Procrastinators Exhibit Suboptimal Retirement Savings Behavior
According to Jeffrey Brown, of the University of Illinois at Urbana-Champaign, and Alessandro Previtero, of the University of Western Ontario, their research provides “new and robust empirical evidence that procrastinators behave differently than non‐procrastinators for five important retirement‐related financial behaviors.” The paper defines a procrastinator as an individual who waits until the last day of their health care open enrollment period to make their plan election—a real-world metric the researchers suggests is more informative and reliable than qualitative definitions of procrastination.
Using three separate administrative data sets, the researchers show that procrastinators defined this way are (i) less likely to participate in a supplemental savings plan; (ii) take longer to sign up for 401(k) plans; (iii) contribute less; (iv) are more likely to stick with default portfolio allocations, regardless of their appropriateness; and (v) are less likely to take the annuity payout option from their defined benefit (DB) plan when offered a lump-sum payout.
“Further evidence shows that these findings are best explained by procrastination being the outcome of present‐biased preferences, consistent with the predictions of leading economic models of procrastination,” the research report says.
The paper notes that this finding is not new. “The idea is certainly intuitive,” the researchers explain. “Planning for retirement involves near-term actions with distant consequences, and it is easy to put it off when faced with more immediate temptations or demands on one’s time. Indeed, the trade‐off between near‐term costs and distance consequences is the reason that economists treat procrastination as a stemming from present‐biased preferences.”
The researchers say, however, there has been very little empirical research to confirm or refute the idea that procrastination has a meaningful effect on financial security in retirement, “and virtually no empirical work showing that procrastination arises from present‐biased preferences.”
They feel their paper fills this void by providing more robust evidence that procrastinators behave differently than non‐procrastinators when it comes to major actions related to financial preparation for retirement. The paper also discusses why the existence of present‐biased preferences is the only theory that is consistent with the entirety of the team’s empirical results—namely by ruling out alternative stories, including optimal delay, rational inattention, or the basic notion that individuals may simply be busy or disorganized.
“For example, we use data on the frequency of interactions with the plan election tool to separate optimal delayers from procrastinators, and show that our results are being driven by the procrastinators,” the paper explains. “We also do a stylized calculation to show that these are consequential decisions, which casts significant doubt that rational inattention could explain our results. We also discuss that a subset of our results—those related to contribution rates conditional on saving, and those related to annuitization—are consistent with present‐biased preferences and not alternative stories. In the end, the only hypothesis that can explain the entirety of our results is that individuals with present‐biased preferences are more likely to procrastinate, and this combination of preferences and behavior lead them to behave quite differently from non‐procrastinators when it comes to preparing for retirement.”
A key advantage of this research approach is that the consequences of underlying decisionmaking by pre-retirees are real rather than hypothetical. In particular, both health care plan elections and retirement plan choices are financially consequential, which increases confidence in the external validity of the findings.
“Another nice feature is that the health care plan election and retirement plan choices are similar enough for the health care plan election behavior to be informative,” the paper concludes, “but not so similar as to introduce any mechanical correlations with our outcomes of interest.”
The full paper is available for download here.