Research

Non-college men have increasingly worked outdoors and left the labor force since the 1970s. I develop and test a task-based model in which heat lowers men’s outdoor productivity and raises the relative value of time at home, drawing them out of work. Using work-hour temperature across US commuting zones in 1980–2019, I find that manufacturing decline pushed men outdoors and that hot days reduce men’s participation through both outdoor and indoor jobs unprotected by air conditioning, with little effect on women. Within commuting zones, more exposed occupations lose both employment and wages, pointing to falling labor demand. Warming explains about 10% of the decline among non-college men, mostly through weak attachment rather than permanent exit. A model-based formula suggests that air conditioning in unprotected indoor workplaces would have avoided about a quarter of the resulting income losses.

Seasonal adjustment removes 80% of within-year variation in unemployment. We develop an econometric framework showing how conventional seasonal filters attenuate estimated climate effects by removing economic variation associated with predictable climate exposure. Linking local weather to unadjusted monthly unemployment rates at the U.S. county level from 1990 to 2019, we find that unemployment would have been 11% lower on average absent extreme temperature days, whereas their effect on seasonally-adjusted series is largely obscured. The effect primarily operates through fewer job openings and hiring, alongside higher separations—especially layoffs—resulting in greater unemployment insurance recipiency and a slacker labor market. We then quantify the nationwide implications of long-run climate change: while milder winters have reduced unemployment, these gains are eroded by accelerating summer unemployment.

Foreign labor represents a growing fraction of risky occupations that appear to be close substitutes of industrial robots. I investigate how dependency on immigrants interacted with robot adoption shaped a workplace injury risk in high-hazard sectors. Associating a wave of unskilled immigrants and workplace injuries across industries during 1992–2019, I find that immigrant workers substantially replaced native fatalities by crowding out natives out of risky jobs. Associated with cross-industry investment of robots, I also find that robot installation dramatically reduced injury risk, but the aggregate nationwide risk remains unabated from poor investments to riskier labor-intensive sectors (e.g., agriculture and construction). Then, I test a hypothesis that immigration inflow impeded the automation and preserved an injury risk for remaining laborers, including natives. Over-dependency on foreign labor may preserve the risky technology generating a social cost (e.g., disabilities; usage of opioids).

This paper revisits the longstanding question of how immigration shapes regional wages through its effects on automation. I develop a dynamic spatial framework in which tasks are allocated among industrial robots, natives, and immigrants, and semi-parametrically recover cross-factor substitution schedules from commuting-zone-level elasticities of economic outcomes with respect to immigration, instrumented by 1940 ethnic settlement patterns. Accounting for immigration-induced automation can reverse the estimated wage effects of immigration over 1980–2015. Low-skilled immigration initially reduces the wages of native workers in routine occupations but raises them in the long run by slowing automation adoption, enhancing domestic welfare. Finally, a universal basic income targeted to natives would increase reliance on both automation and immigrants by raising routine native wages.

This study examines how digital revolutions shape the seniority premium among knowledge workers, using move-level game data from 40 years of professional Japanese chess (Shogi) league. We use two measures of productivity: a relative measure based on annual Grandmaster League rankings and an absolute measure based on the optimality of players’ moves as evaluated by an AI engine. We find that the AI revolution in the 2010s shifted peak performance to older ages, whereas the ICT revolution in the 1990s shifted it to younger ages, according to both measures. These results suggest the AI can reduce age-related disadvantages among knowledge workers by democratizing access to high-quality learning.

We study the impact of climate change on the labor share. Combining newly constructed US county-industry-level labor shares with climate variables, we find that extreme temperatures reduce the labor share, with stronger effects in industries with higher climate exposure and automation potential. Extreme temperatures also accelerate robot adoption. A back-of-the-envelope calculation suggests that the within-county-industry response to climate change accounts for 15% of the decline in labor share since 2000. Over the 20th century, however, the opposing effects of decreased cold days and increased hot days offset each other, consistent with the historical stability of labor share.

Many real-world negotiations are persistently delayed, yet imposing deadlines is costly because it induces inefficient separations. Do all deadlines in one-on-one market transactions need to be perfectly credible? To improve trade efficiency, I propose a mechanism that introduces an intermediate, imperfectly credible deadline to facilitate agreement. Using a canonical seller–buyer dynamic bargaining model with a credible deadline, I analytically characterize the optimal degree of credibility of such a deadline that maximizes trade efficiency, leading to early agreements without triggering separations. Under a non-zero risk of separation, the seller is tempted to discount the price to secure a payoff, while the buyer is more likely to accept at the intermediate deadline because pricing resembles ultimatum offers. A laboratory experiment provides direct evidence of the mechanism’s effects that extend beyond the theoretical benchmark.

A monopolist often exploits a hard deadline to raise their commitment power. I explore whether a group of buyers can employ a soft deadline to counter the monopoly. Using a simple durable goods monopolist model under a deadline, I show that the buyers’ imperfect commitment to an earlier exit may elicit a big sale from the monopolist and generate the buyers’ premium. The soft deadline partially restores the self-competition dynamics of Coase conjecture, which was previously constrained by the hard deadline. In the conventional wisdom on Coase conjecture, the shorter bargaining horizon (or, interpretably, less durability of goods) augments monopoly power. A soft deadline breaks this link: the horizon appears shorter, but the buyers may be better off in expectation.