Product × Finance × AI: Financial Products in the Age of AIPart One From 0 to 1中文
Chapter 12

The institutions are in place, and people still make mistakesBias, inertia and defaults

Saving, lending, insurance, banks, shares, markets and delegation are all in place, and custody, disclosure and suitability have their rules. Yet people themselves still make mistakes. In spring a household resolves to store half the harvest, and when autumn comes it eats a good deal more than planned. Seeing which ship earned the most last year, people switch all their ship shares into it. After a storm they rush to sell their ship shares cheaply. People weight the present too heavily, and they have inertia: when they do not want to make a decision, they do nothing. Inertia can be turned to their advantage. Make a good arrangement the default that happens when nothing is done; let people commit now and increase later; let purchases happen automatically on schedule. The architecture of choice can also be used against people. What matters is who sets the default, and for whom.

1. Planned saving abandoned halfway

At sowing time in spring, a household works it out clearly: after this autumn's harvest it will store half, for its own old age. But once the harvest is in and the granary is full, a good meal or a new piece of clothing seems more pressing than old age decades away, and what gets stored is usually far less than planned. Next spring the household makes the same plan, and in autumn the plan fails in the same way.

The new constraint is this: people do not always act on their own plans.

The problem is that people do not weigh the present against the future consistently over time. Ask the household: would you rather eat 1 extra portion of grain next autumn, or 1.5 extra portions the autumn after? It is willing to wait the extra year. But when next autumn arrives, the same question becomes 1 extra portion today or 1.5 extra portions next year, and it no longer wants to wait. The plan it makes today for next year is one it will not want to carry out when next year comes. If it knows it will behave this way, it will be willing to lock itself in beforehand, for example by storing grain in a granary that is hard to draw from.

Today's counterpart. Many people resolve every year to save more, yet find little left at the end of each month.

Planning to save half every year, falling short every year

When the plan is made in spring, this year seems as far away as the future, and a household intends to save half its harvest; at harvest time in autumn, the grain in front of it weighs more. Drag "weight on the present" to see the gap between plan and outcome over five years; tick "lock in beforehand" to see what happens if half is put in spring into a granary it cannot draw from.

Share planned in springShare actually saved in autumn

Illustrative: logarithmic utility, two periods, no interest or spoilage. With no extra weight on the present, half is saved; when present consumption is given extra weight (Laibson's 1997 quasi-hyperbolic discounting, with β the weight on the future relative to the present), the share actually saved in autumn is β/(1+β). Here "weight on the present" is 1−β.

2. Inertia: doing nothing is also a choice

Another common tendency is inertia. Joining a savings plan means filling in a form, choosing a rate and choosing which ship shares to buy. When people do not want to make these decisions, they do nothing, putting it off to next month and then to next year. As a result, whether people save depends to a large extent on what happens when they do nothing.

Madrian and Shea (2001) studied a US company that changed its retirement savings plan from "opt in voluntarily" to "enrolled automatically, free to opt out". Participation among new employees rose sharply. Most also stayed at the default contribution rate and default fund the company had set, and few adjusted them.

Today's counterpart. Many countries have moved their workplace pension plans to automatic enrolment: employees who make no choice are enrolled by default.

What happens when people do nothing

A hundred new employees, eight in ten of whom want to join the savings plan. Each month, each person has only a certain probability of actually acting to change their status. On the left, "opt in voluntarily"; on the right, "enrolled automatically, free to opt out".

Opt in voluntarily

Enrolled automatically, free to opt out

SavingNot savingStatus differs from the person's wish

Illustrative model: those who want to join and those who do not change their status only in the month they "act", and the probability of acting is the same every month. Madrian and Shea (2001) found that after a US company switched to automatic enrolment, participation among new employees rose sharply, and most stayed at the default contribution rate and fund.

3. The two sides of defaults

Inertia makes defaults especially powerful: whatever is set, most people stay with it. This can help people, and it can also harm them. If the default contribution rate is too low, many people save too little for good. If the default investment is too conservative, many people earn very low returns for decades. A default is not neutral; whoever sets it is in effect deciding for most people.

Defaults are not the only thing at work. For the same set of options, which one is listed first, which one is already selected, and which step requires the person to fill something in all change what people end up choosing, and no arrangement is entirely neutral. Since someone has to decide on the arrangement anyway, it can be arranged deliberately in people's favour, while leaving everyone free to choose otherwise.

