On average, households aged 60 to 69 with personal retirement accounts withdraw only about 2 percent of their account balances each year.
Over the past thirty years, retirement savings for private sector workers have undergone a dramatic shift, from employer-provided defined benefit pension plans typically paid out in the form of lifetime annuities to personal retirement accounts (PRAs) in defined contribution pension plans, such as 401(k)s. In 2008, private sector PRA assets totaled $7.1 trillion while assets in traditional private sector defined benefit programs totaled $2 trillion. How households draw down the balances that they accumulate in these retirement saving accounts can have an important effect on their retirement income security.
At the time of retirement, the PRA participant typically has sole control of the accumulated assets and can decide when to withdraw them. In The Drawdown of Personal Retirement Assets (NBER Working Paper No. 16675), authors James Poterba, Steven Venti, and David Wise present new evidence on how PRA assets are drawn down, focusing in particular on patterns in the early years of retirement. Analyzing data from the Survey of Income and Program Participation and the Health and Retirement Study, the authors find a relatively modest rate of withdrawals prior to the age 70 1/2, when households are required to take minimum required distributions. Only 7 percent of PRA-owning households between the ages of 60 and 69 take annual distributions of more than 10 percent of their PRA balance, and only 17 percent of that group make any withdrawals in a typical year.
The rate of distributions rises sharply after age 70 1/2: the proportion of PRA-owning households making a withdrawal jumps to over 60 percent by age 71, and crosses 70 percent a few years later. The sharp increase in withdrawals when distributions become mandatory suggests that many households in their early 70s would not make withdrawals if it were not for the distribution rules.
The low rate of withdrawals from PRAs during the sample period in this study, 1997-2005, along with investment returns to PRA assets and contributions to PRAs by some individuals who were still employed, generated a pattern of increasing average PRA balances by age. Rather than declining in value after households retire and begin to finance retirement consumption, PRA balances continue to grow through at least age 85 in many cases, although the rate of growth is slower at older ages than at younger ages. On average, households aged 60 to 69 with PRA accounts withdraw only about 2 percent of their account balances each year, considerably less than their rate of return during the period under study. Even after the required minimum distribution age, the percentage of balances withdrawn remains at about 5 percent which, for most years, was below the average return on PRA assets.
While average withdrawal rates are low, there is substantial variation across households, and some of them withdraw a significant proportion of PRA assets. Among households headed by someone between the ages of 60 and 69, roughly 10 percent of PRA owners make an annual withdrawal of 5 percent or more of their PRA assets. Those with higher balances are more likely to make a withdrawal than those with lower balances. Among those who make a withdrawal, the PRA balance is the most important determinant of the proportion of assets withdrawn.
At ages 72 and older in contrast, after required distributions begin, 41 percent of households withdraw more than 5 percent of their PRA balance in a typical year, 23 percent withdraw more than 10 percent of balance, and 11 percent withdraw more than 20 percent of their balance.
There are also substantial differences in PRA balances across households. The authors estimate that while only 8 percent of households in the lowest decile of non-PRA wealth, income, and health status have a PRA as they approach retirement, about 80 percent of households in the top decile of non-PRA wealth, income, and health status have such accounts. Households in poor health are less likely than those in good health to have a PRA.