Most academic economists spend their careers writing research that has little impact on the outside world. It’s safe to say that Harvard’s Brigitte Madrian has made an impact. Her studies on the remarkable power of default options in defined contribution plans led to policy improvements that will help millions of Americans be better prepared for retirement.
Just as important, Madrian’s research shook the foundations of traditional neoclassical economics and changed the way academics think about public policy.
This month, Madrian receives the 2015 Achievement in Applied Retirement Research Award, presented by the Retirement Income Industry Association in partnership with Research magazine and sister publication ThinkAdvisor.com. (See sidebar “A Distinguished Award.”)
Madrian, who received her economics Ph.D. from MIT, is Aetna Professor of Public Policy and Corporate Management at the Kennedy School of Government, Harvard University. She also serves on FINRA’s Board of Governors, among other affiliations.
Madrian’s paper titled “The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior,” coauthored with Dennis Shea, was published in the Quarterly Journal of Economics in 2001. It provided empirical evidence that automatic enrollment had a remarkable impact on employee participation in 401(k)s.
This ran counter to neoclassical economic theory’s expectation that defaults would not affect participation. A rational employee would simply save the right amount of money each month based on how much she valued living well in retirement. In reality, people aren’t quite so rational.
In a series of subsequent papers using actual participant data, Madrian and her co-authors established unequivocally that individuals will accept the employer’s default savings rate and the default investment. Both findings contributed to the remarkable success of the 2006 Pension Protection Act (PPA), which provided subtle carrots for employers to default participants into plans and investments that were theoretically appropriate for their life-cycle stage.
Evidence from record-keepers shows that these changes — which neoclassical economics predicts would have no impact on behavior — have transformed how workers invest and how much they save. In particular, average workers in younger cohorts who began working after the PPA are saving at much higher rates than older cohorts who weren’t able to benefit from the power of defaults. The retirement security of these workers can be traced to the applied research of a group of young economists who have pushed the science of economics to accept people as humans, and to improve policy by accepting and adapting to their humanity.
David Laibson, Harvard colleague and a member of this group of behavioral economists and Madrian’s frequent co-author, describes her as “fearless — she tells people what she thinks, whether or not the listener likes the message. This is one of the reasons her work is so important. She gets to the truth of the matter, even if the result is not politically correct or popular.”
“Brigitte’s work is the gold standard for research that is both intellectually rigorous and impactful in practice,” according to University of Illinois professor Jeffrey Brown, who notes that “she is a very funny and fun person. On top of those great qualities, Brigitte is also an admirable example of selfless service to the economics profession. She can always be counted on to lend a hand at any task, no matter how big or small.”
I can confirm she is also a lot of fun to interview. I asked Madrian to discuss how she developed an interest in applied retirement research, and her opinions on changes in policy since her findings were first published. The following are excerpts of our Q&A session.
How did an economist whose earliest research looked at the impact of health insurance on retirement choices start doing research in the field of plan participation?
That project kind of fell into my lap. The company whose data is analyzed in the paper was looking for someone to do a regression analysis of savings behavior in their 401(k) plans. I’d never done any research on savings but I knew enough about the literature to know that almost nothing had been done using administrative data from employers. Then, when I got the data, I learned that they’d implemented automatic enrollment. I knew, even without looking at the data, that this was going to have a big impact on savings outcomes. What I didn’t know was how big the effect was going to be.
I was at University of Chicago at the time, so I started coming at the problem from a rational decision-maker point of view. Then I looked at the data and, lo and behold, we got almost a 50% increase in participation among employees who had been auto-enrolled! As soon as I found that in the data, I knew this was a home run paper. But I didn’t quite know how to explain it. I went to my Chicago colleagues such as Dick Thaler and others who had more of a psychology background to understand what would lead to this particular outcome in the data.
Most of my other colleagues at Chicago were not particularly enthusiastic about this paper. Other than Dick, who loved that paper, the economic luminaries there kept on trying to rationalize away the results because they didn’t line up with the idea that a rational, fully foresighted individual who ought to be saving for retirement wouldn’t sign up for the plan just because there was a small barrier in terms of the cost of initiating participation, and that if you change the default you’d get a dramatically different outcome. That just didn’t square with the Chicago view of the world.
The paper opened a whole new line of research. I joined David Laibson, my colleague at Harvard, Andrew Metrick, who was at Wharton, and James Choi, who had been a graduate student at Harvard, and wrote a stream of several papers using administrative data on savings to better understand the economic psychology of decision making around retirement savings and to document the importance of plan design features that, in a rational economic model, would have little impact but in reality have a tremendous impact on outcomes.
What kind of criticisms have you heard about the use of defaults in policy?
The biggest criticism is that defaults are paternalistic in a negative sense. It is coercive — you’re coercing people to save who otherwise wouldn’t be saving for retirement. That was a refrain that I heard early and often when this research first started coming out. You don’t hear that objection so much anymore.
I think there are a couple of reasons. One is that it went a long way to point out that you have to have a default. You have to have an outcome for people who don’t choose to make an active decision about whether or not they want to be saving. An employer has to pick a default. You could pick the default as not being in the plan, and you could sign up if you wanted to be in it. Or you could have the default as being in the plan, and you can opt out if you don’t want to be in it. It’s not that one decision is somehow less paternalistic than the other. They’re paternalistic in the same way. It’s just that the norm has been one form of paternalism rather than another. You’re not being more paternalistic, you’re just favoring a different outcome that works better for some people and less well for others.
