Research

This paper examines the macroeconomic impact of chronic disease in the United Kingdom (UK). We use individual-level data to estimate how diagnoses of six major diseases affect labor market transitions and combine these with a tractable growth model with age-specific productivity and labor force participation to quantify the impact of chronic disease on UK economic growth. Using a novel machine learning approach to classify National Health Service (NHS) cost data, we also provide new estimates of disease-specific treatment costs. Our findings indicate that a 20% reduction in disease incidence would increase annual GDP by 0.99% after five years and 1.51% after ten years. Most of the gains are due to increased participation in the labor force, especially among workers aged 50 to 65 years. Reductions in mental health conditions and musculoskeletal conditions contribute the most to these effects. Our analysis points to three important features of preventative health policies: 1) the potential welfare gains are substantial and manifest themselves in terms of both improved population health and increased output growth, 2) only around 40% of long-term effects appear after five years, and 3) the 50-65 age group experiences the largest labor force participation gains. This last feature is due to two factors: improved health at those ages prevents transitions into health-related inactivity and a larger share of workers reaches this age band as a result of reduced transitions into inactivity at earlier ages. This compounding effect underscores the importance of targeting prevention efforts at earlier ages.

This paper studies the effects of bank failure on firm performance. We collect 36 million loan records to build a novel dataset on the credit relationships of 1.8 million US firms, predominantly composed of small and medium-sized enterprises (SMEs). We then implement a staggered treatment difference-in-differences estimation strategy using 179 bank failures from 1990 to 2023 to estimate the impact of bank failure on firm-level survival and employment growth. Although the US regulatory framework resolves failed banks through forced acquisitions by healthier institutions—a process designed to minimize disruption—we find substantial negative effects. Firms with a credit relationship to a bank that subsequently fails are 6.7 percentage points (44.3%) more likely to fail themselves within five years, while surviving firms exhibit 25% lower employment growth compared to those banking with non-failed institutions. These impacts persist for more than 10 years, are evident during both crisis and non-crisis periods, and are strongest among very small firms—a firm size segment that we are the first to study in this context. Our estimated effects are further supported by two natural experiments. Surprisingly, we also observe that a small subset of bank failures had positive effects on firm outcomes, suggesting that, in some cases, bank failure can be fortuitous for affected firms. Overall, our findings suggest that bank failures exert a substantially larger influence on the real economy than previously recognized, possibly requiring a re-evaluation of current regulatory approaches to managing such events.

This paper develops a task-based model of lifecycle occupational sorting in which workers are endowed with three skill dimensions—physical, fluid cognitive, and crystallized cognitive—that follow distinct biological trajectories over the lifecycle. Productivity in a task depends on the distance between a worker's skill vector and the task's requirement vector, generating dynamic comparative advantage as workers age. The framework provides a structural explanation for why older and younger workers are imperfect substitutes: they are not uniformly 'more' or 'less' productive, but differently matched to tasks. I apply the model to study the general equilibrium effects of 'age-friendly' technology that reduces physical task requirements. While older workers capture a larger share of total income, the reallocation of workers to low-physical tasks accrues disproportionately to the young. This result illustrates how price adjustments and re-sorting can generate distributional consequences that diverge from partial equilibrium intuitions.

How do individuals value policies that reduce the probability of chronic illness? I develop a continuous-time lifecycle model in which agents face three sources of risk: mortality, morbidity, and income shocks. The analysis reveals striking heterogeneity: willingness-to-pay (WTP) for prevention as a share of income is highest for young, asset-poor individuals, reflecting the large present value of future earnings at risk and limited capacity to self-insure. Prevention is most valuable to those with the least ability to pay. I further characterize the interaction between income uncertainty and health investment. Two competing forces emerge: a crowd-out effect, where precautionary savings motives reduce the marginal value of prevention, and an amplification effect, where compounding risks become harder to insure. Which effect dominates depends on the agent's position in the state space.

I document and analyze a newly constructed administrative dataset linking the 2011 and 2021 England and Wales Censuses to HMRC PAYE Real-Time Information (RTI) records and to ONS mortality registrations. The data cover approximately 30 million individuals observed over 3 billion person-month records between 2014 and 2023, with detailed information on annual PAYE earnings, demographic characteristics, and mortality outcomes. I use these data to estimate life expectancy across the income distribution in the United Kingdom.

This paper charts technological progress embodied in capital equipment, and the innovation and diffusion patterns therein. To do this, we digitize archival administrative filings from 1998 to 2024 and extract 50 million capital equipment transactions from five large US States. From these documents, we deploy an 'agentic AI' measurement approach, where multiple AI 'agents' collaborate to build and validate the data. The final dataset contains the make and model millions of pieces of IT equipment, heavy machinery, agricultural tools, vehicles, robotics, CNC machines, and much more. It also contains equipment-level characteristics, including time varying prices.

In this report, we argue for a shift towards preventative healthcare measures to address the economic challenges posed by an aging population and increasing disease burden. Using our novel model that combines health interventions with macroeconomic indicators, we estimate that a 20% reduction in six major disease categories could boost annual GDP by £26.3 billion within ten years and generate significant fiscal savings. We highlight the potential of treatments targeting multiple conditions, such as GLP-1 RA drugs, to unlock even greater economic benefits. Our case study on cardiovascular disease interventions demonstrates that even targeted treatments can yield substantial long-term economic gains. We emphasize the need for swift implementation of prevention programs and suggest starting with cardiovascular disease-focused initiatives due to available cost-effective interventions. We conclude that prioritizing preventative health measures can create a virtuous cycle of improved health, economic growth, and sustainable healthcare budgets.