Gustavo de Souza
Gustavo de Souza
Economist
I received my PhD from the University of Chicago in 2021. Former Postdoctoral fellow at the IIES.
In my research, I use micro-data to estimate macroeconomic models and derive policy implications. My research interests are in macroeconomics, labor economics, development, and public finance.
Contact Information
Publications
What is the most cost-efficient way to impose trade sanctions against Russia?
Journal of Monetary Economics
Trade sanctions are a common instrument of diplomatic retaliation. To guide current and future policy, we ask: What is the most cost-efficient way to impose trade sanctions against Russia? To answer this question, we build a quantitative model of international trade with input-output connections. Sanctioning countries simultaneously choose import tariffs to maximize their income and to minimize Russia’s income, with different weights placed on these objectives. We find, first, that for countries with a small willingness to pay for sanctions against Russia, the most cost-efficient sanction is a uniform, about 20% tariff against all Russian products. Second, if countries are willing to pay at least US$0.7 for each US$1 drop in Russian welfare, an embargo on Russia’s mining and energy products – with tariffs above 50% on other products – is the most cost-efficient policy. Finally, if countries target politically relevant sectors, an embargo against Russia’s mining and energy sector is the cost-efficient policy even when there is a small willingness to pay for sanctions.
with Haishi Li
Can anti-dumping tariffs increase employment?
Forthcoming Review of Economics and Statistics
Can anti-dumping tariffs increase employment? To answer this question we compile data on all anti-dumping (AD) investigations in Brazil, which we match to firm-level administrative employment information. Using difference-in-differences, we estimate the effect of AD tariffs on trade, the protected national suppliers, and the sectors linked to these suppliers. In response to an AD tariff, imports decrease and employment increases in the protected sector. Moreover, downstream firms decrease employment, while upstream ones are not affected. To quantify the aggregate effect of these tariffs, we build a model with international trade, input-output linkages, and labor force participation. The model can reproduce the micro-elasticities we find, as well as the aggregate moments of the Brazilian economy. We show that the Brazilian AD policy increased employment by 0.06%, but they decreased welfare by 2.4%. Using tariffs, the government can increase employment by as much as 2.8%.
with André Victor D. Luduvice
A larger monetary requirement increases welfare by reducing moral hazard discouraging low-pay and temporary jobs
Forthcoming International Economic Review
In the US, unemployed workers must satisfy two requirements to receive unemployment insurance (UI): a tenure requirement that stipulates the minimum qualifying work spell and a monetary requirement that determines a past minimum wage. This paper develops a heterogeneous agents model with history-dependent UI benefits in order to quantitatively obtain an optimal UI program design. We first conduct an empirical analysis using the discontinuity of UI rules at state borders and find that both the monetary and the tenure requirement reduce unemployment. The monetary requirement decreases the number of employers and the share of part-time workers, while the tenure requirement has the opposite effect. We then use a quantitative model to rationalize these results. When the tenure requirement is long, workers tend to accept more low paying jobs to become eligible for UI sooner and to protect themselves from risk, while the monetary requirement works conversely. We show that, because it mitigates moral hazard, the monetary requirement can generate higher welfare levels than an increase in the length of the tenure requirement.
VoxDevLit, 22(1), February 2026.
With Francesco Amodio, Markus Poschke, Bruno Caprettini, Jaedo Choi, Hanwei Huang, Yu-Hsiang Lei, Tristan Reed, Rodimiro Rodrigo, Luis Felipe Sáenz, Marco Sanfilippo, Gustavo de Souza, Michael Sposi, and Verena Wiedemann.
Industrialisation has historically been central to economic development, driving productivity growth, structural transformation, and the creation of stable wage employment for workers with limited formal education. Yet the conditions under which industrialisation delivered broad-based prosperity in the past have changed. Automation, the reshaping of global trade patterns, and the rising importance of services mean that industrial growth today is less likely to generate large numbers of jobs automatically.
