This paper studies how workers respond to skill-specific technological innovations through investment in training. We construct a novel measure of technological exposure by combining patent texts with O*NET skill descriptors using natural language processing. Linking this measure to UK longitudinal worker data, we find that greater technological exposure increases hourly wages, particularly among highly educated workers and those in non-routine occupations. Exposure also raises participation in training, especially training for future employment opportunities, and increases voluntary employer mobility. We develop a model of multidimensional skill investment under uncertain future skill returns. Higher skill returns provide more resources to pay for training, and allowing training to respond amplifies the output gains from technological change. Greater uncertainty about future skill returns, such as due to the unpredictable labour-market effects of AI, encourages workers to acquire a more balanced set of skills, giving them more flexibility to respond as skill returns change. These findings highlight workers’ proactive investment in skills and pursuit of new employment opportunities as important margins of adjustment to technological innovation.

This paper studies the dynamics of unemployment (u) and its natural rate (u^*), with u^* measured by real-time estimates for 29 countries from the Organization for Economic Cooperation and Development. We find strong evidence of hysteresis: an innovation in u causes u^* to change in the same direction, and therefore has permanent effects. For our baseline specification, a one-percentage-point deviation of u from u^* for 1 year has a long-run effect of 0.16 points on both variables. When we allow asymmetry, we find, perhaps surprisingly, that decreases in u have larger long-run effects than increases in u.

We use a simple theoretical framework, a building block of many macroeconomic models, to study the prominently debated relationship between the model parametrisation of the Frisch elasticity and the reduced-form evidence on the elasticity of labour. Focusing on tax holidays, we show that the elasticity measured with a reduced-form approach is only equal to the Frisch-elasticity parameter if there are no income or general equilibrium effects. Furthermore, for a wide range of standard values of the Frisch-elasticity parameter, the response of labour generated by a tax holiday in the model is aligned with the reduced-form evidence.

The share of single households in the U.S. has been increasing over the past few decades. Single and married households differ in both their exposure to income risk and in the assets they hold. This paper asks how these differences affect the economy’s response to aggregate shocks. Using the U.S. Current Population Survey and Survey of Consumer Finances, I show that it is rare for both spouses in a dual-earner couple to become unemployed in the same month. Married households also hold more wealth per person, even after controlling for income and demographic characteristics, while singles hold roughly twice as large a share of their portfolios in liquid assets. I develop a two-asset Heterogeneous Agent New Keynesian (HANK) model with single women, single men and married couples. Households choose consumption and savings in liquid and illiquid assets. The model is calibrated to the observed household type-specific employment transition rates and liquid asset shares. Income pooling cushions individual earnings losses within couples, while their larger illiquid asset shares limit consumption smoothing. The aggregate implications of the growing share of single households therefore depend on the interaction between these insurance and savings channels.

We document “zombie products” in U.S. consumer goods markets: Products with predictably weak sales that persistently remain on retailers’ shelves. Using Nielsen scanner data, we find that sales in the first twelve weeks predict subsequent performance, yet weak and strong early performers have the same median observed lifespan of four years in our sample. We measure the prevalence of zombie products across categories and over time and study whether retailer adjustment costs, contracts with manufacturers or demand for niche varieties explain their persistence. We examine whether their persistence limits entry and experimentation with new products, and what this implies for aggregate productivity growth and consumer welfare.