Are we good at making choices?

James Choi: Over the last ten years I've been doing a lot of work on how people make choices in retirement savings plans, and whether they appear to be optimizing in the way that we think that rational economic agents would. One of the main decisions we make in our financial lives is how much to save. And one of the best ways to save for retirement is a 401(k) plan because it's tax-advantaged and typically there's a matching contribution from the employer.

Because this is a high-stakes decision that has major implications for your material well-being in retirement, economists have traditionally thought that people would put a lot of thought into this decision and make it in a sensible way. One feature of sensible decision making is that the outcome doesn't depend on characteristics of the situation that don't affect its fundamental economics. One such characteristic is what happens when somebody doesn't act. In traditional economic theory, not acting is, in and of itself, a decision that reveals something about your preferences. So if I'm passive and as a result don't get enrolled in my 401(k) plan, then traditional economic theory would say that I've revealed that I don't want to be in the plan. If you change the plan so that I'm automatically enrolled unless I opt out, traditional economic theory would say that the outcome should be exactly the same: I would act so that I still end up not enrolled in the plan.

In fact, what we've observed is that when the default is being enrolled in the plan, many more people — the overwhelming majority — end up enrolling in the plan than if the default is not enrolling. Most automatically enrolled people come into the plan at the default, which might be something like a 6% contribution rate invested in a lifecycle fund. Four years after the initial automatic enrollment in the plan, about half the people are still at these defaults, which they would not have chosen if they had to actively opt into the plan.

A major reason why people get pushed around like this is that they procrastinate. Quite frankly, the cost of delaying action in your retirement plan just one more day is not that large, and there are all sorts of things that are pressing today that you'd rather be tending to — the kid is screaming, dinner's not ready, the game is on TV — and thinking about your retirement savings plan is something that can wait until tomorrow. Of course, when tomorrow rolls around, other issues are pressing and you again choose to delay. Before you know it, a couple of years have gone by and you haven't changed from the status quo in your retirement plan. That's why the default is so powerful.

Keith Chen: Do you think that it's exclusively hyperbolic discounting that's driving this phenomenon?

Choi: I think hyperbolic discounting plays a big role. Let me describe what hyperbolic discounting is. Generally, when economists have modeled inter-temporal choice — that is, choices between two different times — we've treated your discount rate as not systematically depending on how imminent the earlier time period is. But if you're a hyperbolic discounter, then the present gets over-weighted relative to the future.

Say the choice is whether to bear a little pain by exercising on Tuesday for the sake of better health on Wednesday. On Monday, you feel that this tradeoff between Tuesday and Wednesday is a good deal, because Wednesday's gain doesn't get discounted so much relative to Tuesday's pain. So you're happy to plan on Monday to exercise on Tuesday. But once you actually arrive at Tuesday, you hugely discount Wednesday's gain relative to Tuesday's pain, and you decide not to exercise on Tuesday. You see these choice reversals: when you're choosing between two future periods, you're relatively willing to delay gratification, but when you're choosing between pleasure now versus pleasure in the future, you're very impatient. This can generate a kind of procrastination, where you delay sacrifice today, thinking that you'll be more patient tomorrow and will get to it then. But in fact when tomorrow rolls around, again, you become impatient and want to push the sacrifice off one more day.

Of course, there are other things that also give the default power. For example, we know that there's something called the endorsement effect, where people believe that a default that the employer has set carries implicit advice about what the right choice is.

Judy Chevalier: Especially if they have what you might think of as high cognitive costs for dealing with this particular problem.

Choi: Exactly. There's some evidence now that people have much better ideas about what a good savings rate is than what a good asset allocation is. You actually see in the data that savings rate defaults are less sticky than asset allocation defaults. When you ask people, "What do you think the right savings rate for you is?" they give answers that sound pretty sensible if you compare them to what professional financial advisors would recommend, usually something like 12, 13, 14% of income. So I don't think the primary problem for people is not knowing how much to save; it's just that they can't get themselves to save that much.

But asset allocation, I think, is completely baffling to most people. This is something that people have to pick up on their own. Most people have no formal education in asset allocation. And so they're looking for anything in the environment that's going to give them clues, and the default is one such clue.

