How Long Do Nudges Last?

How Long Do Nudges Last?

VLADIMIR: Well? Shall we go?

ESTRAGON: Yes, let’s go.

They do not move.

—Samuel Beckett, Waiting for Godot

One question that we have not discussed is whether nudges are long lasting.5 One possibility is that people initially exhibit default behavior for reasons such as status quo bias,6 laziness, procrastination, and so forth, but over time get their act together and make sensible changes to their initial choices. In such a world, the specific design by the choice architect has only a transitory effect. But if the effects of nudges are persistent, then the choice architecture design can be critical, with effects lasting decades. This Swedish experience offers a unique opportunity to provide some insight on this issue because it is possible to track what has happened from the launch up until the end of 2016.

First, some background. After the initial launch of the pension system with all its fanfare, the pension plan became less salient to everyone. The government reduced its advertising significantly, and so did the individual funds, because most of the citizens had already been onboarded. Whereas the initial cohort in 2000, when the plan was launched, included all 4.4 million people who were in the workforce at the time, subsequent cohorts only consist of new entrants into the system; that is, mostly younger people when they start to earn an income, plus new immigrants. For example, the 2016 cohort had only 183,870 people, not enough to make advertising an economically attractive option for fund managers.

In the absence of public and private advertising, over time the public gradually disengaged, and the default had its more typical effect. By 2003, only three years after the new system had launched, fewer than one out of ten (9.4 percent) of the new participants elected to be Active Choosers, a share that declined to only 3 percent in 2010 and was below 1 percent in the most recent years.

Moreover, participants seem to have adopted a “set it and forget it” mindset. When first confronted with a choice, they made a decision, but most failed to revisit it. We can see this by looking at how individuals’ portfolio choices varied depending on when they joined the system. To see how we can study this issue consider two hypothetical retirement savers, Madeleine and Per, who were born on January 1, 1982, and were thus eighteen when the Premium Pension Plan was launched. They were both attending university at the time, but Madeleine had a part-time job when she started school, making her eligible to join the pension system, whereas Per did not begin working until 2002. Both experienced all the advertising accompanying the launch of the system, but only Madeleine was nudged to think about making a choice at that time. Per never gave it a thought. To estimate the differential effect of being exposed to the ads depending on whether they were relevant, we can compare people in Madeleine’s situation to that of Per’s, by comparing the choices of young people who joined in 2000 to those who joined in the next couple of years.

Statistical analyses show that after controlling for other observable characteristics, the odds of becoming an Active Chooser were about six times higher for people in the 2000 cohort compared with those in the two years just following the launch, 2001 and 2002. Thus, the ads mostly affected the people who were in a “deciding” mindset when those ads ran.

Another question of interest is what happens over time. Do people stick with their initial choices or do they reconsider over time as they see how things are working out? One way of looking at that question is to ask what proportion of people changed their mind and switched from being a Delegator to an Active Chooser, and vice versa. It is possible to study the choices of the 4.4 million Swedish retirement savers who joined the system in 2000 and then follow them from the launch until 2016.*

Just over a quarter (27.4 percent) of the initial Delegators changed their minds and decided to become Active Choosers.7 The bulk of these switches occurred during the first decade after the initial choices. What prompted them to become active? Some of these switches were, uh, “helped” by third parties who offered to give participants investment advice. This was particularly common in the early years, when advisers could make changes on behalf of clients easily if they were given a PIN number. (Sensibly, these rules were later changed.) Thus, the 27 percent represents an upper bound on the number of Delegators who independently decided to start managing their own portfolio. Everyone else just stuck with the default for at least the sixteen years we’ve observed.

Perhaps surprisingly, the impact of the advertising to encourage Active Choosing was even more persistent. Only a tiny percentage (2.9 percent) of the initial wave of Active Choosers ever switched to becoming a Delegator. Once an Active Chooser, always an Active Chooser!

We have seen so far that switching between being an Active Chooser and a Delegator was uncommon. Most people chose one strategy or the other and stuck with it. Additionally, the group we are calling Active Choosers was not very active. Recall that we gave them that label on the basis of a single decision (under the influence of big ad campaigns) to pick funds for themselves. That was pretty much the end of their “activity.” For the group, the median number of trades over the entire sixteen-year period was just one. This is a similar level of activity that is observed by American investors in 401(k) plans.

A natural follow-up question is what it would take to capture the attention of the vast majority of investors who were behaving quite passively. Could anything awaken them from their stupor? Two events give us a chance to examine this issue: one that affected the default fund and another that concerned one of the individual funds in the pension system.

The default fund has undergone some changes over time. While it has always been a low-fee fund with global diversification, we mentioned that when it was first launched it had some quirky features such as the home bias toward Swedish stocks and some small investments in hedge funds and venture capital. In 2010, the fund switched to become essentially a global index fund (investing all over the world) and fees were reduced further to 0.11 percent.*

In 2010, the Swedish government approved a much more radical change: It decided to permit the default fund to employ financial leverage at the discretion of the fund’s management. The statute allowed for up to 50 percent leverage. What this means is that the fund managers were allowed to essentially borrow money to buy even more stocks, and they took advantage of the new discretion. Fifty percent leverage means that if the market went up by 10 percent, the fund would grow by 15 percent, but conversely, if prices fell, the fund would fall by 50 percent more. That is pretty risky!

Any investor who was worried about this massive increase in the riskiness of their portfolio had a great alternative: they could switch at no cost to another fund that was identical to the default fund, but without the leverage. But almost no one did. This is particularly surprising because a study of Swedish investors reported that users of the default fund considered themselves relatively more risk-averse than average and said they wanted a safe investment.8 It seems likely that they just didn’t notice (or understand) the change in the fund.

The other event that tested the limits of investor inertia came in January 2017 when a leading Swedish business magazine reported that the CEO of Allra, one of the fund companies in the pension system, had purchased the most expensive house in Sweden the previous year. Oh, and he bought a helicopter. Pro tip: if you plan to steal money from your investors, it is probably not a good idea to flaunt your wealth. Less than a month later, one of the leading Swedish newspapers launched a series of articles pointing to possibly fraudulent practices by Allra. After a few weeks, the Swedish Pensions Agency decided to prohibit people from switching into Allra’s funds, pending a fraud investigation. It is important to emphasize that people were still permitted to shift contributions out of Allra to other funds at any time and at no cost.*

Before the fraud allegations, Allra participated in the pension system with four different funds. A total of 123,217 investors had picked these funds, with about $2 billion of assets invested. One might expect a significant portion of these so-called Active Choosers to dump their Allra investments based on the credible news of possible fraud. In fact, a sudden run on the fund might well have been expected. Yet nothing of the kind occurred. In the week after the revelation of the fraud allegations, only 1.4 percent of the Allra investors sold their shares. And even after the accounting firm Deloitte, which thereafter resigned as auditor, reported Allra to the authorities, only 16.5 percent of the start-of-the-year investors had elected to move their money into some other fund.

What do we conclude from this episode? At the time of the scandal, in early 2017, the number of funds in the system had grown to almost nine hundred, which is obviously too high. In fact, Sweden was quickly approaching the absurd situation in which there would be more funds offered than there were new participants becoming Active Choosers. We think most would agree that one fund per investor seems more than necessary. Furthermore, it is clearly impossible for a small country like Sweden to adequately monitor such a large number of funds. Reporters, not regulators, discovered the scandal involving Allra.