Saved by Design

Saurabh Bhargava, Lynn Conell-Price, Richard Mason and Shlomo Benartzi
Working Paper
August 23, 2018

Despite the success of automatic enrollment (AE) in raising 401(k) participation, many US employees save inadequately for retirement. We assert that an employee’s initial decision to confirm enrollment at the default rate or personalize enrollment at an adjusted rate may strongly influence their prospects for "retirement security" and that non-economic variation in the design of the digital enrollment interfaces from which employees increasingly enroll may strongly shape this consequential decision. While standard economic theory would deem variation in non-economic features unimportant, recent work in Behavioral Economics challenges this presumption. We conduct three field studies, in collaboration with a major 401(k) provider, to test how variation in the "psychological design" of an otherwise standardized enrollment interface — realized through small changes to the presentation of options and the proximal display of plan information — influences several thousand enrollment decisions of employees across over 500 plans. We supplement the field studies with a lab study involving thousands of hypothetical enrollment decisions and a survey of plan administrators.
Our primary contribution is to document the large influence of psychological design on enrollment. The “enhanced design” we tested increased the share of personalized enrollment by 0.09 (0.63 baseline), the share of full match take-up by 0.10 (0.59 baseline), and the average contribution rate by 63 basis points (5.69 percent baseline). The latter effect is equivalent to that predicted from increasing the typical plan's match limit by 67%, or 4 percentage points, and, for the marginal employee personalizing enrollment, translates to a significant increase in expected retirement security. Second, we find that enhanced design not only increases average contributions but appears to amplify employee sensitivity to underlying match generosity, offering evidence for a novel complementarity between a plan’s psychological and economic design. Third, we find little evidence from the lab that the observed patterns can be explained by a standard economic model of enrollment, even allowing for information frictions or institutional distrust. Collectively, these findings imply that part of the substantial variation in retirement preparedness among 401(k) enrollees may be attributable to (previously unexamined) features of plan design that fall outside the current fiduciary and regulatory purview and point to the need for a more psychologically informed understanding of how employees save.

Read More