–35. Chicago: University of Chicago Press.
Jin, Yizhou, and Zhengyun Sun. 2021. “Lifting Growth Barriers for New Firms: Evidence from an
Entrepreneurship Training Experiment with Two Million Online Businesses.” Job Market Paper, Harvard
Web Publishing, Cambridge.
Kanazawa, Kyogo, Daiji Kawaguchi, Hitoshi Shigeoka, and Yasutora Watanabe. 2022. “AI, Skill, and
Productivity: The Case of Taxi Drivers.” Working Paper 30612, National Bureau of Economic Research,
Cambridge.
Katz, Lawrence F., and Kevin M. Murphy. 1992. “Changes in Relative Wages, 1963-1987: Supply and
Demand Factors.” Quarterly Journal of Economics 107 (1): 35–78.
Katz, Lawrence F., Jonathan Roth, Richard Hendra, and Kelsey Schaberg. 2022. “Why do sectoral employment
programs work? Lessons from WorkAdvance.” Journal of Labor Economics 40 (S1): S249–91.
Kingsley, Sara Constance, Mary L. Gray, and Siddharth Suri. 2015. “Accounting for Market Frictions and
Power Asymmetries in Online Labor Markets.” Policy & Internet 7 (4) :383–400.
Kluve, Jochen, Susana Puerto, David Robalino, Jose M. Romero, Friederike Rother, Jonathan Stöterau,
Felix Weidenkaff, and Marc Witte. 2019. “Do youth employment programs improve labor market
outcomes? A quantitative review.” World Development 114 (February): 237–53.
Kogan, Leonid, Dimitris Papanikolaou, Lawrence D. W. Schmidt, and Bryan Seegmiller. 2021. “Technology,
Vintage-Specific Human Capital, and Labor Displacement: Evidence from Linking Patents with
Occupations.” Working Paper Series 29552, Cambridge.
Konstandini, Genti, Elton Mykerezi, and Cesar Escalante. 2014. “The Impact of Immigration Enforcement
on the U.S. Farming Sector.” America Journal of Agricultural Economics 96 (1): 172–92.
Kostøl, Fredrik B. and Elin Svarstad. 2023. “Trade Unions and the Process of Technological Change.”
Labour Economics 84.
Koolwal, Gayatri and Dominique van de Walle. 2013. “Access to Water, Women’s Work, and Child
Outcomes.” Economic Development and Cultural Change 61(2):369-405.
Koomen, Miriam, and Uschi Backes-Gellner. 2022. “Occupational Tasks and Wage Inequality in Germany:
A Decomposition Analysis.” Discussion Paper 15702, Institute of Labor Economics, Bonn.
Korinek, Anton, Martin Schindler, and Joseph E. Stiglitz. 2021. “Technological Progress, Artificial
Intelligence, and Inclusive Growth.” Working Paper 166, International Monetary Fund, Washington, D.C.
Krenz, Astrid, Klaus Prettner, and Holger Strulik. 2021. “Robots, Reshoring, and the Lot of Low-Skilled
Workers.” European Economic Review 136.
Krenz, Astrid, and Holger Strulik. 2021. “Quantifying Reshoring at the Macro Level-Measurement and
Applications.” Growth and Change 52 (3): 1200–29.
References 100
Kugler, Adriana D., Maurice Kugler, Laura Ripani, and Rodimiro Rodrigo. 2020. “U.S. Robots and their
Impacts in the Tropics: Evidence from Colombian Labor Markets.” Working Paper 28034, National
Bureau of Economic Research, Cambridge.
Lachaud, Michee Arnold, Boris E. Bravo-Ureta, and Carlos E. Ludena. 2017. “Agricultural Productivity in
Latin America and the Caribbean in the Presence of Unobserved Heterogeneity and Climatic Effects.”
Climate Change 143: 445–60.
Lamb, Creig. 2016. “The Talented Mr. Robot: The Impact of Automation on Canada’s Workforce.” Research
Paper, Brookfield Institute for Innovation and Entrepreneurship, Toronto.
Lassébie, Julie, and Glenda Quintini. 2022. “What Skills and Abilities Can Automation Technologies
Replicate and What Does It Mean for Workers? New Evidence.” Social, Employment and Migration
Working Paper 282, Organisation for Economic Co-operation and Development, Paris.
