Research

I study how knowledge facilitates firm-level growth across space. Using US Census microdata, I identify establishments specialized in knowledge production for other establishments of the firm. Firm growth is disproportionately concentrated near these establishments, suggesting that within-firm geographic frictions inhibit the perfect replication of knowledge across production units in space. I show that these specialized knowledge establishments relate to increased learning and adoption of firm knowledge in the firm’s production establishments. These findings motivate a dynamic model of firm growth via the accumulation of knowledge in which firms jointly determine where to locate their production of knowledge and output. The firm’s problem is dynamic, combinatorial, and features many continuous state and choice variables, so I solve it using a novel computational algorithm. I estimate the model and use it to understand the effects of geographic shocks. Counterfactual analysis demonstrates that firms’ knowledge investment decisions amplify the welfare effects of local productivity shocks by 21% and propagate these effects to other regions.

Relatively flat US productivity growth versus rising R&D expenditures is often interpreted as evidence that ideas are getting harder to find. We build a new 45-year panel tracking the universe of US firms’ patenting to investigate the micro underpinnings of this conclusion, separately examining the relationships between research inputs and ideas (patents) versus ideas and growth. We find that average patents per R&D input are increasing, the elasticity of patents to R&D inputs is flat or rising, and there is not systematic evidence of a secular decline in patenting after controlling for research inputs. We then document a positive, significant, and fairly steady relationship between firms’ patent and labor productivity growth rates. Average firm growth after controlling for patent growth, however, declines. Together, these results suggest that firms’ innovative efforts play a key role in sustaining growth that has not diminished over the last four decades.

How similar are labor and product market power? Using US Census Bureau data on establishment shipments, we show that measuring product market concentration at the location where sales takes place, as opposed to where production occurs, breaks the otherwise tight connection between the two. We find that the strength of the relationship is dictated by the tradability of output in the sector. We develop a multi-region general equilibrium model featuring both types of market power that can rationalize our empirical findings. Equipped with our model, we study how labor and product market power jointly shape the passthrough of productivity to the real wage following a reduction in trade costs.

Using a unique dataset of 30,000 ZIP codes, we examine wage convergence and divergence patterns in the US since 1994, revealing several key insights. First, wage growth follows a U-shaped pattern as a function of initial wages, with convergence below the median and divergence above it. Second, convergence stems from rapid wage growth in low-wage ZIP codes across all commuting zones, driven by their specialization in low-wage services experiencing significant aggregate growth. Third, divergence is driven by “Superstar ZIP Codes” concentrated in high-wage commuting zones that specialize in high-wage business service industries that exhibit fast local wage growth. Finally, while Superstar ZIP codes contribute nearly half of US wage growth over the past 30 years, they generate limited local wage growth spillovers but significantly drive up house prices in their commuting zones.

We investigated the impact of workplace heat and cold on work time loss. Field experiments in different industrial sectors were conducted in multiple countries across all seasons between 2016 and 2024. Hundreds of workers were video recorded, and their full shifts were analyzed on a second-by-second basis. Environmental data were recorded using portable weather stations. The Workplace Environmental Labor Loss (WELL) functions were developed to describe work time loss due to work-place temperature. The WELL function revealed a U-shaped relationship whereby the least work time loss is observed at 18°C (64°F) and increases for every degree above or below this optimal temperature.

Empirical researchers often have to map data provided for a "reporting" spatial unit, say counties in 1900, to a "reference" one, say, counties in 2010. We discuss a general method to create such crosswalks: computing the share of the area of each reporting unit nested in a given reference unit. Using these shares, data can be re-aggregated from the reporting to the reference units. We apply the method to construct a crosswalk for US county-level data since 1790 to present-day counties or commuting zones. We also provide the code to generate other crosswalks given maps of reporting and reference units.