Principal investigator: Bruno Benevit
Authors: Luis R. Díaz Pavez and Inmaculada Martínez-Zarzoso
Original title: The impact of automation on labor market outcomes in emerging countries
Location of the Intervention: Pais in desenvolvimento
Sample Size: 1.105 sectors-country-year
Sector: Labor Economics
Primary Variable of Interest: JOBS
Type of Intervention: Job automation
Methodology: IV-FE, FE
Summary
Automation is seen as a transformative factor in labor markets, with the potential to alter how tasks are performed and how added value is distributed. The increased adoption of technologies, both through local and foreign robots, occurred within a context of productive restructuring observed in emerging countries in recent years. To assess the impacts of this process, this study used an instrumental variables (IV) methodology to analyze 16 sectors in 10 emerging countries between 2008 and 2014. The results revealed that exposure to foreign robots negatively affected employment and the fraction of added value attributed to labor, while the effects of local robots varied according to the sector. Additionally, the presence of [missing information - likely a continuation of the previous sentence] was identified. spillovers intersectoral factors that adjusted labor indicators.
- Policy Problem
Driven by the increasing adoption of advanced technologies, automation has significantly altered the way work activities are organized (Díaz Pavez; Martínez-Zarzoso, 2024). This change allows devices and systems to replace repetitive tasks, directly impacting the structure of jobs and modifying the qualification requirements of workers. The transformation of production processes through automated technologies generates changes in the daily lives of organizations, opening space for debates about the redistribution of operational functions and the adaptation of professional profiles.
In recent years, the implementation of automated solutions has intensified amidst periods of economic instability and restructuring of productive sectors. The accelerated advancement of these technologies, especially in industrial environments, has highlighted a rapid transformation in the way activities are performed, with the use of robots and intelligent systems becoming increasingly widespread. This evolution has led to the replacement of simple and routine tasks, driving internal reorganization within companies and modifying the demand for certain skills.
In developing countries, the automation process presents varied characteristics according to the local economic structure. It is observed that, in certain sectors, the adoption of automated technologies results in a reduction of activities traditionally performed by humans, while in others there is a reconfiguration of production methods. This scenario highlights that the impacts of automation manifest themselves in different ways, reinforcing the need to understand the consequences of this process in order to adjust production strategies and promote a more effective adaptation in the markets of these countries.
- Policy Implementation Context
Literature indicates that the effects of automation on the labor market depend on the origin of the technology (Díaz Pavez; Martínez-Zarzoso, 2024). Regarding the adoption of local robots, evidence suggests that, although the replacement of routine tasks is observed, productivity gains can offset job losses, resulting in less pronounced impacts on employment (Acemoglu; Restrepo, 2019). Conversely, when it comes to exposure to foreign robots, the implementation of technologies from countries with higher automation intensity tends to be associated with more noticeable reductions in labor demand, especially in sectors with standardized activities.
In addition to direct changes in the labor supply, the adoption of automated systems can affect other market indicators, such as wage levels and the share of work. In environments with greater exposure to imported robots, competitive pressures between sectors can lead to adjustments not only in employment but also in the remuneration paid.
In contrast, the implementation of technology from local sources is usually associated with more subtle changes in these secondary variables. Thus, the impact of automation can vary depending on the origin of the robots and the production contexts analyzed.
- Evaluation Details
The study considered two main databases, originating from World Input-Output (WIOD) and of International Federation of Robotics (IFR). WIOD provides structured data on sectoral economic indicators, such as employment, wages, capital, and value added, allowing for the analysis of productive and intersectoral relationships. IFR, on the other hand, provides information on the stock of industrial robots, which serves as a basis for measuring the degree of automation in production processes. The integration of this information makes it possible to examine technological transformations and their implications for the economic organization of emerging countries.
The variables considered included employment, measured by the total number of workers, nominal wages per worker, inventory and return on capital, as well as the labor fraction, which indicates the share of labor in value added. Specific automation indicators were also developed, such as the number of local robots per thousand workers and an index of exposure to foreign robots, which allow for the identification of nuances in the transformation of production processes.
WIOD data from 2004 to 2014 demonstrate that, while developed countries showed high levels of automation from the early years analyzed, emerging economies started with almost no robot presence, only reaching significant levels over time. The labor share in emerging countries remained consistently lower than in developed countries, indicating a distribution of added value that favors capital. At the end of the analyzed period, developing countries still had fewer than 500 robots and a labor share of around 0,47, while in developed countries, these indicators approached 2 million robots and a labor share of 0,57, respectively.