Lederman, Daniel, Julián Messina, Samuel Pienknagura, and Jamele Rigolini. 2014. Latin American
Entrepreneurs: Many Firms but Little Innovation. Washington, DC.: World Bank.
Lekfuangfu, Warn N., Suphanit Piyapromdee, Ponpoje Porapakkarm, and Nada Wasi. 2020. “On Covid-19:
New Implications of Job Task Requirements and Spouse’s Occupational Sorting.” Discussion Paper
Series 12/20, Centre for Research & Analysis of Migration, London.
Levin, Victoria Indhira Vanessa Santos, Michael Weber, Syedah Aroob Iqbal, Ashwani Aggarwal, Paul John
Comyn, Hiromichi Katayama, Margo A. Hoftijzer. Building Better Formal TVET Systems: Principles and
Practice in Low- and Middle-Income Countries. Washington, DC.: World Bank Group.
Lewandowski, Piotr, Karol Madoń, and Deborah Winkler. 2023. “The Role of Global Value Chains for Worker
Tasks and Wage Inequality.” Policy Research Working Paper 10433, World Bank, Washington, DC.
Lewandowski, Piotr, Albert Park, Wojciech Hardy, Yang Du, and Saier Wu. 2022. “Technology, Skills, and
Globalization: Explaining International Differences in Routine and Nonroutine Work Using Survey
Data.” The World Bank Economic Review 36 (3): 687–708.
Lewis, Ethan. 2011. “Immigration, Skill Mix, and Capital Skill Complementarity.” The Quarterly Journal of
Economics 126: 1029–69.
Licetti, Martha Martinez, Tanja Goodwin, Dennis Sanchez Navarro, and Noelia Carreras Schabauer. 2021.
Fixing Markets, Not Prices: Policy Options to Tackle Economic Cartels in Latin America and the
Caribbean. Washington, DC.: World Bank.
Lin, Jeffrey. 2011. “Technological Adaptation, Cities, and New Work.” The Review of Economics and
Statistics 93 (2): 554–574.
Lo Bello, Salvatore, Maria Laura Sánchez-Puerta, and Hernan Winkler. 2019. “From Ghana to America: The
Skill Content of Jobs and Economic Development.” World Bank Policy Research Working Paper 8758,
World Bank, Washington, DC.
Loukos, Panos, and Leslie Arathoon. 2021. “Landscaping the Agritech Ecosystem for Smallholder Farmers
in Latin America and the Caribbean.” Technical Note 2084, IDB, Washington, DC.
Madariaga, Javier, Cesar Buenadicha, Erika Molina, and Christoph Ernst. 2019. “Platform Economy and
Employment: What Is It Like to Work for an App in Argentina.” Executive Summary, enter for the
References 101
Implementation of Public Policies for Equity and Growth, Inter-American Development Bank, and
International Labour Organization, Buenos Aires.
Maloney, William F. 2002. “Missed Opportunities: Innovation and Resource-Based Growth in Latin
America.” Policy Research Working Paper 2935, World Bank, Washington, DC.
Maloney, William F., and Carlos Molina. 2016. “Are Automation and Trade Polarizing Developing Country
Labor Markets, Too?” Policy Research Working Paper 7922, World Bank, Washington, DC.
Maloney, William F., and Carlos Molina. 2019. “Are Automation and Trade Polarizing Developing Country
Labor Markets, Too?” Policy Research Working Paper Series 7922, World Bank, Washington, DC.
Mandelman, Federico S., and Andrei Zlate. 2022. “Offshoring, Automation, Low-Skilled Immigration, and
Labor Market Polarization.” American Economic Journal: Macroeconomics 14 (1): 355–89.
Manyika, James, Michael Chui, Mahdi Miremadi, Jasques Bughin, Katy George, Paul Willmott, and Martin
Dewhurst. 2017a. A Future that Works: Automation, Employment, and Productivity. Washington, DC.:
McKinsey Global Institute.
Manyika, James, Susan Lund, Michael Chui, Jacques Bughin, Jonathan Woetzel, Parul Batra, Ryan Ko, and
Saurabh Sanghvi. 2017b. Jobs Lost, Jobs Gained: Workforce Transitions In A Time Of Automation.
Washington, DC.: McKinsey Global Institute.
