“Workforce Age Structure and Frontier Technology Adoption: Evidence from Newly Released Software”
Workforce aging is often viewed as a drag on economic growth because it reduces labor supply and raises old-age dependency. But aging may also affect growth through a less studied channel: the speed at which economies adopt new technologies. If older workers are less likely to learn new tools, then older labor markets may adapt more slowly when new technologies arrive. Despite the importance of this channel for productivity growth, there is little systematic quantitative evidence on whether workforce age composition affects technology adoption.
This paper fills this gap by estimating the causal effect of local workforce age composition on firms’ adoption of newly released software technologies in the United States, instrumenting the local old-worker share with a Bartik-style measure constructed from historical commuting-zone age composition and national survival rates while allowing for a rich set of controls. We use Lightcast job postings to track local employer demand for 15 newly released software technologies between 2012 and 2016, each matched to an older technology used for similar tasks. Technology adoption is measured by employer demand for frontier software skills, and workforce age composition is measured by the local share of college-educated workers ages 50 to 69.
The empirical results show that, within five years of release, employers in older labor markets post increasingly fewer vacancies requiring the newest software technologies. Moreover, the estimates indicate that the age-related adoption gap is larger when the new software is more difficult to master.
To explain our empirical findings and evaluate policy counterfactuals, we develop and calibrate a simple model linking workers’ retraining decisions to firms’ demand for frontier-technology skills. The key mechanism is finite-horizon retraining: because older workers have shorter remaining careers, they have less time to recover learning costs, making the same skill investment less attractive late in the career. The calibrated model shows that policy margins changing workforce age structure can have different implications for technology adoption. Aging caused by weak inflows of young workers, as under low fertility or low migration, slows adoption by reducing the share of workers with long retraining horizons. By contrast, longer expected working lives, as under delayed retirement, can mitigate this force by increasing older workers’ return to skill investment.
