Rethinking Retirement: Saving Insights | Katy Milkman

How Do Retirement Funds Work Today?

Dan Loney: The decision to head into retirement is not always an easy one. People consider a variety of factors, including whether they will have enough savings, especially since people are living longer right now. This is an area of research by our guest here today, Katy Milkman, who’s a professor of operations, information and decisions here at the Wharton School.

Katy, great to talk to you again, as always. From the research you’ve done and people you’ve talked with, is it a harder process to make some of these retirement decisions now than maybe it was a few decades ago?

Katy Milkman: Things have changed a lot. We now live in a society in the U.S. where it’s common to work for an employer who offers you what’s called a defined contribution plan instead of a defined benefit plan. It used to be that you would be guaranteed some amount of income for the rest of your life if you worked for an employer. That’s a defined benefit, even after retirement.

There’s been a major shift over the last 30 or 40 years towards giving people the opportunity to contribute a defined amount and maybe have an employer match to a retirement fund. But then the income they’ll have in retirement is a function of how much they choose to save, whether they potentially dip into those funds prior to retiring. There’s no guarantee because it depends on the performance of those assets. So, it’s a really different world, and it has led to not great results, honestly, in terms of the retirement income for people who have who’ve worked their whole careers. A lot of people are working longer or living less comfortably than they would have liked because of this new era.

I should say I’m not an expert on either defined contribution or defined benefit plans, but that’s my rough understanding of the change. The work I do with my collaborators really focuses on, in this current era where defined contribution plans have become so common, it’s a big challenge to convince people to save. We have to make sure people are saving an adequate amount. And we face a lot of biases when we’re trying to make that pitch. One of them is present bias. The fact that we tend to be more attuned to the instant gratification we get from, say, spending a paycheck now, rather than saving it for later. We focus more on the here and now and undervalue future everything. Whatever money I’ll have in the future, it’s worth less to me. “That’s forever away,” and I don’t value it as much. That’s a major bias. It’s something I teach my MBA students about. Given that we’re fighting that uphill battle when we’re trying to convince people to save for, essentially, future me, we try to use a lot of psychological insights to propel people to save what they’ll need.

The ‘Fresh Start’ Phenomenon

Loney: You research this through the idea of a fresh start. Give us an idea of what it is.

Milkman: This is a topic I’ve studied with my former Wharton PhD student, Hengchen Dai of UCLA. Hengchen and I look at this idea that not only at the beginning of a new year, but there are lots of moments in our lives that we feel motivated by the sense that we have a clean slate or a fresh start. We’re most familiar with New Year’s resolutions as part of the fresh start effect. At the start of every new year, we think, “Oh, it’s a clean slate. It’s a new year, it’s a new me. I can achieve more.” And we set these resolutions.

But we have documented that this phenomenon of feeling like we have a fresh start and are more motivated to make change arises at lots of other moments, too. At every birthday, on Mondays, at the start of a new month, following the celebrations of any major event that feels like the start of a new cycle. All sorts of new beginnings give us that fresh start feeling, which increases our motivation to pursue our goals and makes us feel more disconnected from past failures. Because we can say, “That was the old me, and this is the new me.” A lot the psychology in this space relates to how I relate to my past self and my future self.

Thinking about that carefully led us to want to use the fresh start effect, or this motivation, to try to propel people to save more for retirement. To think more about future me. We thought fresh start moments would be an ideal time, because people have that extra motivation to pursue goals and to think about the future and to feel disconnected from past failings.

Loney: Were employees receptive to looking at this kind of moment in their life as a fresh start and a great way to maybe head down a different path?

Milkman: We ran an experiment with thousands of employees who were not yet saving for retirement, or who were saving but at a very low rate well below what they would need to have a comfortable retirement. We partnered with four different organizations and sent mailings to employees who were in these categories: non-savers or a small number of very low savers. And we tried to use this insight about fresh starts to increase the likelihood that people would save.

We sent mailings that invited people to start saving right away. But we know that people like to procrastinate on anything that sounds difficult, like starting towards a savings goal, so we also invited them an opportunity to delay. We said, “If you don’t want to save now, you could start saving on this future date.” And what we randomized in our experiment was, for some people the future date was a fresh start date, and for some people it wasn’t. For instance, if you had a birthday coming up in two months, we might flip a coin and decide, are you going to be invited to start saving now or after an upcoming birthday, or after your next birthday? That would be what your mailing might say, in one condition. Or — and this is what’s tightly controlled about it — we would say, would you like to start saving now or in two months?

In both cases, it’s exactly the same offer. But in one case, we’ve tied it to your birthday, which makes it clear that this is an opportunity you might want to align with that fresh start moment. We tried this with several fresh start dates. We tried aligning New Year’s. We invited you to start saving after the start of spring, after a birthday. You just get one of these offers. It was random assignment. But we tested all of those different fresh start opportunities. And we tested it against just inviting people to save at an equivalent time delay. We also tried some dates that don’t feel so much like fresh starts — we think of them as placebos. Valentine’s Day. It’s also a day that is sort of notable on the calendar that we can label, but nobody typically thinks of Valentine’s Day as a fresh start moment. If you met the love of your life then, maybe for you, it’s meaningful. But for most people, that’s not a day that feels like the beginning of a new cycle in life.

What we found is that when we invited people to begin saving after a fresh start date, and we compare what happened to people’s savings rates who were invited to save now at an equivalent time but without that fresh start date call out, we see significantly more savings over the next eight to nine months. We see, depending on how you model it, a 20% to 30% increase in the next eight months of savings among the population that is invited to start saving after a fresh start date. So, this does seem to motivate more people. It doesn’t lead them to decline saving now, and it increases the total number of people who are saving, leading to this higher savings rate over the subsequent months.