Martínez-Plumed, Fernando, Songül Tolan, Annarosa Pesole, José Hernández-Orallo, Enrique Fernández-
Macías, and Emilia Gómez. 2020. “Does AI Qualify for the Job? A Bidirectional Model Mapping Labour
and AI Intensities.” Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, February
2020, 94–100.
Martins-Neto, Antonio, Nanditha Mathew, Pierre Mohnen, and Tania Treibich. 2021. “Is There Job Polarization
in Developing Economies? A Review and Outlook.” Working Paper 9444, CESifo Institute, Munich.
Mas, Alexandre, and Amanda Pallais. 2017. “Valuing Alternative Working Arrangements.” American
Economic Review 107 (12): 3722–59.
Mas, Alexandre, and Amanda Pallais. 2020. “Alternative Working Arrangements.” Working Paper Series
02138, National Bureau of Economic Research, Cambridge.
Maurizio, Roxana. 2021. “Challenges and Opportunities of Teleworking in Latin America and the
Caribbean.” Labour Overview Series, Latin America and the Caribbean 2021, International Labour
Organization, Geneva.
McKenzie, David. 2017. “How Effective are Active Labor Market Policies in Developing Countries? A Critical
Review of Recent Evidence.” World Bank Research Observer 32 (2): 127–54.
McKenzie, David. 2021. “Small Business Training To Improve Management Practices in Developing
Countries: Re-assessing the Evidence for ‘Training Doesn’t Work’.” Oxford Review of Economic Policy
37 (2): 276–301.
Mealy, Penny, R. Maria del Rio-Chanona, and J. Doyne Farmer. 2018. “What You Do at Work Matters: New
Lenses on Labour.” Mimeo.
Meindl, Benjamin, Morgan R. Frank, and Joana Mendonça. 2021. “Exposure of Occupations to
Technologies of the Fourth Industrial Revolution.” Working Paper, arXiv.org.
References 102
MEPyD (Ministerio de Economía, Planificación y Desarrollo). 2022. Encuesta Nacional para la Detección
Necesidades de Habilidades y Cualificaciones en el Empleo. Santo Domingo: National Statistics
Office - Ministry of Economy, Planning and Development.
Meschi, Elena, Erol Taymaz, and Marco Vivarelli. 2011. “Trade, Technology and Skills: Evidence from
Turkish Microdata.” Labour Economics 18: S60–70.
Meschi, Elena, Erol Taymaz, and Marco Vivarelli. 2016. “Globalization, Technological Change, and Labor
Demand: A Firm-Level Analysis for Turkey.” Review of World Economics 152: 655–80.
Messina, Julián, and Joana Silva. 2018. Wage Inequality in Latin America: Understanding the Past To
Prepare for the Future. Washington, DC.: World Bank.
Messina, Julián, and Joana Silva. 2021. “Twenty Years of Wage Inequality in Latin America.” The World
Bank Economic Review 35 (1): 117–47.
Michaels, Guy, Ashwini Natraj, and John Van Reenen. 2014. “Has ICT Polarized Skill Demand? Evidence
from Eleven Countries over Twenty-Five Years.” Review of Economics and Statistics 96 (1): 60–-77.
Michaels, Guy, Ferdinand Rauch, and Stephen J. Redding. 2019. “Task Specialization in U.S. Cities from
1880 to 2000.” Journal of the European Economic Association 17 (3): 754–98.
Michel, Veronica, and Ian Walker. 2019. Honduras: Jobs Diagnostic. Jobs Series. Washington, DC.: World
Bank.
Minnesota Population Center. 2020. Integrated Public Use Microdata Series, International: Version
7.3 (dataset). Censuses of Costa Rica, the Dominican Republic, El Salvador, Guatemala, Honduras,
Nicaragua, and Panama. Integrated Public Use Microdata Series-International, Minneapolis. https://
doi.org/10.18128/D020.V7.3.
Mondolo, Jasmine. 2021. “The Composite Link between Technological Change and Employment: A
Survey of the Literature.” Journal of Economic Surveys 36: 1027–68.
Mongey, Simon, Laura Pilossoph, and Alexander Weinberg. 2021. “Which Workers Bear the Burden of
Social Distancing?” The Journal of Economic Inequality 19: 509–26.