The second fact that went a long way toward quelling the fear of paternalism is the research, part of it done by me and my little team of co-authors, suggesting that most people actually do want to save. Automatic enrollment is not coercing them into doing something they don’t want to do, it’s just facilitating what they actually want to do. Having some evidence that this is actually what people want to do when you automatically enroll them — they don’t get upset about it, they don’t opt out in droves, they don’t get back and see the money’s been taken out and say “aha, something went wrong.” By and large, people were very happy with it.
The third piece is that you can opt out. It’s less paternalistic than a mandate would be.
Some of the criticism levied at the PPA is that many employers are selecting a default savings rate that may be too low. It seems like it was a blunt instrument. Not a lot of thought has been given to deciding what the right default savings rate is. Obviously that’s not going to be the same for every employee.
This is a great example of well-intentioned policymaking gone awry in some sense. Back in the late 1990s early 2000s well before the PPA, employers were still trying to figure out if automatic enrollment was even legal. So the Treasury Department issued a series of letter rulings outlining hypothetical examples of what a company could do in their 401(k). They issued a handful of these about automatic enrollment, and the examples that were used in the rulings were in companies that were using defaults of 2% or 3% of pay.
These percentages were picked just for illustrative purposes. There was no attempt to pick the numbers because they thought that it was what companies should be doing. Someone just picked a number and said ‘let’s go with that.’ They picked a number that ended up sticking in the regulation. In some sense, the letter rulings ended up creating the default and the policy never deviated from it, but it wasn’t a great default to start with. And it wasn’t a well-intentioned default — it was accidental.
As economists, we should also be thinking of having defaults that are perhaps more flexible for individual employees.
We have that on the asset allocation side. Another piece of the PPA was the mechanism for having a qualified default investment alternative (QDIA) that would buy you some regulatory relief. One of the investments that was deemed to satisfy the requirements for being a QDIA was a target date retirement fund.
The industry overnight glommed onto target date funds as a default investment option in their savings plans. If you think about a target retirement being funded, the investment allocation is different depending on how old a worker is. My understanding is that there is nothing in the regulation that would prevent companies from changing the contribution rate. It’s just that no one is actually doing that.
That’s probably in part because it’s a lot more difficult to come up with some prescriptive guidance about what an appropriate default savings rate should be for an individual. The amount of information you would want is probably much larger than the amount of information you’d need for issuing some prescriptive guidance on the investment allocation side. This would include a lot of information that the employer wouldn’t have. Nonetheless, I think employers could do more of that than they currently are doing.
It would seem though that we have enough technology to allow us to do a much better job of customizing defaults for individuals than we currently do. What is the holdup?
My guess is that it’s some conservatism on the part of companies that’s driven by the legal office. What’s in it for me? I can see a lot of risks from doing this — if I innovate and do it wrong then I have a lawsuit on my hands and if I do nothing then nobody’s going to say anything. The upside is for the participant, but there’s not a lot of upside for the company.
Along that line of thinking, explain to me how important the idea of safe harbor was in the PPA?
It was huge. I do think we will probably see more of it in the retirement space. For example, one of the current policy issues is how do we encourage greater annuitization of wealth when individuals get to retirement? I think the only way we’re going to see a dramatic change is if there’s some safety guidance about the kinds of retirement income products that companies can put into their 401(k) plans without being on the hook for huge amounts if that investment somehow turns out to be problematic.
For plan sponsors, there’s no motivation to do much to create DC plans. This is actually a big problem when it comes to maintaining DC plans for the post-retirement stage of the life cycle. There’s not much incentive for employers to innovate in the best interest of employees.
Yeah, it’s the same argument with the differential contribution rate. There’s the potential downside to the employer from doing anything because of liability. All of the benefits, to the extent there are any, would accrue to individual participants and there’s no benefit to the plan sponsor from being proactive and innovative in this space.
How would you change that?
Number one, you can give some safe harbors. Number two, you can create more regulatory pressure. Up until very recently, the U.K. actually mandated partial annuitization. You could try to create some norms or financial incentives around annuitization. Right now, it’s not an easy thing to do. We make it hard rather than easy. I think that’s exactly the challenge. Given that there’s not a lot in it for plan sponsors individually, how do you get them to actually do it?
A Distinguished Award
For nearly a decade, RIIA’s Achievement in Applied Retirement Research Award has honored economists who have contributed to solving the retirement income puzzle. A review of past winners shows how thinkers of diverse outlooks have advanced our understanding.
Theoretical economists such as Zvi Bodie, Moshe Milevsky and Laurence Kotlikoff have provided a much needed rational framework for creating a safer and more efficient retirement income planning strategy. Behavioral economists such as Richard Thaler, Shlomo Benartzi and Brigitte Madrian have provided insight into how real people actually behave.
Scholars such as Olivia Mitchell, Jim Poterba and Jeff Brown embody a new wave of economists who combine economic and behavioral theory to help craft policy solutions that can best move the average worker toward retirement security.
At its best, the social problem of retirement income effectively marries the insights of academics with private sector solutions guided by effective policy. Through the efforts of all these award winners, today’s worker is far more likely to meet the goal of retirement security. —MF