This VoxDevLit synthesises evidence on industrial development, drawing on historical experience while focusing primarily on recent evidence and lessons relevant for contemporary policy.
Working Papers
Developing countries rely on technology created by developed countries. This paper demonstrates that such reliance increases wage inequality but leads to greater production in developing countries.
Reject and Resubmit at the Quarterly Journal of Economics
Developing countries rely on technology created by developed countries. This paper shows using model and data that the dependence of developing countries on technology made by developed countries increase wage inequality but leads to higher production in developing countries. I study a Brazilian innovation program that taxed the leasing of international technology to subsidize innovation. Exploiting heterogeneous exposure, I show that the innovation program led firms to replace technology licensed from developed countries by in-house innovations. The replacement of international technology by national technology led to a decline in employment and in the share of high-skilled workers in the firm. I explain these facts with a model of directed technological change and cross-country technology transactions. Firms in a developing country can either innovate or lease technology from a developed country. These two technologies endogenously differ in productivity and skill bias due to factor supply differences in the two countries. I show that the difference in skill bias and productivity can be identified with closed-form solutions by the effect of the innovation program on the firm’s expenditure share and employment. Calibrating the model to reproduce these elasticities, I find that increasing the share of firms patenting in Brazil by 1 p.p. decreases the skilled wage premium by 0.02% and production by 0.2%.
In developing countries, innovation subsidies drive firm growth by facilitating firm entry into high-tariff markets with domestically produced versions of foreign goods.
R&R at the Journal of Political Economy
I study the effect of an innovation subsidy on the growth of firms in a developing country. Using administrative microdata for Brazil and difference-in-differences, I find that innovation subsidies drive firm growth by facilitating firm entry into high-tariff markets with domestically produced versions of foreign goods. After receiving an innovation subsidy, firms issue more patents, expand their workforce, and diversify their product line. However, these patents receive minimal citations, while also heavily citing foreign patents. Firms increase imports of foreign inputs and expand their product line towards products with high import tariff. Despite that, in the most conservative estimate, every $1 of innovation subsidy generated $10 in present value wages.
with Haishi Li
Technological progress in tools – machines that complement labor – has greatly mitigated the negative impact of robots on employment.
R&R at the Review of Economic Studies
What is the effect of robots and tools on employment and inequality? Using natural language processing and an instrumental variable approach, we discover that robots have led to a sizable decrease in the employment and wages of low-skill workers in operational occupations. However, tools — machines that complement labor — have led to an equally large reinstatement of these workers, increasing their employment and wages. Using a quantitative model, we find that the lower prices of robots and tools over the last 20 years have reduced inequality and increased welfare without a significant effect on employment.
Exogenous increase in the relative income of voters causes an increase in public goods provision, contrary to standard political economy theories.
R&R at the European Economic Review
I show that an exogenous increase in the relative income of voters causes an increase in public goods provision, contrary to standard political economy theories. I explain this result with a model of complementarity between consumption and public goods. Estimating the model to reproduce the micro-elasticities, I find that compulsory voting would decrease the government size by 7% despite reducing the average income of voters by 6%.
with Ruben Gaetani, and Martí Mestieri
Import tariffs encourage companies to license their technology abroad, rather than exporting finished products, leading to knowledge diffusion across countries.
How does international trade affect technology diffusion? We show that tariff increases in Brazil lead to more international technology transfers to Brazilian firms and more citations to foreign patents. The highest increase in citations occurs among firms located near those receiving technology transfers, and it is driven largely by citations to firms transferring technology to Brazil. These findings suggest that import tariffs can facilitate the diffusion of foreign technology by promoting technology transfers. We quantify this effect in a growth model that incorporates trade, technology transfers, and their effect on diffusion. When tariffs in Brazil rise, foreign firms transfer their technology rather than export their products, boosting the diffusion of foreign knowledge. An optimal subsidy to technology transfers significantly amplifies the welfare gains from trade liberalization.
