Status Quo Bias and the Default Effect: The Cost of Doing Nothing

Status quo bias is a preference for whatever state a decision is already in, held more strongly than the merits of that state justify. The default effect is what appears when somebody else selected that state. Both produce the same record, nobody moved, and separating them decides whether a take-up figure describes what people want or what was arranged for them.

Doing nothing is a choice with consequences, and it is the only choice that never feels like one. Every arrangement a person is inside has some state it occupies when nobody acts. A monthly instruction keeps running. A scheme keeps holding what it holds. A connection keeps renewing on the same terms it renewed on last time. The state was selected by somebody, or it fell out of a process nobody examined. For the person inside it, the two amount to the same thing. Once the state an arrangement occupies when nobody acts becomes visible, a large amount of behaviour stops looking like preference and starts looking like design.

What is status quo bias, and what is being preferred?

Status quo biasPreferring the current state beyond what its merits justify. is a preference for the current state of an arrangement, held over and above whatever is genuinely good about that state. The last five words are the ones that get dropped, and dropping them makes the idea useless. If the current arrangement really is the best of the available options, then remaining in it is not a bias at all. Remaining is then a correct decision that happens to require no movement, and calling it a bias would mean calling every well-taken decision a bias whenever the answer came out as stay. The bias is only the extra weight: the amount by which the current state wins beyond what its own merits earn it.

Samuelson and Zeckhauser set this out in the Journal of Risk and Uncertainty in 1988, and the design of the work is what makes the finding hard to argue with. Put the same set of options in front of people. Then change one thing only. Change which of those options is described as the one already in place. Watch the answers move. If the first option beats the second when the first is the one already in place, and the second beats the first when the second is the one already in place, then whatever is doing the work is not the first option and it is not the second. The word doing the work is already.

Take it out of finance before taking it back in. A household has bought milk from the same vendor for eleven years. Asked why, the answer comes back as a list: he is reliable, he comes early, the account is settled monthly without fuss. Every item on that list may be true. A list always exists, so the test is not whether the list exists. The test is what happened the last time a second vendor started on the same street. If nobody in the household looked, the list was assembled afterwards to explain a position that was already being held. In status quo bias the position itself is what is preferred, and the reasons arrive later to keep the position company.

What is the default effect, and who chose the current state?

A defaultThe state an arrangement occupies when nobody does anything. is the state an arrangement occupies when nobody does anything. Every arrangement has one, whether or not anybody sat down and picked it. A monthly instruction either runs or it does not run when the date arrives and nobody has touched it. A person is either inside a scheme or outside it on the morning they take no action at all. There is no third setting in which the arrangement waits politely until somebody decides.

The default effectThe large influence a default has on where people end up. is the influence that setting has on where people finish. The default effect is not the same idea as status quo bias, and the difference is worth holding onto. Status quo bias is a statement about a person: this person weights the current state too heavily. The default effect is a statement about an arrangement: this arrangement sends most people to one particular end state, whatever they may or may not weight. Status quo bias is a fact about a mind, and the default effect is a fact about a process, and the second one is the one somebody built.

Two studies are the ones to name. Madrian and Shea, in the Quarterly Journal of Economics in 2001, examined what happened to a saving arrangement when the state it occupied for a person who did nothing was flipped from outside to inside. Johnson and Goldstein, in a paper called Do Defaults Save Lives in Science in 2003, compared arrangements that differed in whether a person had to act in order to be included or act in order to be excluded. Both report differences much larger than anybody expected before the work was done. How large the difference runs depends on the arrangement studied, so a magnitude carried away from one of these studies travels badly to the next.

Status Quo Bias vs Inertia: is there a real difference?

Yes, and it is the difference between a preference and the absence of one. Status quo bias means a person considered the alternatives, in some rough way, and came out preferring where they already were, with the current state carrying more weight than it earned. InertiaNot acting, without any preference for the current state being involved. means nothing of the sort happened. Nobody weighed anything. The alternatives were never brought into view, so no preference over them was ever formed, and the person stayed where they were for the same reason a parked vehicle stays parked.

Now the uncomfortable part. Consider what each of those two people leaves behind in any record. The first leaves no change. The second leaves no change. A record stores what moved, and neither of them moved, so the record is character by character identical. An hour of staring at it will not resolve the question. Status quo bias and inertia are distinguishable in principle and indistinguishable in evidence, unless somebody goes and asks.

The identity of the two records is not a philosophical complaint. The identity changes what a person is entitled to conclude from a record, and what they would have to do to conclude more. If a survey of 60 investors shows that most of them made no change to a standing instruction across eight quarters, the honest description is that no change was recorded. Everything beyond that, every sentence about what those investors wanted or valued or were comfortable with, is being supplied by the reader rather than by the record.

Why does the distinction decide who is responsible?

