As defined benefit pensions become increasingly rare, many employers have begun offering 401(k) plans. One of the defining characteristics of 401(k) plans is that individuals choose how much of their salary that they want to contribute; however, individuals are often unsure how much to contribute. Prior research has shown that individuals’ contributions are heavily influenced by the default contribution rate or the maximum level of contributions that the employer will match. While increasing the employer match would, on average, increase 401(k) contributions, this is an expensive and burdensome way of encouraging individuals to save more for their retirement.

The research team made slight modifications to informational emails about the 401(k) plan sent to a subset of employees of one company. The additional content in the emails differed based on how much an individual was projected to contribute to their 401(k) plan that year.

• For individuals projected to contribute between $0 and $3,000, the standard informational email was modified by including one of two additional sentences: either a $3,000 threshold cue: “The next $X of contributions you make between now and December 31 will be matched at a 100%” rate, where X is $3,000 minus the amount that the individual had contributed so far; or a more distant $16,500 threshold cue: “Contributing $Y more between now and December 31 would earn you the maximum possible match,” where Y is $16,500 minus year-to-date contributions.
• The moderate contributors, those projected to contribute between $3,000 and $6,000 in the year, received an informational email with either no modifications, a sentence providing a savings goal example of $7,000, or a sentence providing a savings goal example of $11,000: “For example, suppose you set a goal to contribute $X for the year and you attained it. You would earn $Y more in matching money this year than you’re currently on pace for,” where X is the goal amount and Y is $500 for the $7,000 goal and $2,500 for the $11,000 goal.
• The substantial contributors are those projected to contribute between $6,000 and $16,500 in the year. They received the informational email with either no modifications or the inclusion of the following short paragraph containing an “anchor” (represented by X%) of 3%, 10%, or 20%. “For example, you could increase your contribution rate by X% of your income and get more of the match money for which you’re eligible. (X% is just an example, and shouldn’t be interpreted as advice on what the contribution increase is for you.)”

Individuals were randomly assigned to receive the standard informational email or one of the modified informational emails based on their projected contribution level. The researchers used data from Vanguard to track how employee 401(k) contributions as a portion of their salary changed over the 11 months after they received the informational email.

The researchers used a randomized control trial to test the effects of different types of messaging on employee 401(k) contributions. They found that among the employees who contributed the least to their 401(k) plans, those who received the $16,500 cue contributed higher proportions of income on average than those who received the $3,000 cue. The difference between these groups slowly grew for 3 months after the email was sent, to a peak contribution rate difference of 1.5% of income, on average, for those who received the $16,500 cue compared to those who received the $3,000 cue. This difference, however, faded substantially after the first 3 months and was effectively gone 9 months after the emails were sent.

The moderate contributors who were encouraged to set an $11,000 savings goal increased contribution rates by an average of 1.1% of income over the first five months compared to those who received the standard email, but this effect diminished after the 5^th month. The employees who were encouraged to set a less ambitious savings target of $7,000 did not contribute at rates significantly different from those who received the standard email. Thus, only the more ambitious goal had a strong, positive effect on contributions to one’s retirement plan.

Among those who contribute the most to their retirement plans, providing anchors of 3%, 10%, or 20% all resulted in significantly higher average contributions, although there were no differences between the anchors, and the increased contributions came at a significant delay. Between 5 and 11 months after receiving the email, the individuals who received an email with one of the anchors increased their contribution rates by 1.0-1.1% of income more, relative to the individuals who received the standard email.

Inspired to implement this design in your own work? Here are some things to think about before you get started:

• Are the behavioral drivers to the problem you are trying to solve similar to the ones described in the challenge section of this project?
• Is it feasible to adapt the design to address your problem?
• Could there be structural barriers at play that might keep the design from having the desired effect?
• Finally, we encourage you to make sure you monitor, test and take steps to iterate on designs often when either adapting them to a new context or scaling up to make sure they’re effective.

Additionally, consider the following insights from the design’s researchers:

• Individuals often end up choosing the default because it is the only reference point that they are provided with; as a result, providing other reference points can work to move them away from the default.
• In a previous intervention, researchers found that low anchors, on average, decreased 401(k) contributions, and they found that cues were most effective among individuals where the cued threshold was furthest away from their current contribution rate. This study sought to answer the questions left by the previous study by testing a wider range of interventions on each subset of employee contributors.
• The researchers tailored the different modifications to be most relevant for each contribution level, applying different treatments based on how much individuals were expected to contribute that year. Individuals already on track to contribute $16,500 (the maximum allowed within a calendar year) were excluded from the study.
• The employees of this company are generally highly educated, very well paid, and technology savvy. Consequently, the interventions that were most effective for employees of this company might not be as effective for employees at other companies.
• It is important to consider one-time events that might occur during the period of interest and how they might affect variables of interest. In this case, the researchers sent the emails in October and they had varying effects over the next 6-12 months; however, bonuses were paid in March and caused individuals to make substantive changes to their contribution independent of which email they received.