Now the problem is that, historically, these defaults were not chosen because the employer necessarily thought that they were the best choice for people. They were chosen because employers did not want to get sued. There was a concern that if employers chose to include stocks in the default 401(k) asset allocation and then the stock market fell, employers would face litigation risk. And so the defaults became very conservative — very low savings rates, typically invested in a money-market fund or stable-value fund, which most experts think is not a good idea. But the Pension Protection Act of 2006 created a safe harbor where employers could create asset allocation defaults and contribution rate defaults that are more aggressive and not worry about getting sued by their employees.

Chevalier: There's an interesting question about which kinds of decisions people are good at making, and which they're not. I have a paper that I wrote with Austan Goolsbee, looking at the college textbook market. You might think, these are college students, they're not thinking rationally about these decisions. And I've certainly seen lots of things written about college textbooks that make you think that not everybody thinks about the college textbooks market the right way. But what we find in the paper is that students actually get it right.

One thing we know about college textbooks is that they tend to be revised every few years. When a book is not revised, the student can buy it at the beginning of the semester and sell it back at the end of the semester, usually for half of the purchase price. So, a $100 book will have cost the student $50 in total for the semester if it's sold back at the end. But when a book is revised, the book can't be sold back. A $100 book costs $100 to use for the semester. In our paper, we look at books in economics, psychology, and biology, and we look at student purchasing behavior. And we show that the students' elasticity of demand looks about the same when you compare changes in book prices and changes in the probability that the students are going to be able to sell back the used book. So if the book is a new edition, the probability that a student is going to be able to sell back the book is really high, and the students appear in the data to be willing to pay plenty for it, because they are almost certainly going to be able to sell it back for half price to the college bookstore. But when the book gets close to the end of its expected life, when there is a high risk that they're not going to be able to sell it back, then the same $100 looks very expensive, and students are less likely to buy it.

In general, we find that the students do a pretty good job planning for the future in that market. So what is the set of things about which people do a pretty good job? They manage to avoid buying books that they're not going to be able to sell back; why do they do that well, if they don't invest well in their 401(k) plan?

Training is probably part of the answer. It's not training by us as their professors, but there's a sort of received wisdom that is more or less right. In the example about asset allocation that James gave, it's not that crazy to think that your employer would set the default in a way that is more or less appropriate for the average employee. It turns out that for various institutional reasons that's a bad assumption. But I think that textbooks is an area where there's this intense cultural training where students pass on this wisdom to each other.

Chen: Maybe another difference between your setting and James's setting is that in James's setting, what we've seen, if anything, is that the short run feedback may actually push people to make the wrong decision, like when they pay too much attention to short-run fluctuations in the stock market and they push their allocation further toward bonds. And the ultimate feed-back that would push them toward the right decision is far into the future — it's hypothetical and requires some kind of mental simulation that they can't really do. Whereas the short-run feedback in your setting actually pushes people toward the right decision.

Chevalier: Yes, if they blindly purchase their books in their first semester in college and they get hammered on the one that they can't sell back, they won't make that mistake again.

Choi: I think the feedback is very important. There is a big difference in the quality of the feedback. With savings, it is not at all clear whether you've made the right asset allocation, ex ante. It's very easy, ex post, to play Monday morning quarterback and say, oh, I shouldn't have been in the stock market in 2008. That's not a very useful form of feedback when you're trying to make asset allocations for the next 30 years of your life.

Chen: There is some new and very exciting work in the animal literature in
this area. For example, it's not a paper of mine, but there's a new bird paper out where pigeons are pecking levers. These levers are going to pay off in food, and they are going to pay off in different amounts of food in different amounts of time. So if the pigeon pecks the left lever, it gets one cup of bird seed in one min-ute. If it pecks the right lever, it gets three cups of bird seed in, say, two minutes. But the levers are mutually exclusive, so once you peck a lever, both levers disappear.

Now what's interesting is that about two-thirds of the pigeons peck the left lever when they're presented with the levers. So the pigeons appear to have an unusually high discount rate, which is consistent with the behavior that James is talking about…

Chevalier: The other third of pigeons are the ones that take over the world.

Chen: Right, the other third of the pigeons actually go to the Ivy League schools. And all of the other pigeons are working for them.

But what's interesting in this new study is: Suppose you take a pigeon — and you already trained it in the other settings so it would know that this is what the levers do — suppose you present the pigeons another two levers. And what these levers do is prevent that original left lever and the original right lever from appearing to you in the future. So you have Left One and Right One. If you peck Left One, Left Two never shows up. And if you peck Right One, Right Two never shows up. So now the pigeon can decide to make sure that he doesn't see one of these levers in the future.