Because the two descriptions point at different people. If a person examined the alternatives and preferred where they were, the decision belongs to that person and the only remaining question is whether the weighting was sound. If a person never examined anything, the position they occupy was determined by whoever set the state the arrangement sits in when nobody acts. The person supplied the absence of action. Somebody else supplied the destination that absence leads to.

The two ideas are almost always taught together, and almost always confused. Confusing them moves responsibility silently. To describe a room full of inertia as a room full of preference is to credit every person there with a decision they never took, and quietly to excuse whoever built the arrangement from having made one. The distinction is not about labelling a person correctly, it is about locating where a choice was actually made.

Consider Meera Sundaram, an invented investor. A standing instruction of Rs 25,000/- a month goes into her arrangement. Suppose she never pauses it across eight quarters. Two readings are available. Either she examined it, thought about the amount and left it where it was, and leaving it was a decision. Or the date arrived twenty four times and nothing happened, and nothing happening is not a decision. Devika Rao, the adviser at Palash Advisory Services Private Limited, cannot tell from the record which of those two she is looking at, and any conclusion she reaches about Meera Sundaram from that record alone is a conclusion she has invented.

How large is the default effect, and how would anybody know?

Large enough that the honest answer depends almost entirely on one thing, and the one thing is how much it costs somebody to leave. The cost of leaving is worth a name. Call it frictionWhat it costs in effort or time to move away from the default.: the effort, the time, the form, the phone call, the small dread of getting it wrong. Friction has nothing to do with whether the arrangement is good. Friction is a property of the exit, not of the thing being exited.

The scale below is not a measurement of anything. Its numbers are chosen so the shape can be seen rather than argued about. Take a population whose underlying preferences are exactly even, half of them genuinely preferring the arrangement and half genuinely preferring the alternative. The even split is stipulated, not measured. Stipulating it gives something fixed to watch the reading move against. Now change only the cost of leaving. When leaving is free, 50.0 per cent remain, exactly what an even split looks like. Add a quarter of the full friction and 60.0 per cent remain. At half, 70.0 per cent. At three quarters, roughly what a form and a short delay feel like, 80.0 per cent remain. At full friction, where leaving is genuinely awkward, 90.0 per cent remain.

The comparison between the two ends is where the point sits. The reading climbs from 50.0 to 90.0, a movement of 40 points. Not one of those 40 points came from anybody preferring anything, because the preferences were held exactly even by construction at every single setting. The 40 points came from the exit. A reader handed only the final number, 90.0 per cent, and told nothing about the cost of leaving, would be looking at a measurement of a door and calling it a measurement of a wish.

Why does a default reveal so little about what people want?

Because the reading available is the sum of two things and only the total is ever handed over. One part of it is what people prefer. The other part is what the exit costs. Added together they give a share remaining, and the share remaining is the number that gets quoted. There is no way to run the addition backwards from the total alone. Any pair of parts that sums correctly is consistent with the total shown.

The gap between a choice and an absence of one is what makes revealed preferenceAn ordering read from what somebody chose rather than from what they said. such a delicate tool in this particular setting. Reading an ordering out of what somebody chose is a reasonable move when a choice was actually made. When the outcome was produced by an absence of action, there was no choice to read anything out of, and the ordering that gets written down is being inferred from a door rather than from a person. A default converts the absence of a decision into what looks exactly like the record of one.

So which figure would carry information? The share who remained when leaving was free. At zero friction the exit adds nothing, so the reading is the preference and nothing else. The zero-friction share is the number worth having, and it is also the number almost nobody has. Almost no arrangement in the world makes leaving free. Forms exist. Notice periods exist. The half hour on a phone line exists. The informative measurement is the one the arrangement is least likely to have produced.

What does the decision log show when nobody acts?

Now put the idea against a record instead of a scale. The Palash decision log is an account of 240 decisions taken by 60 investors across eight quarters. Sorted by kind, it holds 96 buys, 84 sells, 36 switches and 24 pauses of a standing instruction, and those four counts add to 240 exactly. The last of the four is the one that matters here.

Twenty four pauses out of 240 logged decisions is 10.0 per cent. Turn it over and the other side is the interesting side. Taking 24 from 240 leaves 216 occasions on which a standing instruction was left alone, and 216 out of 240 is 90.0 per cent of the chances anybody had to interrupt one. Written down like that, 90.0 per cent looks like a finding. The 90.0 per cent is not a finding about people but a count of the occasions on which nobody touched the arrangement.

To read that 90.0 per cent as evidence that those investors wanted their instructions to run is to make, in one move, exactly the error named here. The instruction running is what happens when nobody does anything, so a count of it is a count of absences and an absence has no opinion in it. The log has columns for what was bought, what was sold and when. The log has no column for whether anybody thought about the instruction and decided to let it stand. No such column was ever kept, and no amount of care with the columns that were will reconstruct it. One log is one log, and a single record is not evidence that any rule holds generally.

