Human Capital, Female Employment and Electricity: Evidence from the Early 20th-Century United States

The Review of Economic Studies, 2024

This paper revisits the link between electrification and the rise in female labor force participation (LFP), and presents  theoretical and empirical evidence showing that electrification triggered a rise in female LFP by increasing market opportunities for skilled women. I formalize my theory in an overlapping generations model and find that my mechanism explains one quarter of the rise in female LFP during the rollout of electricity in the United States (1880-1940), and matches the slow decline in female home-production hours during this period. I then present micro-evidence supporting my theory using newly digitized data on the early electrification of the United States.

Risk-Taking Adaptation to Macroeconomic Experiences (with Remy Levin)     

We study how lifetime experiences of macroeconomic volatility shape individual risk attitudes. We build a Bayesian model where risk aversion endogenously adapts to agents’ beliefs about an exogenous income process. We combine panel data from Indonesia and Mexico containing elicited measures of risk aversion with state-level real GDP growth time series capturing individuals’ lifetime macroeconomic experiences. In line with the model’s predictions, we find that measured risk aversion increases with macroeconomic volatility, and that this is a first-order driver of risk attitudes. These results are robust to many alternate specifications and controls and extend to risk-taking behavior in other domains.

How do Workers Learn? Theory and Evidence on the Roots of Lifecycle Human Capital Accumulation (with Xiao Ma and Alejandro Nakab)  [NEW VERSION COMING SOON]

How do the sources of worker learning change over the lifecycle, and how do these changes affect human capital and wages? We use data from Germany and the US to document that internal learning (learning from colleagues) decreases with workers' experience, whereas external learning (on-the-job training) follows an inverted U-shape pattern. We build a search model featuring a two-source learning technology where the incentives to learn from each source change as workers climb the human capital distribution within firms. Quantitative results highlight the role of internal learning in driving lifecycle wage growth due to compensation for learning spillovers among coworkers.

Revisiting the Link between Electrification and Fertility: Evidence from the Early 20th-Century United States

The decline in fertility occurring throughout the first half of the 20th century in the United States and preceding the baby boom remains largely unexplored. This paper presents empirical and theoretical evidence linking this decline to the spread of electricity. Using data on early electrification efforts, I empirically disentangle two channels linking electrification and fertility: the introduction of time-saving appliances that reduce the time needed for child-rearing; and the rise in female wages which raises the opportunity cost of childcare. I then use these empirical estimates to calibrate a model that features both channels and quantifies the aggregate impact of electrification on fertility. I find that electrification explains 3.1% of the overall fertility decline in 1900-1940 in the US, and that this decline is driven by young childless women who can reap the labor market gains of electricity.

Who Pays for Training? Theory and Evidence on Firm-Level Differences in Training Investments (with Xiao Ma and Alejandro Nakab)

We investigate how on-the-job training patterns vary with firm characteristics, and how this informs the distribution of training costs between firms and workers. Analyzing data from over 100 countries, we show that on-the-job training opportunities are consistently lower in smaller firms. Using administrative firm-level data from China and Mexico, we then show that differences in labor share and productivity levels across firms are key mechanisms driving this pattern. We build a general equilibrium model with firm heterogeneity and training expenditures to provide insights into these findings. We explore various training cost-sharing schemes, and show that only those where firms cover a significant portion of  costs after the match is formed align with our empirical observations. We then consider a calibrated version of the model and show that a scenario where firms pay a calibrated fixed share of training costs generates the most reasonable training returns matching the literature. Within this model, we also find substantial inefficiencies in the provision of training, which are more pronounced in smaller firms. Finally, we conduct a policy exercise that indicates the optimal training subsidy rate is larger for smaller firms, but that even a policy providing the same subsidy rate to all firms can generate a 7% increase in net output in the US.

A Historical Measure of Risk Preference (with Remy Levin)