with Francisco (Paco) Buera, Joseph Kaboski, and Richard Rogerson
The Review of Economic Studies
PDF Publication Replication Package Online Appendix BibTex Citation
Using a broad panel of advanced economies, we document that increases in GDP per capita are associated with a systematic shift in the composition of value added to sectors that are intensive in high-skill labor, a process we label as skill-biased structural change. It follows that further development in these economies leads to an increase in the relative demand for skilled labor. We develop a quantitative two-sector model of this process as a laboratory to assess the sources of the rise of the skill premium in the US and a set of ten other advanced economies, over the period 1977 to 2005. For the US, we find that the sector-specific skill-neutral component of technical change accounts for 18-24% of the overall increase of the skill premium due to technical change, and that the mechanism through which this component of technical change affects the skill premium is via skill-biased structural change.
with Omar Licandro
Paper Online Appendix CES ifo Working Paper Replication Files Slides
We study how to measure welfare-relevant growth in economies undergoing structural change. Within a continuous-time dynamic general equilibrium model of structural transformation, we compare two welfare-based measures of real income growth: a current-base equivalent variation measure and a chained Fisher–Shell index. The chained Fisher–Shell index coincides with the chained Divisia index, providing theoretical foundations for national accounting methodology. The current-base index, by contrast, requires systematic revisions and introduces base-dependent biases that distort the evaluation of past growth. Our quantitative analysis shows that these biases are substantial, increase over time, and conflict with key stylized facts in the structural transformation literature.
with Selim Elbadri
We study how structural change shapes aggregate productivity through the allocation of production and innovation. Using U.S. data, we document that production employment shifts from Goods toward non-research-intensive Services, while research employment shifts predominantly toward research-intensive Services. We develop a general equilibrium model in which sectoral composition, capital deepening, and directed innovation are jointly determined, allowing for heterogeneous markups and cross-sector knowledge spillovers.
Holding initial value-added shares fixed while retaining baseline sectoral productivity paths raises average annual aggregate productivity growth by 0.13 percentage points over 1950-2023. Removing the forces driving changes in value-added shares and re-solving the equilibrium raises growth by 0.37 percentage points, leaving aggregate productivity approximately 31% higher in 2023. An exact decomposition attributes 62% of this growth difference to the endogenous response of sectoral productivity.
Innovation provides the largest positive contribution: sustaining the Goods sector's market share supports research and idea accumulation, an effect reinforced by the sector's reliance on its own knowledge stock. These findings show that the productivity consequences of structural change depend not only on the reallocation of production, but also on the innovation incentives generated by that reallocation.
This paper examines how occupational structure shapes cross-country differences in human capital stocks. Using harmonized microdata, I document that cognitive occupations are associated with higher schooling and higher returns to education and experience, underscoring their greater skill intensity. I then develop a dynamic model of schooling, on-the-job training, and occupational choice, calibrated and structurally estimated with data from Brazil and the United States. The results show that while average human capital in cognitive occupations in the U.S. is only about 10 percent higher than in Brazil, the gap widens to nearly threefold in routine jobs. A gap decomposition attributes 25 percent of the U.S.–Brazil gap to differences in the occupational structure, 30 percent to the higher human capital embodied in U.S. routine work, and the remainder to relative occupational efficiency and cognitive skill differences.
with Jorge Sabat
We estimate a structural life-cycle model of post-secondary education to evaluate the long-run effects of expanding student credit, using Chile's 2006 introduction of the Crédito con Aval del Estado (CAE). Students differ in ability and parental resources, are assigned to institutions of differing quality, and choose a programme, a full-time or working track, and whether to borrow. Enrolment is financed out of a parental transfer, earnings while studying, grants and loans, so credit constraints bite through a forced-dropout margin as well as through enrolment. Two frictions shape who benefits. Students from households without a university education misread observed average earnings as their own return, which distorts programme choice; and the suitability of a programme is learned only after enrolling, so part of what credit finances is the option to find out. In the working phase, borrowing constraints and loan repayment depress on-the-job investment, flattening the earnings profiles of exactly the graduates the policy was meant to help. We estimate the model on administrative and survey data on attainment and loans linked to longitudinal earnings from Chile's pension system, and decompose the reform's effect on enrolment, completion, programme choice and lifetime earnings into access, financing and information channels, with heterogeneity by gender, parental education and field of study.
with Selim Elbadri
We study the optimal allocation of research across sectors in an economy undergoing structural change. In a three-sector growth model with directed technical change, non-homothetic demand, heterogeneous markups, and knowledge spillovers, the private value of a patent differs from its social value for three reasons: innovators do not fully appropriate the gains their patents generate; they fully ignore the knowledge spillovers they create, within and across sectors; and free entry into research ignores congestion. We characterize the gap as a decomposition of the ratio between the planner's shadow value and the market value of a patent, and we quantify it in a model calibrated to US sectoral data. Appropriability and spillovers push the private return below the social return, while congestion pushes it above, so the sign and size of the gap vary across sectors. Spillovers are strongest in goods, and the optimal allocation shifts research toward that sector. Because sectoral outputs are complements, this reallocation lowers the relative price of goods and moves expenditure further toward services, so the optimum raises the measured Baumol drag on aggregate productivity growth while raising welfare. We use the model to compare the optimal sector-specific research subsidy with a sector-neutral subsidy, as in current US policy.
"Monetary Policy with Declining Deficits: Theory and an Application to Recent Argentine Monetary Policy"
with Rody Manuelli
St. Louis Federal Reserve Review, Fourth Quarter, 2017.
with Rody Manuelli
In this paper we develop a continuous time stochastic growth model that is suitable for studying the impact of natural disasters on the short run and long run growth rate of an economy. We find that the growth effects of a natural disaster depend in complicated ways on the details of expected foreign disaster aid and the existence of catastrophe insurance markets. We show that aid can have an influence on investments in prevention and mitigation activities and can delay the recovery from a natural disaster strike.