Montenovo, Laura, Xuan Jiang, Felipe Lozano-Rojas, Ian Schmutte, Kosali Simon, Bruce A. Weinberg, and
Coady Wing. 2022. “Determinants of Disparities in Early COVID-19 Job Losses.” Demography 59 (3):
827–855.
Montobbio, Fabio, Jacopo Staccioli, Maria Enrica Virgillito, and Marco Vivarelli. 2022. “The Empirics of
Technology, Employment and Occupations: Lessons Learned and Challenges Ahead.” Discussion
Paper 15731, Institute of Labor Economics, Bonn.
Morris, Michael, Ashwini Rekha Sebastian, and Viviana Maria Eugenia Perego. 2020. “Future Foodscapes:
Re-imagining Agriculture in Latin America and the Caribbean.” Report, World Bank, Washington, DC.
Moroz, Harry, and Indhira Santos. 2022. “Platform Work Globally and in Latin America and the Caribbean:
Overview and Directions in Labor Regulation and Social Protection Policy.” Report, World Bank,
Washington, DC.
Nayyar, Gaurav, and Marcio Cruz. 2018. “Developing Countries and Services in the New Industrial
Paradigm.” Policy Research Working Paper 8659, World Bank, Washington, DC.
References 103
Nayyar, Gaurav, Marcio Cruz, and Linghui Zhu. 2018. “Does Premature Deindustrialization Matter? The
Role of Manufacturing versus Services in Development.” Policy Research Working Paper 8596, World
Bank, Washington, DC.
Nayyar, Gaurav, Mary Hallward-Driemeier, and Elwyn Davies. 2021. At Your Service? The Promise of
Services-Led Development. Washington, DC.: World Bank.
Nedelkoska, Ljubica, and Glenda Quintini. 2018. “Automation, Skills Use and Training.” Working Paper
202, Organisation for Economic Co-operation and Development, Paris.
Nicoletti, Giuseppe, Christina von Rueden, and Dan Andrews. 2020. “Digital Technology Diffusion: A
Matter of Capabilities, Incentives or Both?” European Economic Review 128.
Noy, Shakked, and Whitney Zhang. 2023. “Experimental Evidence on the Productivity Effects of Generative
Artificial Intelligence.” Science 381: 187–192.
Nyman, Sara, and Rodrigo Barajas. 2021. Antitrust and Digital Platforms: An Analysis of Global Patterns
and Approaches by Competition Authorities. Washington, DC.: World Bank.
Ñopo, Hugo. 2012. “Gender Differences in Education in Latin America and the Caribbean: Girls Outpacing
Boys.” In New Century, Old Disparities Gender and Ethnic Earnings Gaps in Latin America and the
Caribbean, edited by Hugo Ñopo, 21–36. Washington, DC.: IDB.
OECD (Organisation for Economic Co-operation and Development). 2017a. Multidimensional Review of
Panama: Volume 1: Initial Assessment. Paris: OECD.
OECD (Organisation for Economic Co-operation and Development). 2017b. OECD Employment Outlook
2017. Paris: OECD.
OECD (Organisation for Economic Co-operation and Development). 2017c. OECD Reviews of Labour
Market and Social Policies: Costa Rica. Paris: OECD.
OECD (Organisation for Economic Co-operation and Development). 2018a. Bridging the Digital Gender
Divide: Include, Upskill, Innovate. Paris: OECD.
OECD (Organisation for Economic Co-operation and Development). 2018b. How Immigrants Contribute
to the Dominican Republic’s Economy. Paris: OECD.
OECD (Organisation for Economic Co-operation and Development). 2020a. Latin American Economic
Outlook 2020: The Digital Transformation for Building Back Better. Paris: OECD.
OECD (Organisation for Economic Co-operation and Development). 2020b. Production Transformation
Policy Review of the Dominican Republic Preserving Growth, Achieving Resilience. Paris: OECD.
OECD (Organisation for Economic Co-operation and Development). 2021. The Digital Transformation of
SMEs. Paris: OECD.
OECD (Organisation for Economic Co-operation and Development). 2022a. Multidimensional Review of
the Dominican Republic: Towards Greater Well-being for All. Paris: OECD.
OECD (Organisation for Economic Co-operation and Development). 2022b. Skills for the Digital Transition:
Assessing Recent Trends Using Big Data. Paris: OECD.
