CURRENT Research
CURRENT Research
with Juan I. Vizcaino
New draft coming soon!
We study how structural transformation shapes sectoral and aggregate productivity dynamics in advanced economies. Persistent productivity gaps between manufacturing and services induce a shift of production resources toward services, generating a compositional slowdown in aggregate productivity. We show that this sectoral reallocation also alters incentives to innovate across sectors, affecting the evolution of sectoral productivity. We assemble a new dataset for the United States (1963–2023) combining sectoral production, employment, and R&D inputs and outputs. The data reveal a divergence in sectoral allocations: production labor shifts from manufacturing to non–research-intensive services, whereas research labor moves from manufacturing toward research-intensive services. To interpret these patterns, we develop a general equilibrium model of structural transformation and directed technical change with endogenous sectoral allocations of labor and innovation, heterogeneous and time-varying markups, and cross-sector knowledge spillovers. In the model, innovation flows toward sectors expanding in size or relative price, while income effects and relative prices govern production reallocation. Calibrated to match key features of U.S. structural change, the model shows that sector-specific markups and knowledge spillovers are central drivers of the observed divergence. Counterfactual exercises imply that, absent structural change in production, aggregate productivity would be 12 percentage points higher, with income effects accounting for half of this gap. Knowledge spillovers from raise productivity by 6 percentage points, whereas heterogeneous markups reduce it by 21 percentage points.
Premature deindustrialisation suggests that services will play an increasingly pivotal role in driving the long-run growth trajectory of developing countries. This paper documents three new facts on productivity growth in services. First, the variance of services productivity growth is higher in developing countries than advanced ones. Second, services productivity growth (or lack thereof) is driven by broad-based improvements (deteriorations) across the different services sub-sectors (`within' channel) as opposed to sub-sector reallocations of employment to more productive services industries (`between' channel). Third, the `within' channel's effect is partly driven by conditional convergence in services productivity between countries. Combined, these facts motivate a theoretical framework where, like the one-sector neoclassical growth model, the social planner faces a standard consumption-investment trade-off. Differently from the benchmark model however, the planner faces an additional static problem which involves allocating investment between two competing technologies. The first is a catch-up technology while the second is a CRS technology. The gains of the former depend on the level of development that is, the further away from the frontier the higher the returns. Crucially, a country-specific growth efficiency parameter interacts solely with the catch-up technology, scaling the marginal gains from this allocation. Productivity growth in the transitional dynamics therefore depends on three factors: the distance to the frontier, the investment rate and the growth efficiency level. Using the World Input-Output Database (WIOD), I calibrate the growth efficiency level to match the 2000-2005 compound annual growth rate of services value added per worker. The calibration outcomes are used for several quantitative exercises, including a variance decomposition of productivity growth in services as well as two counterfactual exercises. A key result of the paper is that assigning sample maximum growth efficiency levels and investment rates raises average growth among developing countries from an observed 3.6% to a counterfactual rate of 18.7%, closing 86.7% of the gap between observed average growth and the maximum potential growth rate.