Default Effect

The default effect is the disproportionate share of people who end up with the option that requires no action. Despite the name, this is not the credit sense of the word. In consumer finance "default" names a missed obligation, as in a loan default, and it also names a preset choice, as in a default investment option or a default withholding election. Both uses are standard and nothing distinguishes them but context; this page is about the second. The effect matters because a default is unavoidable. Any decision that can be postponed has an outcome attached to postponing it, so whoever designs the form has already answered the question for everyone who does not answer it themselves, whether or not they meant to.

Three explanations are on offer, and a 2011 paper by Isaac Dinner, Eric Johnson, Daniel Goldstein and Kaiya Liu separates them cleanly. Past research, in their words, "has offered potential reasons why no-action defaults matter: (a) effort, (b) implied endorsement, and (c) reference dependence. The first two of these explanations have been experimentally demonstrated, but the latter has received far less attention." Effort is the obvious one: changing something takes a form, a login, or a phone call. Implied endorsement is the interesting one, because it means a default is read as advice. Someone encountering a preset option infers that it was chosen for a reason by somebody who knew what they were doing, which may or may not be true. Reference dependence is the third: the default becomes the point from which every alternative is scored as a gain or a loss, which is the machinery loss aversion and status quo bias run on.

The evidence that separated effort from the rest is the organ-donation work by Eric Johnson and Daniel Goldstein, published in Science in 2003. In an online experiment with 161 respondents, agreement to be a donor was 42 percent when the default was not to be one and 82 percent when the default was to be one, with a third condition requiring an active choice landing at 79 percent, which did not differ significantly from the opt-out version. Changing the answer took a mouse click, which as the authors note largely rules out effort as the explanation. Their country comparison is the figure everyone quotes: four opt-in countries and six opt-out countries with no overlap between the two distributions and nearly 60 percentage points separating them.

The discipline this page owes its reader is the result in the same paper that is almost never quoted alongside that one. When the same authors looked at actual donation rather than registration, presumed consent raised the donor rate from 14.1 to 16.4 per million, an increase of 16.3 percent. A second analysis covering a broader set of countries for a single year put the increase at 56.5 percent, and the authors attribute the spread to which countries each analysis included. So the default moved the paperwork almost completely and the outcome by less, on either estimate, because the paperwork is not the only thing standing between a policy and a transplant. A default effect measured on a form is an upper bound on the default effect measured on the world.

The sharpest financial evidence compares a default against the obvious alternative, which is paying people to decide well. Raj Chetty and colleagues used 41 million observations covering the population of Denmark. About 15 percent of people are active savers who respond to tax subsidies, and they respond largely by moving money between accounts: 99 cents of each krone contributed to a retirement account came from money that would have been saved in a taxable one anyway. The result is that each krone of government expenditure on the subsidy produced less than one cent of new saving, with the upper bound of the confidence interval at 28 cents. The other 85 percent behave passively, and for them the automatic contribution does the work: an employee moving to an employer who contributed one krone more saved about 0.8 krone more in total, without offsetting it elsewhere. The lesson, stated plainly, is that paying people to decide well barely works and changing the default works. The honest limits are that this is one country's system and that the long-run size of automatic-saving gains is smaller once job changes, cash-outs and opt-outs are counted, which savings automation covers.

Because a default is read as advice, setting one is an exercise of authority whether or not it is intended as one, and the reverse case follows immediately. A pre-selected add-on at a checkout, a subscription that renews unless canceled, a service tier already highlighted when a page loads: each is the same instrument with the preferred option chosen by someone whose interests are not the reader's. Nothing in the psychology distinguishes a helpful default from a self-serving one. The only reliable test is the one status quo bias also points to, applied deliberately rather than when something goes wrong: for each arrangement currently running by inertia, would it be chosen again today by someone starting fresh?

The sequence is short. A choice is presented with one outcome attached to inaction. Acting costs effort, invites regret, and moves the person away from a reference point that now belongs to the preset option. The preset option therefore wins a share of the population far above what its merits alone would earn, and that share is largest where the decision is unfamiliar, unpleasant, or easy to postpone. Retirement saving is all three.

A hypothetical, using the passive share the Danish data implies. An employer sets a 3 percent default contribution rate and 1,000 employees each earning $60,000 are eligible. If 85 percent take no action, those 850 people contribute $1,800 each ($60,000 × 0.03), which is $1,530,000 a year (850 × $1,800). Raise the default to 6 percent, change nothing else, and the same 850 contribute $3,600 each, or $3,060,000. Contributions from the passive majority doubled without anyone being asked to do anything, and an employer who instead offered a richer match would have reached mainly the minority who respond to one. Real plans are messier than this: some people opt out, some cut the rate back, and what a plan may set as a default is constrained by statute, which the automatic enrollment page covers.

Pros

• It is the cheapest lever available. Changing which box is pre-selected costs a form revision and reaches everyone.
• It reaches the people that price incentives miss entirely, who are the large majority and who tend to be the least prepared.
• It preserves the choice. Nothing is removed, and the person who wants something else can still have it.
• A well-chosen default is a reasonable answer for someone who would otherwise make no choice at all, which is the realistic alternative rather than a careful one.

Cons

• It works whether the preset option is good or bad, and nothing in the mechanism tells the chooser which one they are looking at.
• Because a default is read as an endorsement, it carries authority its designer may not have earned or intended.
• Effects measured on registrations, enrollments and forms overstate effects on outcomes, sometimes by a wide margin.
• The person who sets the default is often not the person who lives with the result, and the two can want different things.
• Gains from automatic saving leak back through job changes, cash-outs and opt-outs, so a default is a real intervention rather than a complete one.