Abstract
Household financial outcomes are affected both by neoclassical preferences and constraints and by systems of institutional and behavioral frictions that impede full active choice. This dissertation studies how policy design and workplace financial arrangements affect saving, withdrawals, and compensation in environments shaped by default rules, non-participation, liquidity needs, and limited attention. Across three essays, I examine how these frictions influence retirement saving, working-life liquidity, and the distribution of financial gains across workers and firms. Together, the chapters show that institutional design can have persistent effects on financial behavior, provide valuable liquidity during adverse shocks, and distribute gains unevenly across groups.
In chapter 1, I study optimal default design in retirement saving when passivity extends beyond the default option itself. Using administrative tax records from early-adopting U.S. states, I study optimal default design in state auto-IRA programs, finding persistent increases in retirement saving, with participants retaining their balances even after job separation. However, higher default rates cause many participants to exit default saving and choose a zero saving rate, even when zero saving is an interior choice. Existing models of default design assume a single passive choice—the default option—implying divergent optimal policies depending on whether default effects reflect real adjustment costs or behavioral biases; in the latter case, these models can favor punishment defaults that induce opt-out. To explain the empirical patterns, I extend existing models by allowing passivity to apply to two competing passive choices, default saving and non-saving. Exploiting variation in auto-IRA default rates, I structurally estimate the model and find that the optimal default rate remains between 2.8% and 3.7% whether frictions reflect real costs or behavioral biases. This stability arises because changes in the default partially reallocate individuals across passive options rather than inducing large shifts toward active choice. Once passivity is not limited to the default, the case for punishment defaults weakens, and the results imply a narrow range of moderate default rates even when default effects reflect behavioral biases. In this case, the default acts as a second-best policy that mitigates other distortions to saving behavior.
In chapter 2, I examine the extent to which retirement wealth accumulated through automatic enrollment provides liquidity during the working life. Standard models of saving in retirement accounts trade off working-life liquidity against resource accumulation for retirement. This chapter shows that, for passive savers, automatic retirement saving can expand working-life access to liquidity. I study state auto-IRA programs, which generate quasi-experimental variation in retirement account wealth through automatic enrollment. I find that each additional dollar of induced IRA wealth raises withdrawals by 36 cents in years with large earnings declines, with no comparable response in normal earnings years. This state-contingent pattern indicates that induced retirement balances are used for consumption smoothing during adverse earnings states rather than primarily reflecting regret-driven leakage. I develop and estimate a welfare framework with earnings risk, incomplete precautionary saving, limited displacement of other saving, and withdrawal frictions and calculate that, net of offsets, the average working-life liquidity value of each dollar of induced IRA saving is $0.16. These findings inform the conventional view of retirement account design. When individuals face frictions to active financial adjustment or do not fully build liquid precautionary wealth, automatic retirement saving can provide both retirement resources and working-life liquidity, while those same frictions partially substitute for formal commitment generated by account illiquidity.
In chapter 3, I study the incidence of Employee Stock Ownership Plan (ESOP) adoption across workers and firms. Employee Stock Ownership Plans (ESOPs) are tax-advantaged retirement plans through which firms compensate employees with company stock and transfer ownership to workers. Using Department of Labor Form 5500 filings matched to tax return data, I construct a panel of firms that adopted ESOPs in 2009 or 2010 and compare them to similar non-adopting firms through 2024. I find that ESOP adoption increases wages for incumbent employees but decreases wages for non-incumbent employees, while increasing retirement wealth for all employees. A compensation-accounting exercise that combines wage and retirement effects implies that incumbent employees gain from ESOP adoption, while non-incumbent employees do not. For firms, ESOP adoption reduces tax payments, increases employment, and is associated with higher valuations, with little change in profits per worker. These findings imply that the gains from ESOP adoption are distributed unevenly across worker cohorts. The results for non-incumbent employees are consistent with workers valuing the non-pecuniary benefits of ESOP participation.