The quota for disabled workers in Brazil increased the employment of disabled workers at a large employment and welfare cost for nondisabled workers.
R&R at the JPE: Micro
I study the effect of a quota for disabled workers on the labor market and on welfare. Using a task-based model, I show that the effect of a quota will depend on the productivity of disabled workers and their labor supply elasticity. I estimate the productivity of disabled workers using variation from inspections of the quota. I find that the quota increased the hiring of disabled workers, but it reduced wage and employment of non-disabled workers, suggesting that disabled workers are of low-productivity. I estimate the labor supply elasticity of disabled workers using heterogeneous exposure across regions. I find that the quota increased the wage and labor force participation of disabled workers. Using the model calibrated to the empirical estimates, I find that the quota for disabled workers decreased welfare by 0.33% and forced the government to increase marginal taxes. However, alternatively, a subsidy for disabled workers could increase welfare by 0.29%.
Voters who could benefit from welfare policies vote against them because they hold negative ideologies against redistribution.
What are the quantitatively relevant determinants of redistribution? I describe a structural method to identify different channels affecting the social choice of redistribution and estimate the counter-factual effect of institutional reforms. The method relies on a dynamic heterogeneous agents model estimated using micro-data on voting and on the support for redistribution. I found that pecuniary gains play a negligible role in shaping redistribution. Voters who could benefit from redistribution support and vote for low transfers motivated by social preferences and not pecuniary gains. Because social preferences are more important than income in predicting support for redistribution, increasing voter turnout or capping campaign contributions would have no effect on redistribution.
with Jack Mannion, and Jacob Herbstman
The demand for new skills in computer programming has contributed to wage inequality both within the programming field and between programmers and other occupations
R&R at Labour Economics
We study how the demand for programming skills has impacted inequality. We create a new dataset with information on wages, employment, and software of Brazilian programmers, covering the period from the birth of information technology (IT) to the rise of artificial intelligence (AI). High-ability, high-wage, and highly educated individuals in key technology hubs are more likely to become programmers. Creating software boosts both wages and career prospects of programmers, especially for those with specialized skills in AI and cybersecurity. These wage gains are concentrated among top programmers, increasing inequality within the profession. Therefore, increased demand for specialized skills in programming has contributed to wage inequality both within the programming field and between programmers and other occupations..
.
I use administrative data on artificial intelligence (AI) software created in Brazil to study its effects on the labor market. Owing to a unique copyright system, Brazilian firms have registered their software with the government since the 1980s, creating a detailed record of nearly all commercial AI applications developed in the country. Drawing on this registry, I show that AI is widely used not only in administrative tasks but also in production settings, where it primarily supports process optimization and quality control. Using an instrument based on variation in software development costs, I find that AI affects administrative and production workers differently. Among office workers, AI reduces employment and wages, particularly for middle-wage earners. Among production workers, it increases employment of low-skilled and young workers operating machinery. These results suggest that AI displaces routine office tasks while making machines more productive and easier to operate, leading to a net increase in employment.
with Andrew J. Fieldhouse, Karel Mertens, Ishan B. Nath & Valerie A. Ramey
U.S. public R&D spending raises total factor productivity in other countries.
This paper provides the first causal estimates of the global social returns to U.S. public R&D. We use a narrative identification strategy to quantify the effects of exogenous shocks to U.S. R&D appropriations on the dynamic TFP response of 69 foreign economies from 1980–2019. A U.S. R&D appropriations shock equal to 1 percent of the federal R&D capital stock raises foreign TFP by approximately 1 percent after 12 years. This response is driven primarily by nondefense rather than defense R&D and is concentrated in non-OECD economies. These patterns are most consistent with diffusion occurring through openly accessible scientific content, capital-embodied trade, and technological leapfrogging by economies further from the global frontier. A back-of-the-envelope calculation suggests that the global social returns to U.S. public nondefense R&D are roughly twice as large as the domestic returns, meaning the U.S. captures about half of these productivity benefits..