What happens when a take-up figure is read as a preference?

Here is the failure in its full form, and it is worth slowing down for. Suppose an analyst is told that 90 per cent of a group are in a particular arrangement, and that the arrangement was the state they occupied unless they acted. The question is what that figure says about what those people want. The answer is close to nothing, and the reason is that at least three quite different groups of people would generate that identical figure.

Who chooses the default, and on what basis?

Somebody always does. The claim is more contested than it sounds. The usual objection is that plenty of arrangements were never designed at all. They grew. Nobody sat in a room and picked the state a person occupies when they do nothing. The objection is true about the room and false about the outcome. An arrangement with nobody minding it still occupies some state when nobody acts, and the state it occupies is whatever the process happened to produce: the setting the software shipped with, the form that was easiest to print, the instruction the previous person left running.

So the choice was made either way. Declining to make it, even on purpose, does not leave the position empty. Not choosing a default is not an alternative to choosing one, it is choosing whichever state falls out of the process, without looking at it. Once that is accepted, a second question becomes unavoidable: on what basis should whoever sets that state be setting it, and what do they owe the people it applies to? Thaler and Sunstein set out the design question under the name choice architecture in Nudge in 2008, and choice architecture takes up how a default gets designed, tested and justified.

When is staying put the right answer?

Often, and often enough that it needs saying without hedging. An account of the cost of doing nothing is easily misread as a case for moving, and it is nothing of the sort. Movement has costs of its own, and those costs are measured under excess trading. Staying put is right whenever the alternatives were examined and the current arrangement won on its merits. Remaining on the merits is not status quo bias. Remaining on the merits is a decision whose answer happened to be stay.

So the useful question is not whether somebody stayed. The useful question is when the alternatives were last examined. A clear and recent answer to that question turns staying put into a decision, and the absence of an answer means the arrangement was simply running. Notice that the question is answerable. The answer has a date in it. Answering it needs no survey and no introspection about preferences, and introspection is the thing people are worst at. Meera Sundaram can answer it in about four seconds about her own standing instruction of Rs 25,000/- a month. The instruction moves Rs 3,00,000/- across twelve months and is therefore not a small thing to leave unexamined. The answer is either a date or a silence.

How does a practitioner read a take-up figure without recommending anything?

Devika Rao, the adviser at Palash Advisory Services Private Limited, receives a figure saying that most of the people in an arrangement are still in it. The first thing she does with that figure is refuse to do the obvious thing with it. She does not write that the clients are comfortable, she does not write that the arrangement suits them, and she does not write that it should therefore be left alone. Each of those sentences converts a count of absences into a claim about minds, and none of them is supported.

She writes three lines instead, and the three lines are the practical form of everything above. First, what was the state when nobody acted, stated plainly. A share remaining is meaningless until the reader knows what it is a share of. Second, what did leaving cost: was there a form, a notice period, a call to make, a fee. Third, when was this last put in front of the people it applies to, with a date if one exists. The three lines turn an uninterpretable figure into an interpretable one, and none of them requires knowing what anybody prefers.

A person deciding alone, with no adviser and no committee, runs the same three lines on their own arrangements once a year, and it takes a short evening. Which arrangements are running by themselves. What would it cost to stop one. When did I last look. The exercise usually produces no changes, and usually should not. The value is that afterwards the things still running are running because a person looked at them, and that is a different state of affairs from the evening before, even where every arrangement is identical.

What does the distinction settle, and what does it leave open?

One job has been done and one only. A preference for the current state has been separated from the absence of any preference, the two have been shown to be identical in evidence, and a figure describing where people ended up has been shown to be a figure about an arrangement rather than about the people inside it. The separation is a judgement about what evidence can carry, and it stops precisely there.

Three questions stand next to this one and are answered elsewhere. How a default is designed, tested and compared against alternatives is set out by Thaler and Sunstein in Nudge in 2008, and taken up under choice architecture. The duty attaching to a person or a practice that sets a default for somebody else is a conduct question, taken up under suitability and appropriateness, where the requirements can be sourced properly rather than gestured at. And what happens to a holding over long stretches of time, in the form of activity that never occurs, is taken up under portfolio inertia. The whole of the claim is that the choice exists and that somebody made it, and no smaller claim would still change how a take-up figure should be read.

One last caution, and it is the one most easily lost. No default is good or bad in itself. A default is good or bad only against what the people it applies to would have chosen had somebody asked them, and the claim being made about defaults throughout is a claim about evidence: a share remaining is not a preference, and reading it as one attributes a decision to people who may never have taken it.

Meera Sundaram, Devika Rao, Palash Advisory Services Private Limited, the Palash decision log and the friction scale are invented.
Educational material. Not advice on any investment, tax, budget or market position.