Working Papers

“Default Options and Retirement Saving Dynamics”. Revise and resubmit, American Economic Review. [This version: June 2021]


I document that employees offset the short-run positive effect of auto-enrollment in retirement plans by saving less in the future. Consequently, a structurally estimated lifecycle model predicts that the long-term effect of auto-enrollment on wealth is negligible except at the bottom of the earnings distribution. The observed inertia at the savings default is explained by an optout cost of around $250. My estimate is smaller than the thousands of dollars estimated in previous studies because non-autoenrolled workers can compensate for not contributing now by contributing more later. Auto-enrollment improves welfare if the policymaker is paternalistic or has strong redistributive preferences.

“What Drives Investors’ Portfolio Choices? Separating Risk Preferences from Frictions” with Tim de Silva (MIT). [This version: April 2022]


We separately identify the role of risk preferences and frictions in portfolio choice. Commonly-used cross-sectional moments, such as the life cycle of participation in the stock market, cannot distinguish between the two: agents may choose not to participate because they prefer safer assets (e.g. due to loss aversion) or frictions impacting their choices (e.g. participation costs). We overcome this identification problem by using variation in the default asset allocation of 401(k) plans and estimate that, absent frictions, 95% of investors would prefer stock market participation with a stock share that declines over the life cycle, which differs markedly from their observed choices. We use this variation to structurally estimate a life cycle portfolio choice model and find the evidence consistent with a relative risk aversion of 3.6. This estimate is significantly lower than most estimates in the life cycle portfolio choice literature and highlights how choice frictions can hamper the identification of risk preferences.

Who Benefits from Retirement Saving Incentives in the U.S.? Evidence on Racial Gaps in Retirement Wealth Accumulation“, with Jorge Colmenares, Cormac O’Dea (Yale), Jonathan Rothbaum (Census), and Lawrence Schmidt (MIT).  [Slides: April 2022] draft available.


In 2020, U.S. private employers and the federal government devoted a combined $300bn to encourage contributions to retirement savings plans. In this paper, we study the distributional impact of these tax and employer match incentives across racial groups using a new linked employer-employee data set covering millions of Americans. On average, White workers contribute 4.6% of their salary to employer-sponsored retirement accounts whereas Black (Hispanic) workers contribute 2.9% (3.3%) of their salary. Differences in income across racial groups explain only one-third of this racial gap in retirement saving, and large disparities remain even after further controlling for education, occupation, county of residence, employer fixed effects, and homeownership. This gap in contributions, amplified by the tax and matching incentives, implies that the average Black (Hispanic) American would retire with around $350,000 ($280,000) less wealth at age 60 than the average White American. We explore two mechanisms driving differences in contribution between employees with similar individual-level characteristics. First, household composition and parental characteristics can explain nearly half of the residual gap in retirement contributions across racial groups. Second, we find evidence that Black and Hispanic individuals face tighter liquidity constraints. Black retirement savers are twice as likely as Whites to take an early withdrawal from their retirement account in any given year, a sign of limited access to alternative means of liquidity. These findings suggest that the institutional design of U.S. retirement plans, which rewards those who can and do save more, exacerbates racial wealth gaps and propagates wealth inequality across generations.

“Efficiency in Household Decision Making: Evidence from the Retirement Savings of US Couples”, with Lucas Goodman (US Treasury) and Cormac O’Dea (Yale). Draft available.


Pareto efficiency is a core assumption of most models of the household. We test this assumption using a new dataset covering the retirement saving contributions of 1.3 million U.S. couples. While a vast literature has failed to reject household efficiency in developed countries, we find evidence of widespread inefficiency in our setting: retirement contributions are not allocated to the account of the spouse with the highest employer match rate. This lack of coordination cannot be explained by inertia, auto-enrollment, or simple heuristics. Instead, we find that indicators of weaker marital commitment correlate with the incidence of inefficient allocations.

“The One Child Policy and Household Saving” with Nicolas Coeurdacier (SciencesPo) & Keyu Jin (LSE). Conditionally accepted,  Journal of the European Economics Association. [This version: December 2021]


We investigate whether the ‘one-child policy’ has contributed to the rise in China’s household saving rate and human capital in recent decades. In a life-cycle model with intergenerational transfers and human capital accumulation, fertility restrictions lower expected old-age support coming from children—inducing parents to raise saving and education investment in their offspring. Quantitatively, the policy can account for at least 30% of the rise in aggregate saving. Using the birth of twins under the policy as an empirical out-of-sample check to the theory, we find that quantitative estimates on saving and education decisions line up well with micro-data.

“How Do Consumers Finance Increased Retirement Savings?” with Christopher Palmer (MIT).

Policy Paper

“The Evolution of U.S. Firms’ Retirement Plan Offerings: Evidence from a New Panel Data Set” with Antoine Arnoud (IMF), Jorge Colmenares (MIT), Cormac O’Dea (Yale) and Aneesha Parvathaneni (Yale). [April 2021]