References 104
OECD (Organisation for Economic Cooperation and Development). 2023. Skills in Latin America: Insights
from the Survey of Adult Skills (PIAAC). OECD Skills Studies. OECD. https://doi.org/10.1787/5ab893f0-en.
Oesch, Daniel. 2013. Occupational Change in Europe: How Technology and Education Transform the Job
Structure. Oxford: Oxford University Press.
Oesch, Daniel, and Jorge Rodríguez Menés. 2011. “Upgrading or Polarization? Occupational Change in
Britain, Germany, Spain, and Switzerland, 1990-2008.” Socio-Economic Review 9: 503–31.
Oettinger, Gerald S. 2011. “The Incidence and Wage Consequences of Home-Based Work in the United
States, 1980-2000.” The Journal of Human Resources 46 (2): 237–60.
Oyer, Paul. 2020. “The Gig Economy: Non-traditional Employment Is a Great Opportunity for Many, But It
Won’t Replace Traditional Employment.” IZA World of Labor 471.
Peng, Sida, Eirini Kalliamvakou, Peter Cihon, and Mert Demirer. 2023. “The Impact of AI on Developer
Productivity: Evidence from GitHub Copilot.” Research Paper, arXiv.org.
Posadas, Josefina, and Mauro Testaverde. 2022. “Investing in Labor Market Information Systems: Why
and How?” Presentation at the World Bank Brown Bag Lunch online series, December 2, 2022.
Pouliakas, Konstantinos. 2018. “Determinants of Automation Risk in the EU Labour Market: A Skills-Needs
Approach.” IZA Discussion Papers 11829, Institute of Labor Economics, Bonn.
PwC. 2018. Will Robots Really Steal Our Jobs? An International Analysis of the Potential Long-Term Impact
of Automation. London: PwC.
Quirós, Carlota Tarín, Juan Pablo Villar García, Julio Blázquez Soria, Juan Cruz Trecet, Antonio Tena de la
Nuez. 2022. “La dimensión de género en la transformación digital empresarial de América Latina y el
Caribe.” IDB, Washington, DC.
Rahman, Ahmed S. 2021. “Why Can’t Everybody Work Remotely? Blame the Robots.” COVID Economics
Vetted and Real-Time Papers 36 (10): 105–28.
Reijnders, Laurie S.M., and Gaaitzen J. de Vries. 2018. “Technology, Offshoring and the Rise of Non-Routine
Jobs.” Journal of Development Economics 135: 412–32.
Rodrik, Dani. 2016. “Premature Deindustrialization.” Journal of Economic Growth 21: 1–33.
Ross, Matthew B. 2017. “Routine-Biased Technological Change: Panel Evidence of Task Orientation and
Wage Effects.” Labour Economics 48: 198–214.
Sanchez, Daniel Garrote, Nicolas Gomez Parra, Caglar Ozden, Bob Rijkers, Mariana Viollaz, and Hernan
Winkler. 2021. “Who on Earth Can Work from Home?” World Bank Research Observer 36: 67–100.
Schlogl, Lukas, and Andy Sumner. 2018. “The Rise of the Robot Reserve Army: Automation and the Future
of Economic Development, Work, and Wages in Developing Countries.” Working Paper 487, Center
for Global Development, Washington, DC.
Schroeder, Kateryna, Julian Lampietti, and Ghada Elabed. 2021. What’s Cooking: Digital Transformation
of the Agrifood System. Washington, DC.: World Bank.
References 105
Schultheiss, Tobias, and Uschi Backes-Gellner. 2022a. “Different Degress of Skill Obsolescence across
Hard and Soft Skills and the Role of Lifelong Learning for Labor Market Outcomes.” Industrial Relations
62: 257–87.
Schultheiss, Tobias, and Uschi Backes‑Gellner. 2022b. “Does Updating Education Curricula Accelerate
Technology Adoption in the Workplace? Evidence from Dual Vocational Education and Training
Curricula in Switzerland.” The Journal of Technology Transfer.
Sebastian, Raquel, and Federico Biagi. 2018. “The Routine Biased Technical Change Hypothesis: A Critical
Review.” JRC Technical Reports, European Commission, Luxembourg.
Shafiq, M. Najeeb, Amanda Devercelli, and Alexandria Valerio. 2018. “Are There Long-Term Benefits from Early
Childhood Education in Low- and Middle-Income Countries?” Education Policy Analysis Archives 26 (122).
