Association for Public Policy Analysis & Management

Promoting retirement income security is the primary objective of U.S. policies governing employer-provided pensions. To encourage workers and employers to participate legislation gives very large tax advantages for private-pension savings. These effectively represent large “tax expenditures” to the federal government in the form of foregone tax revenues.

To that end, U.S. policymakers have a substantial interest in the results of these large expenditures for retirement income security. Is the private-pension system effectively providing retirement-income security? What are the losses or impediments to achieving retirement-income security among U.S. workers? Which groups of workers are at greatest risk of falling short?

One feature of the U.S. pension system in particular may jeopardize the objective of promoting retirement-income security: the ability of workers to cash out (i.e., withdraw funds from) their private-pension plans upon job separation. Federal rules aim to discourage such pre-retirement cash-outs.  For example, the Tax Reform Act of 1986 introduced a 10 percent tax penalty on withdrawals prior to the age of 59 ½. Burman et al. 2012 showed that this tax penalty reduced pre-retirement cash-out of pension balances and increased rollovers into individual retirement accounts that preserve the tax-advantaged status of the pension balances.  Similarly, they found reductions of cash-outs in response to the 1992 reform that imposed 20 percent tax withholding (without affecting the total tax liability).

These policy changes have reduced, but not eliminated, early withdrawals.  Several studies have investigated what led people to take early pension withdrawals despite incurring the tax penalty. It appears that a significant portion of early withdrawals pertains to households facing liquidity constraints and experiencing financial shocks (Amromin and Smith, 2003; Scherpf, 2010). Still, according to Butrica et al. (2010), about half of early withdrawals from 401(k) defined-contribution pension accounts and Individual Retirement Accounts (IRAs) could not be attributed to the events observed in the data, possibly indicating “unnecessary loss of retirement savings.”

In this project we propose to use the long panel of data collected in the Health and Retirement Study, spanning up to 20 years for the earliest cohorts, to add new insights to prior research findings on this topic.  We investigate trends in pension cash-outs among older workers, assessing cohort differences, and find that although cash-out propensity has increased recently, these cash-outs are associated with more frequent employment or financial shocks, with the unemployment rate and mortgage-related troubles strongly linked to the cash-out decision. Additionally, there is a sizable increase post-2006 in the use of cash-outs to pay down debt.

We then exploit the long panel structure of the HRS to compare 1992 HRS respondents’ outcomes in 2012 as a function of cash-out behavior. Although a correlational approach shows that those who cash out have higher mortality, lower wealth, lower household income, lower pension income, and worse health 20 years later, they are also worse in these metrics in 1992, suggesting a strong selection effect. We pursue a causal estimation strategy and find these differences greatly reduced, and for some measures, reversed.