Similarly, data from the IFR indicated that the stock of robots in emerging countries grew consistently between 2008 and 2014. This upward trajectory demonstrates the late start in the incorporation of these technologies in these economies, in contrast to the high levels recorded in developed countries, highlighting variations in the pace of adoption among different countries.
The integration of WIOD and IFR data also allowed for the measurement of an index of exposure to foreign robots, calculated from bilateral flows of intermediate inputs. This indicator revealed that, during the same period, some sectors and emerging countries show high levels of exposure, especially in segments linked to automotive, electronics, and metallurgical production, while others register lower indices. These patterns highlight differences in the penetration of imported technologies and emphasize the variability in the organization of production processes in the economies analyzed.
- Method
To analyze the impact of automation on the labor market in developing countries, a model was defined based on a specification of the production function that related employment, wages, capital, and added value. This approach integrated technological factors when considering the incorporation of automation into the production process. The methodology employed instrumental variable (IV) estimation to address the problem of endogeneity of automation indicators.
The instruments were chosen based on data that reflected exogenous variation in countries with similar production segments, representing the variation in robot inventories in countries with comparable production structures and that functioned as proxy for robotics adoption in each emerging country. The models considered the sectoral panel structure, including fixed effects (FE) that controlled for unobserved differences between countries and sectors, as well as annual shocks. The analyses covered 16 sectors in 10 emerging countries, with data recorded between 2008 and 2014.
In the main analysis of the study, FE and IV-FE models were considered. Regarding the outcome variables, the effects of automation on: (i) employment, (ii) nominal wage per worker, and (iii) the fraction of value added directed to labor were estimated. To control for the productive context in the estimated models, value added and the index of... were incorporated as covariates. inshoring and measures relating to capital, among others.
The study also explored specifications that investigated indirect effects and interactions between sectors. These analyses maintained the FE and IV-FE strategies and the controls present in the main analysis, but incorporated variables from spillovers, which measured the impact of robotic adoption on other productive segments (spillover effect). The additional specifications integrated indicators of exposure to foreign technologies and the stock of local robots at aggregate levels. Furthermore, the authors estimated a model considering the effects of automation separately on each of the sectors.
- Main results
The results of the main analyses indicated that, in both the FE estimation and the IV-FE approach, the adoption of local robots did not significantly affect employment in the samples from emerging countries. Conversely, exposure to foreign robots showed a negative and consistent effect on the number of workers, such that a percentage increase in this index implied a proportional reduction in employment. Furthermore, the models demonstrated that the variables relating to nominal wages per worker and the fraction of value added directed to labor did not register statistically significant variations.
In the set of analyses on indirect effects of spilloversIt was found that automation in neighboring sectors resulted in a residual negative impact on employment and nominal wages. This effect highlighted competitive pressures that led to reductions in job supply and wage adjustments, albeit of low magnitude. However, the share of added value attributed to labor remained stable, showing no significant changes in response to the identified indirect effects.
Regarding the sectoral analysis, it was observed that the impacts of foreign robots on employment were negative and more pronounced in sectors with greater exposure to automation from abroad, which also showed a reduction in the fraction attributed to work. In turn, the effects of local robots varied according to the segment, with some cases associated with increases in employment and others with decreases. This evidence indicates that the impact mechanisms of automation presented themselves differently among the sectors analyzed.
- Lessons in Public Policy
This article analyzes how automation has influenced labor markets in emerging countries, considering the adoption of locally produced robots and exposure to foreign robots. The results indicate that the implementation of automated technologies through imported robots has had a negative effect on employment and labor's share of added value, while the effects of locally produced robots varied according to the sectoral context. Furthermore, the analysis identified the presence of indirect effects resulting from the interaction between sectors, reinforcing the variations in labor outcomes.
The evidence found in this study allowed us to understand the mechanisms that link production processes to technological transformations, providing support for the identification of differentiated patterns by sector. This configuration of results helps guide the formulation of public policies that seek to adjust production processes to the challenges posed by automation, supporting the definition of strategies that mitigate the impacts resulting from the convergence between local and foreign technologies.
References
ACEMOGLU, Daron; RESTREPO, Pascual. Automation and New Tasks: How Technology Displaces and Reinstates Labor. Journal of Economic Perspectives, v. 33, n. 2, p. 3–30, 1 May 2019.
DÍAZ PAVEZ, Luis R.; MARTÍNEZ‐ZARZOSO, Inmaculada. The impact of automation on labor market outcomes in emerging countries. The World Economy, v. 47, no. 1, p. 298–331, Jan. 2024.