Sinha, Rishabh. 2020. “Talent Allocation and Post-Reform Growth in Central America.” Policy Research
Working Paper 9380, Washington, D.C., World Bank.
Sinha, Rishabh. 2022. “Central America’s Deindustrialization.” Policy Research Working Paper 10203,
World Bank, Washington, DC.
Sousa, Liliana D., and Andrés García-Suaza. 2018. “Remittances and Labor Supply in the Northern
Triangle.” Policy Research sinWorking Paper 8597, World Bank, Washington, DC.
Spiezia, Vincenzo. 2012. “ICT Investments and Poductivity: Measuring the Contribution of ICTs to Growth.”
OECD Journal: Economic Studies 2012 (1): 199–211.
Spitz-Oener, Alexandra. 2006. “Technical Change, Job Tasks, and Rising Education Demands: Looking
outside the Wage Structure.” Journal of Labor Economics 24 (2): 235–70.
Stemmler, Henry. 2019. “Does Automation Lead to De-Industrialization in Emerging Economies? Evidence
from Brazil.” Discussion Paper 382, Center for European, Governance and Economic Development
Research, Göttingen.
Stöterau, Jonathan. 2019. “Job training programs for youth in low income countries.” PEGNet Policy Brief
15/2019, Kiel Institute for the World Economy, Kiel.
Tapia, Catalina Rodriguez. 2023. “From Informality to Inclusion: Harnessing Public-private Partnerships to
Enhance Social Protection Coverage of Online Gig Workers.” Working Without Borders: The Promise
and Peril of Online Gig Work Short Note Series Issue 4, World Bank, Washington, DC.
Tolan, Songül, Annarosa Pesole, Fernando Martínez-Plumed, Enrique Fernández-Macías, José Hernández-
Orallo, and Emilia Gómez. 2021. “Measuring the Occupational Impacts of AI: Tasks, Cognitive Abilities
and AI Benchmarks.” Journal of Artificial Intelligence Research 71: 191–236.
Ulku, Hulya, and Gabriel Zaourak. 2021. Unleashing Central America’s Growth Potential. Washington,
DC.: World Bank.
UNCTAD (United Nations Conference on Trade and Development). 2023. Technology and Innovation Report
2023: Opening Green Windows: Technological Opportunities for a Low-Carbon World. New York: UNCTAD.
UNESCO (United Nations Educational, Scientific and Cultural Organization). 2017a. Cracking the Code:
Girls’ and Women’s Education in Science, Technology, Engineering and Mathematics (STEM). Paris:
UNESCO.
References 106
UNESCO (United Nations Educational, Scientific and Cultural Organization). 2017b. Digital Skills for Life
and Work. Report by the Broadband Commission for Sustainable Development. Paris: UNESCO.
UNESCO (United Nations Educational, Scientific and Cultural Organization). 2020. Global Education
Monitoring Report – Gender Report: A New Generation: 25 Years of Efforts for Gender Equality in
Education. Paris: UNESCO.
UNESCO (United Nations Educational, Scientific and Cultural Organization). 2022. Global Education
Monitoring Report Paris, UNESCO.
Urquidi, Manuel, and Miguel Chalup. 2023. “The Gender Earnings Gap in Latin America and the
Caribbean: An Analysis of Its Components.” Technical Note 2650, Inter-American Development Bank,
Washington, DC.
USAID (United States Agency for International Development). 2017a. “Guatemala Labor Market
Assessment.” USAID, Washington, DC.
USAID (United States Agency for International Development). 2017b. “Honduras Labor Market
Assessment.” USAID, Washington, DC.
USAID (United States Agency for International Development). 2017c. “Proyecto de USAID Puentes para el
Empleo: Evaluación del Mercado Laboral.” USAID, Washington, DC.
USAID (United States Agency for International Development). 2020. “Dominican Republic Labor Market
Assessment.” USAID, Washington, DC.
USDA (United States Department of Agriculture). 2023. International Agricultural Productivity Data
Product. U.S. Department of Agriculture, Economic Research Service, Washington, DC.
Vegas, Emiliana, and Lucrecia Santibáñez. 2009. The Promise of Early Childhood Development in Latin
America. Washington, DC.: World Bank.
