Technological change and employment in Brazil, Colombia and Mexico: Which workers are most affected?

• DOI: http://doi.org/10.1111/ilr.12104
• Published date: 01 June 2020
• Date: 01 June 2020

Copyright © The authors 2020

Journal compilation © International Labour Organization 2020

International Labour Review, Vol. 159 (2020), No. 2

* Department of Economics and Finance, Universidad del Tolima, Colombia, email: jfariza@

ut.edu.co (corresponding author). ** Department of Economics and Economic History, Universitat

Autònoma de Barcelona, Spain, email: Josep.Raymond@uab.cat.

Responsibility for opinions expressed in signed articles rests solely with their authors, and

publication does not constitute an endorsement by the ILO.

Technological change

and employment in Brazil,

Colombia and Mexico: Which

workers are most affected?

John ARIZA* and Josep Lluís RAYMOND BARA**

Abstract. This article adopts a task-based approach to analyse employment pat-

terns in terms of skill distribution and occupations in the urban labour markets

of Brazil, Colombia and Mexico during 2002–15. The results suggest that employ-

ment fell strongly for some medium-skilled occupations, and increased slightly for

both low-skilled and high-skilled occupations. Decomposition results suggest that

the decreasing share of employment of secretaries and stenographers is fully ex-

plained by changes within industries (routinization hypothesis), whereas the de-

crease in machinery operation and handicraft jobs is mainly explained by changes

between industries. By socio-demographic group, technological changes negatively

aected women but beneted younger workers and those with higher educational

attainment.

Keywords: employment, skills, technological change, Latin America, Brazil,

Colombia, Mexico.

1. Introduction

Advances in computer technology, followed by the development and spread of

information and communications technology (ICT) have been pervasive and

decisive for many global economies over the last three decades. According to

the Organisation for Economic Co-operation and Development (OECD, 2015),

in 2014, the average number of devices online per 100 inhabitants in a sample

of 24 OECD countries was around 14.8 , while for developing economies such

as Brazil, Colombia and Mexico the average was 7.3. The spread of such techno-

logical advances to rms and the population in Latin America has been partly

driven by lower prices and by considerable nancial and commercial reforms

implemented at the beginning of the 1990s, increasing the degree of economic

International Labour Review

138

exposure to international dynamics. We are currently witnessing an acceler-

ation in the eect of ICTs on key economic variables such as employment, prod-

uctivity and wages.

In the eld of labour economics, recent theoretical work has considered

models in which this rapid adoption of computer technologies changes the tasks

performed by workers in their jobs.1 The task-based approach presented by

Autor, Levy and Murnane in 2003 argues that technological changes have led

to workers performing routine tasks being replaced by computers and other

machines (routinization hypothesis). According to this scenario, as the price of

computer capital falls, the demand for workers performing routine activities

decreases, and the demand for workers performing non-routine activities in-

creases. The authors have assumed that routine tasks can be replaced by tech-

nology, that non-routine cognitive tasks are complementary to it, and that

non-routine manual tasks are not directly aected by technological advances.

Contrary to advanced economies, where these new theoretical models have

been widely applied in empirical work, few studies have analysed the demand

for skills in Latin America using a task-based framework. Most of the related

studies in the region focus on the basic theoretical framework that considers

just two groups of workers (skilled and unskilled) and predicts that techno-

logical advances only aect the most skilled (constantly increasing their wages).

So, this “canonical” model, as Acemoglu and Autor (2011) call it, only accounts

for the monotone changes in the demand for skills once technological change

has taken place, and it is only consistent with the traditional skill-biased techno-

logical change hypothesis. This is of course a very restrictive model that rules out

the heterogeneous eect of technology on a variety of tasks and jobs.

One example of non-monotone changes in the demand for skills as a result

of technological advances is the employment pattern referred to as “job polar-

ization” (Goos and Manning, 2007 ). In this pattern, relative employment grows

in both well-paid skilled jobs and low-paid less-skilled jobs, but decreases in jobs

requiring a medium level of skills. The main explanation of job polarization is

the routinization hypothesis, which is linked to the replacement process of cer-

tain tasks previously carried out by workers and now performed by machines.

Empirical evidence of employment polarization in developed countries can be

found in Autor, Levy and Murnane (2003), Autor, Katz and Kearney (200 6), Autor

and Dorn (2009) and Acemoglu and Autor (2011) for the United States; in Goos

and Manning (2007 ) for the United Kingdom; in Spitz-Oener (200 6) and Dust-

mann, Ludsteck and Schönberg (2009) for Germany; and in Goos, Manning and

Salomons (2009) for 16 European OECD countries.

1 The rationale for these kinds of models is straightforward. Workers are endowed with vari-

ous skills used to perform tasks in order to obtain output. A range of tasks can be codied for exe-

cution by computers or other machines and are referred to as “routine” tasks. Record-keeping,

calculations and repetitive assembly are examples of such tasks. By contrast, forming and testing

hypotheses, managing personnel and driving trucks are dened as “non-routine” tasks. The litera-

ture also distinguishes between cognitive and manual tasks within routine and non-routine

classications.