Jonathan Falk sent me an email to David Tabak and me with the above subject line and a pointer to this article, Smaller than We Thought? The Effect of Automatic Savings Policies, which begins:
Medium- and long-run dynamics undermine the effect of automatic enrollment and default savings-rate auto-escalation on retirement savings. Our analysis of nine 401(k) plans incorporates the facts that employees frequently leave firms (often before matching contributions from their employer have fully vested), a large percentage of 401(k) balances are withdrawn upon employment separation, and many employees opt out of auto-escalation. Steady-state saving rates increase by 0.6% of income due to automatic enrollment and 0.3% of income due to default auto-escalation. Only 40% of those with an auto-escalation default escalate on their first escalation date, and more opt out later.
I replied: Interesting. The remaining effect is still impressive, though.
Falk responded:
The effect is still there, but to argue that there was never going to be an effect of defaults is a strawman…. after all, there are idiots out there. (That’s a characterization I once heard about the insights of the behavioral economics literature.) Missing the effect by a factor of 4 seems pretty important to me, and I guarantee it won’t be reflected in textbooks and marketing materials any time soon.
In response to that last statement, Tabak pointed to this press release, Do Nudges Help Americans Save for Retirement? Not as Much as We Thought.
The only thing is that this press release was from the institution of one of the authors of this new study. What I’d really like to see is a “full Sunstein” on this: magazine articles, newspaper op-eds, media interviews, a couple of books, some promotion on social media, a description of the experiment as a “masterpiece,” a claimed discovery of a new continent . . . You know, the usual thing we see when hyping some shaky claim coming from the Nudgeoverse.
Still, 1/4 of the claimed effect isn’t zero. It’s the Edlin factor.
“a large percentage of 401(k) balances are withdrawn upon employment separation”
this is really misleading. I would presume that the 401(k) is rolled over into the person’s own retirement account (IRA). This is certainly the case when I retired from my job.
I had the same thought but have no idea if the rollover part is true. The first link in Andrew’s post is broken so I couldn’t check that, but the press release link says this: “About 4% of U.S. workers leave their job every month, and in the authors’ sample, employees who leave their job cash out about 40% of their 401(k) balances”.
From the paper: “Alight (2019) reports that in their universe of more than 2 million participants, 40% of
departing workers cash out their entire 401(k) balance.”
A substantial part of the paper discusses the estimates of cash withdrawals not rolled over. They can distinguish between rollovers and cash withdrawals, but since year-end balances are the data they have, they restrict the direct estimate of cash withdrawals to separations in July, August and September. Later leavers may not have cashed out yet and earlier leavers have overlap between investment returns and their cash withdrawals. But to answer Alan query above, people do not routinely roll over balances into personal IRAs, unlike people who know what they’re doing.
Pages 14-17 discuss this. On table 2 I notice that most of the cashed-out 401(k) retirement funds are small, less than 20% of a year’s salary for an living-wage job in the North Atlantic. I wonder if this is really saying that people who leave their employer shortly after they are hired need cash in a hurry? Or don’t know the most efficient way to move the funds to their main retirement account? Most people don’t know the “best” way to do a banking procedure.