Vivarelli, Marco. 2014. “Innovation, Employment and Skills in Advanced and Developing Countries: A
Survey of the Economic Literature.” Journal of Economic Issues 48 (1): 123–54.
Vona, Francesco, Giovanni Marin, and Davide Consoli. 2019. “Measures, Drivers and Effects of Green
Employment: Evidence from US Local Labor Markets, 2006–2014.” Journal of Economic Geography
19 (5): 1021–48.
Voorend, Koen, Daniel Alvarado, and Luis Ángel Oviedo. 2021. “Future Demand for Migrant Labor in Costa
Rica.” KNOMAD Working Paper 40, World Bank, Washington, DC.
Webb, Michael. 2020. “The Impact of Artificial Intelligence on the Labor Market.” SSRN Electronic Journal.
WEF (World Economic Forum). 2018a. The Future of Jobs Report 2018. Cologny/Geneva: WEF.
WEF (World Economic Forum). 2018b. Towards a Reskilling Revolution: A Future of Jobs for All. Cologny/
Geneva: WEF.
Weinberger, Catherine J. 2014. “The Increasing Complementarity between Cognitive and Social Skills.”
The Review of Economics and Statistics 96 (5): 849–61.
References 107
Weller, Jürgen, Sonia Gontero, and Susanna Campbell. 2019. “Riesgos de la sustitución tecnológica del
trabajo humano y desafíos de la generación de nuevos puestos de trabajo.” Working Paper 201,
Comisión Económica para América Latina, Santiago.
Winkler, Hernan, and Miriam Montenegro. 2021. Jobs Diagnostic: Dominican Republic. Jobs Series.
Washington, DC.: World Bank.
Wood, Alex J. 2019. “Good Gig, Bad Gig: Job Quality and New Sources of Insecurity in the Gig Economy.”
Oxford Internet Institute, Oxford.
World Bank. 2012a. A Gender (R)evolution in the Making. Washington, DC.: World Bank.
World Bank. 2012b. Better Jobs in Nicaragua: The Role of Human Capital. Washington, DC.: World Bank.
World Bank. 2016. World Development Report 2016: Digital Dividends. Washington, DC.: World Bank.
World Bank. 2018. World Development Report 2018: Learning To Realize Education’s Promise. Washington,
DC.: World Bank.
World Bank. 2021a. Digital Skills: The Why, the What, and the How: Part 1. Washington, DC.: World Bank.
World Bank. 2021b. Digital Skills: The Why, the What, and the How: Part 2. Washington, DC.: World Bank.
World Bank. 2022a. A Roadmap for Climate Action in Latin America and the Caribbean, 2021–2025.
Washington, DC.: World Bank.
World Bank. 2022b. “Central America Human Capital Review Phase 1: Costa Rica, Dominican Republic,
Guatemala, and Panama.” World Bank, Washington, D.C.
World Bank. 2022c. Digital Economy for Latin America and the Caribbean Country Diagnostic: El
Salvador. Washington, DC.: World Bank.
World Bank. 2022d. “My Skills, My Career Superpowers: How ‘SkillCraft’ Helps Jobseekers to Self-Assess,
Identify and Develop Skills for Jobs in South Africa.” World Bank, Washington, DC.
World Bank. 2022e. Two Years After: Saving a Generation. Washington, DC.: World Bank.
World Bank. 2022f. “Understanding Migration in North Central American Countries: El Salvador,
Guatemala, Honduras, and Nicaragua.” World Bank, Washington, DC.
World Bank. 2023a. Country Climate Development Report: Dominican Republic. Washington, DC.: World Bank.
World Bank. 2023b. Country Climate Development Report: Honduras. World Bank, Washington, DC.
World Bank, UNESCO (United Nations Educational, Scientific, and Cultural Organization), and ILO
(International Labor Organization). 2023. Building Better Formal TVET Systems: Principles and
Practice in Low- and Middle-Income Countries. Washington D.C.: The World Bank.
World Bank and WTO (World Trade Organization). 2023. Trade in Services for Development: Fostering
Sustainable Growth and Economic Diversification. Washington, DC: World Bank and WTO.
Yusuf, Shahid. 2017. “Automation, AI, and the Emerging Economies.” CGD Notes, Center for Global
Development, Washington, DC.
References 108