Using Large-Scale Administrative Tax Data to Assess Housing and Retirement Policies

Abstract

In this dissertation I use administrative tax data to evaluate two policies in the fields of low income housing and retirement. In particular, I construct novel data sets to assess whether growing up in housing constructed under the Low Income Housing Tax Credit leads to higher educational attainment and earnings later in life, and to measure the effect of automatic enrollment in firm-sponsored retirement plans on individual and household savings. On the topic of housing, I find that each additional year spent in LIHTC housing as a child is associated with an average 3.9 percent increase in the likelihood of attending a higher education program for four years or more, and a 5.2 percent increase in future earnings. I conclude that the reason I see a positive LIHTC effect is likely that the housing subsidy provides families with a more stable living situation and with more disposable income. On the topic of retirement, I estimate an increase in plan participation of 32.4 percentage points due to automatic enrollment, and a more modest increase in average saving of 0.9 percentage points, with a negative nudge towards lower default rates for employees who opted in prior to the policy change. However, I also find that automatic enrollment leads to a 36.1 percent increase in withdrawals taken from other retirement accounts, and that this completely offsets the increase in contributions from firm-sponsored retirement plans for employees in the bottom wage quintile, although the net effects remain positive for employees with higher wages.