Working Papers: 

This paper investigates how to recover households' expectations from consumption data when the consumption basket undergoes dramatic changes. I study consumption behavior during the Covid recession and subsequent recovery. Initially, household consumption declined significantly due to Government restrictions and changing behavior. I propose a methodology for computing a measure of robust consumption that identifies a subset of goods and services that is informative about prevailing economic circumstances. Constructing this measure in micro data, reveals new facts about the Covid recession, such as differences in the response of robust consumption by education. I combine this measure of consumption with a sophisticated life-cycle model, featuring three consumption goods, and use it to estimate households expectations during the recession and recovery when subject to a range of shocks. This delivers estimates of changing household perceptions of the persistence of the economic shocks during the course of the Covid recession, and the strength and causes of the subsequent recovery. The analysis highlights the importance of capturing dynamically changing expectations of households during a recession and the contribution of government transfers to the recovery spending boom.

Slides, Latest data 

This paper investigates how the labor market conditions before the Great Recession influenced the magnitude of the consumption decline. It highlights a fall in labor market risk since the 1980s and documents a simultaneous change in households’ asset allocation towards more illiquid assets. I study a model with a job ladder and multiple assets. Reallocation towards illiquid assets and progression up the job ladder are important sources of state dependence. The tranquil labor market prior to the Great Recession, amplified the negative response of consumption by up to 60 percent. Job risk heterogeneity raises the marginal propensity to consume. 

Online Appendix

This paper studies role of credit shocks in household consumption dynamics. It emphasizes the option value of previously agreed credit terms, which fundamentally change the response of the economy to credit shocks. Agents' ability to retain previously agreed credit terms provides a powerful propagation channel by lowering the probability of durable adjustment long after a credit shock hits. Evidence in the micro-data is consistent with this mechanism dominating durable consumption decisions. When credit shocks are concurrent with a large negative productivity shock, the aggregate consumption dynamics provide a better fit to the Great Recession than the standard specification. Household specific credit terms also introduce substantial state dependency into the response of the economy to shocks, as not only the distribution of assets matter but also the distribution of credit terms, which moves over time due to the history of aggregate shocks. Surprisingly, aggregate credit conditions are not amenable to policy intervention. Government intervention to relax credit conditions and reduce market incompleteness results in a decline in consumer welfare.

Work in progress:

The Rise of Women and Decline of Education

with Kolm, A-S, Krusell, P., and Mitman, K. 

This paper investigates how a decline in barriers to market production facing women has affected female participation, occupation decisions, and aggregate human capital formation. Since the 1960s, female labor force participation has increased, the share of college-educated women in the labor force working as teachers has fallen, and average teacher quality has declined. We investigate the implications for the aggregate economy of reallocating highly productive women from teaching and into the market sector. Teaching is unusual because the government fixes the contract most teachers face and does not fully reflect differences in teacher productivity. We derive conditions under which this can distort the first best allocation, even if the planner can choose the contract instruments optimally. We find falling market barriers have raised aggregate welfare but lowered human capital production by 27 percent. As a result, male output in the market sector has fallen. The optimally chosen teacher contract comes close to replicating the planner allocation, whereas a rigid contract combined with falling market barriers implies a substantial welfare decline.

Mortgages, Monetary Policy, and the Great Inflation of 2021

with Hedlund, A., Mitman, K. and Ozkan, S.



with M. Ravn, O. Attanasio, and M. Padula

Econometrica (2022)


US households’ consumption expenditures and car purchases collapsed during the Great Recession and more so than income changes would have predicted. Using CEX data, we show that both the extensive and the intensive car spending margins contracted sharply in the Great Recession. We also document significant cross-cohort differences in the impact of the Great Recession including a stronger reduction in car spending by younger cohorts. We draw inference on the sources of the Great Recession by investigating which shocks can explain household choices in a 60 period life-cycle model with idiosyncratic and aggregate shocks fitted to aggregate and life-cycle moments. We find that the Great Recession was caused by a combination of large aggregate income and wealth shocks, while cross-cohort adjustment patterns imply a role for life-cycle income profile shocks. We also find a role for car loan premia shocks in accounting for car spending and car loans.

Federal Reserve Bank of St. Louis Review, First Quarter 2021