Automation and the Shadow Economy: A Negative Association Across 54 Countries
A new HAPRI study, co-authored by Nguyen Doan, Canh Phuc Nguyen (HAPRI Senior Research Associate) and Nadia Doytch, finds a negative association between automation in production and the size of the shadow economy across 54 countries. The authors suggest that automation might have unexpected benefits for the economy, given its negative association with the size of the shadow economy.
Context
Earlier research has found the shadow economy to cause several problems, such as a reduction in government tax revenue, bias in official indicators, and social fairness distortion. One line of argument in the literature links it to automation: automation might cause higher unemployment, the increase in unemployment is one of the determinants of the shadow economy, and so automation might cause a larger shadow economy.
Yet, as the authors note, the evidence of the impact of automation on unemployment so far is conflicting. The study therefore sets out to identify a new driver of the shadow economy in the modern economy by analyzing whether automation in production is associated with this sector.
Methods
The authors estimate the relationship between automation in robotic operations and the shadow economy in a panel sample of 54 countries from 2002 to 2018. Automation is measured with the operational stock of industrial robots, collected from World Robotics Reports (International Federation of Robotics). The shadow economy, as a percentage of GDP, is estimated using three different models: the MIMIC and DGE models from the World Bank and the Medina & Schneider model from the IMF. The estimates control for country heterogeneity, economic structure and institutional quality, and are repeated with alternative measures of robot intensity. A supplementary system of equations also examines unemployment and the service sector; the authors state that this analysis is not intended to identify causal mediation effects.
Key findings
Automation shows a negative association with the size of the shadow economy as a percentage of GDP.
This relationship is relatively consistent after controlling for country heterogeneity, economic structure and institutional quality, and when alternative measures of robot intensity are used.
When the sample is split by income level, the coefficients of robot adoption in low- and middle-income economies differ in sign across the shadow economy measures, and none of them is statistically significant.
The supplementary analysis provides little evidence of a systematic relationship between robotics operation and unemployment.
Policy implications
The authors write that while governments cannot directly control the pace of automation in private industries, and policies promoting automation would not necessarily lead to greater economic formalization, more investment in workforce reskilling, digital infrastructure and regulatory reforms in the labor market are crucial as economies become more automated. The study suggests that concerns about automation should not focus only on possible employment displacement, given the negative association between robotics operation and the shadow economy.
Keywords:
Automation
Industrial Robots
Shadow Economy
Robot Intensity
Institutional Quality
Panel Data
Link:
Citation:
Doan, N., Nguyen, C. P., & Doytch, N. (2026). Automation and the shadow economy. Research in Economics, 80(4), 101186.
