Individual and employer retirement plans (namely IRAs and 401ks) serve as important supplements to Social Security income during retirement. How widely these plans are used, and by whom, are fundamental facts needed to inform retirement policy. However, these exact numbers remain elusive. Surveys, which use a representative sample of the U.S population, suggest that participation in retirement accounts is much lower for low-income and nonwhite households. However, there are limitations to how much we can learn from survey data. This is due to small sample sizes, potential measurement error, and the inability to follow participants over time.
To provide a more accurate and comprehensive measure of how retirement plan participation varies across geography, income levels, and racial groups, we use administrative tax data. Using the universe of tax returns and information returns, we can observe which households have employer retirement plans. We can also measure how these participation rates relate to a household’s location, income, and employment history.
Relative to survey data, we find larger disparities in retirement account participation across the income distribution. Participation rates also vary greatly with geography, even within commuting zones. This geographic variation seems to be driven by differences in income across locales, rather than racial composition or access to financial services. Finally, using an instrumental variables design, we estimate the impact of an employer offering a retirement plan on whether an employee participates. Access to an employer provided plan has a meaningful impact on participation, particularly for low- to middle-income households.