And what it turns out is that very quickly, almost every pigeon learns to peck Left One, in some sense to prevent themselves from having the temptation in the future of getting the immediate but poor payoff. Now, whether or not the pigeons are actively thinking of this as a self-control device or something like that is unclear. I think it's suggestive, but there could be just some very simple reinforcement story underlying this.

Choi: There's an interesting tension between this work and what we see in the market. The pigeons seem to have a strong demand for self-commitment. And yet when you look at the marketplace among humans, we see surprisingly little demand for self-commitment. Suppose you're trying to lose five pounds, and I offer you a self-commitment contract: you post a $1,000 bond to me, and if you don't lose your five pounds, then I'll keep your $1,000. If you lose the five pounds, then I'll return the money to you. People are generally very reluctant to write those kinds of contracts, even though they would like to lose those five pounds.

Of course, self-commitment devices exist in the world. For example, personal trainers are arguably a self-commitment device. They provide some know-how, but do I need a personal trainer to tell me to work on the treadmill for 20 minutes at a reasonable pace? Probably not. Primarily I think the value is in the commitment to show up three times a week, and if I don't show up, then I'm out $100.

Chevalier: There is some literature about the role of educational interventions. I can really explain to people why they need to save more, but it turns out that it's not all that effective, I believe. What's the state of the art on educational interventions?

Choi: There is still disagreement, but my view is that education is remarkably ineffective, unless it's tied to an action that people can take in response to the education immediately. You walk out of the financial seminar room, go home, and turn on the TV and eat some dinner — at that point you've basically lost it. You need to have some way in the financial seminar itself to push a button and increase your savings rate while it's still at the front of your mind. Generally when you interview people coming out of these financial seminars, they say that they are going to raise their savings rate; they're going to change their asset allocation. You check the administrative data afterwards, and they've pretty much done nothing.

Chevalier: To get back to the pigeons, are there examples of self-commitment devices that work?

Choi: One thing that we have seen in the marketplace is food packages that serve out very small portions — cans of soda that are extremely small and hundred-calorie snack packs. Of course, the question is whether or not that actually reduces consumption. The way it's marketed is that with these smaller package sizes, you commit yourself not to consume too much. And there is work by Brian Wansink of Cornell University suggesting that portion size has a large impact in how much we end up consuming.

Chen: There's actually a huge amount of self-control literature in the animal literature. Just anecdotally, I see monkeys, when they know that they should pull Lever B, but are tempted by Lever A — they'll sit on their hands or they will turn away or go to the other side of the cage, as far away from Lever A as they can, and stare into the corner. There's a lot of self-generated self-control devices that you see.

Choi: You would think that technology could lead to more sophisticated and more pervasive self-commitment devices. You could program your refrigerator not to open between the hours of 7:30 p.m. and 6:00 a.m. And that could conceivably do a lot to reduce food consumption. But this is not in demand, as far as I can tell.

Chevalier: I wouldn't want that. I also think that the lack of a commitment device in the marketplace could well be informative about people's motives. I mean, people don't want to starve in retirement. We know that to be true. So we feel comfortable imposing on them some of these defaults which are really quite paternalistic. But there may be a lot of people who, if you asked them, would say, yeah, I'd rather be five pounds lighter than I am, but not actually at the expense of not eating. I'd like it if I didn't want my midnight snack.

Chen: I have some new research on cognitive dissonance that basically says that the choices that people make say a lot about what they really want.

There has been a really long psychology literature that claims that the mere act of choosing changes the way we feel about things, because we like to think of ourselves as people who make good decisions. These experiments ask a bunch of people to make choices between simple consumer goods. So, which of these two things would you like to take home with you today? Which of these posters would you like to take home with you today and hang in your dorm room? And there has been a really, really long line in the psychology research that claims that, not only in adults but also in monkeys and in four-year-old children and in people who have neural damage and don't even remember what choices they make, that the mere act of choosing between CD A and CD B — say, a Debbie Gibson CD and a Madonna CD — causes people to like the good that they chose more after making the choice than before.

My research is about how that appears to be slightly true, but not nearly as much as the psychologists were claiming. In fact, the reason that this literature has actually been misleading for such a long time is because it hasn't taken into account how, on some level, people tend to be very, very careful in the choices they make, and choose things that are going to make them happier or choose things that make them better off, even though they may not be able to express why this will make them happier or make them better off. We may actually be better about expressing our preferences by making choices about things than when we try to describe why we think like we do.

Interview conducted and edited by Ben Mattison.