Loney: I think it’s also interesting because you mentioned in the paper that we tend to make decisions that don’t help us in the long term. By doing this, we’re also changing behaviors.

Milkman: That’s right. We are changing behavior by getting people to recognize this opportunity is one that’s aligned with their goals and seeing, “Yeah. Actually, I do want to start saving following my birthday. That sounds exactly like the right moment to do this.” Or, “Yeah, at the start of spring really does feel like a moment when I should be upping my savings contributions.” But if I’d asked you, “Do you want to start saving next month?” which happens to be the start of spring, but I haven’t called your attention to it, you don’t have that same resonance, and you don’t make the same decision. So, it’s changing behaviors. It’s changing long-term outcomes for people by increasing their savings rates. We think that’s a really important insight. And the more we can leverage these kinds of moments that people see as fresh starts to increase savings or any other goal-directed behavior, the better.

How Peers Influence Retirement Savings

Loney: You’ve also done research that looks at peer information in retirement savings decisions. Tell us about that.

Milkman: This was another randomized controlled trial, also inviting people who were either non-savers or low savers to save. I should also say there’s a common co-author on both of these. John Beshears of Harvard Business School is a fantastic researcher, does a lot of work on retirement savings, and was involved in both of these projects.

In this case, we partnered with one big company that had a lot of employees who had been part of a union. As a result of their union bargaining, they had not been automatically enrolled in the savings program, the contribution program that they had on offer. It’s a 401(k) plan, where you put a portion of every paycheck into this plan, and you get a tax benefit when you do so. That money is not taxed. It’s put in before taxes.

Lots of people were not saving, because if you’re not automatically enrolled, they had to take steps to start saving, and a lot of people don’t bother to do that. We tested this both with union employees whose union had not negotiated for them to be automatically enrolled, as well as a non-union population. But the non-union non-savers were different because they had intentionally opted out of saving. They’re sort of selection-bias, if you will. That’s our nerd term for saying there’s slightly different populations.

Loney: That effect that you saw play out, was it similar for pretty much everyone across the company?

Milkman: It wasn’t. It actually turned out to matter which population they were in. What we were testing was whether or not telling them about how many of their peers were already saving or were already saving at a higher rate, whether or not that might increase their savings likelihood. We tried two things. One, we varied whether or not they got a mailing that told them about the high number of their peers who are already enrolled and encouraged them to follow suit. The other thing we did is that we varied what number they saw. We were never deceptive, but we randomly assigned people to either find out about the savings rate of peers in their five-year age cohort. So if you’re age 40 to 45 you might find out about others age 40 to 45. Or in their 10-year age cohort. If they were age 44, they could also see the 40 to 50 year age bucket.

It meant we had some variation in the number people saw. We could test not only what’s the impact of finding out how many of your peers are saving, but also what is the impact on finding out a slightly different number when you learn that your peers are saving. When you’re in the 40 to 45 age bucket, you might find out 75% of other 40- to 45-year-olds are saving. But if you saw the 40 to 50 age bucket, you might find out that 80% of folks in that group are saving. We have these two ways we can look at what’s the impact of peer influence. One, finding out that lots of peers are saving. And two, what’s the actual number you see, and how does that matter?

What was really surprising, given what we know about how influenced we are by our peers, is that in this study we found a backfire effect, specifically among union employees. This was the group we expected to be most malleable because they hadn’t previously made an active decision about retirement. They just passively were not signed up. This group, when they saw a peer comparison and learned “75% of your peers are saving,” that reduced the likelihood that they chose to save. In addition, when we look at the specific number they saw, we see the higher the number they observe, the less likely they are to save.

These are two things that go against what we normally expect to see in terms of peer effects. Because one, when I find out everybody else is doing something and the majority of other people are doing it, I normally decide to do it too at a higher rate. And two, the higher the number of my peers doing something, the more likely I am to want to join. But we see the opposite in both cases.

This was really puzzling. We don’t see this among people who weren’t members of the union. So, they hadn’t actively opted out. There, we don’t see any effect. But we wanted to dig into this backfire effect. And I want to say, first of all, I still feel that we don’t know for sure what happened. But our best explanation at this point, based on additional analyses we ran, is that it seems to be sort of an upward social comparison reaction, where people are feeling like they could never possibly catch up, because it’s driven by lower income members of the population.

When we do a median split on earning, and look at people who are below median earners in all the different states around the country where this company has employees who received our mailings, we see that the effect is really driven by the lower income folks. That leads us to conclude, potentially, this is driven by that sense that I can never catch up. The idea of social norms is, “I want to keep up with the Joneses, so I’m going to try to do what the Joneses are doing.” But if you feel you can’t possibly keep up with the Joneses because their income is so much higher and they’re already way ahead of you on so many dimensions, it may just be demotivating to hear, “Yeah, they bought another luxury car, and they saved more for retirement,” and so on. It may make you feel that it’s hopeless. That’s our guess as to what happened in this particular study. It was very disappointing, but also useful to know.

Loney: Is there a next logical step that you would like to take this research?

Milkman: Well, the answer to everything right now is AI. A lot of the work that my collaborators and I are talking about in the space of behavior change is related to, how can we use these large language models to incorporate some of the best behavioral insights we know into dialogues to help people make better decisions? That’s the next natural step. Certainly, the LLMs that we train to try to help support people’s retirement savings ambitions will be armed with the knowledge from this research — that it may not be as effective as we thought to use social norming, particularly on low-income consumers, when we’re trying to encourage them to save more for retirement. And that it can be effective to leverage fresh start dates as moments when people feel it’s appropriate and optimal to begin saving.