Today's counterpart. Automatic-enrolment retirement plans usually also set a default contribution rate and a default investment, such as a target-date fund that adjusts the stock–bond mix automatically with age.

Whatever the default, most people stay with it

A hundred employees each have the contribution rate they would choose after thinking it through, about 8% on average. Six in ten never change the default. Choose different default rates to see the resulting distribution and how many save less than they themselves want.

Stayed at the defaultChose for themselves

Illustrative: each person's desired rate lies between 2% and 14%; the six in ten who never change the default are chosen at random, unrelated to the rate they want. "Default 14%" is higher than most people want, to show that a default set too high also pulls people away from their own intentions.

4. Commit now, increase later

Since people weight the present too heavily and have inertia, both tendencies can be put to work: commit now to raising the savings rate step by step out of future pay rises. The commitment is made today, but nobody has to eat less today. At each future pay rise the savings rate goes up automatically, and those who do not want to change anything need do nothing. Thaler and Benartzi (2004) introduced this "Save More Tomorrow" plan at one company. Participants' average savings rate rose from 3.5% to 13.6% within 40 months, and most stayed in the plan. A research review by the US Department of Labor noted that the study did not randomly assign participants, that the decision to join may itself reflect a willingness to save, and that the causal evidence is therefore limited.

Today's counterpart. Some retirement plans offer an option to raise the contribution rate automatically each year; regular fixed-amount investment plans make purchases happen automatically on schedule.

Commit now, increase out of future pay rises

An employee currently saves 3.5% of pay and gets a 3.5% pay rise each year. Compare two approaches: raising saving to 10% right now; or committing now to save a few more percentage points automatically at each future pay rise. The upper panel shows the savings rate; the lower panel shows monthly take-home pay.

No changeRaise to 10% nowSave more at each pay rise

Illustrative: pay starts at 100 and rises 3.5% at the start of each year; the savings rate is capped at 15%. Thaler and Benartzi (2004) report that at a company that actually introduced this plan, participants' average savings rate rose from 3.5% to 13.6% within 40 months; a research review by the US Department of Labor found that the study did not randomly assign participants, so the causal evidence is limited.

5. Chasing gains and selling into losses: getting less than the product itself

Bias also shows up in buying and selling. Whichever ship earned the most last year, people pile into it, often buying near the top. After a storm, when ship shares fall sharply, people rush to sell, often near the bottom. The return on the ship shares themselves is one thing; the return principals actually receive, given when they bought and sold, is another. Dichev (2007) compared returns that take the timing of money flowing in and out into account with buy-and-hold returns, and found that on the New York and American stock exchanges from 1926 to 2002 the former were about 1.3 percentage points a year lower. Keswani and Stolin (2008) argue that this result is not robust: over the last two-thirds of the sample, the sign of the difference for the New York and American exchanges even reverses.

Today's counterpart. Fund investors buy heavily after good performance and redeem heavily after declines, so the return investors as a whole receive is often lower than the return of the fund itself.

Chasing gains and selling into losses: the principal gets less than the ship shares

The price of ship shares fluctuates around a long-run trend of 6% a year. A principal intends to invest 10 portions each year; when chasing gains and selling into losses, the principal invests more after a strong year and less, or even sells, after a fall. Drag how strongly the principal chases, and compare three returns.

Ship share priceInvested that yearSold that year

Illustrative: price = long-run trend × (1 + 0.3 × cyclical swing), with an eight-year cycle. "The ship shares themselves" is the annualized buy-and-hold return; the principal's return is calculated from the timing of money flowing in and out (internal rate of return), which is what Dichev (2007) calls the return that takes the timing of flows into account. The difference between regular scheduled investment and the ship shares themselves comes only from when money goes in, not from any judgement.

6. The architecture of choice can help people or harm them

Defaults, ordering and the steps required can help people keep to their plans, and sellers can also use them for profit: pre-ticking an add-on service nobody needs, hiding the cancel button deep in a menu, putting high-fee products at the top of the list. The power of a default comes from people's inertia, so who sets it and for whom matters more than the default itself. A default set for the principal should be the one most people would choose after thinking it through. A default set for the seller turns inertia into the seller's revenue.

Today's counterpart. The EU Consumer Rights Directive requires the consumer's express consent before any extra payment beyond the price of the main contract. If a trader infers consent through a default option, meaning the consumer must untick a box to avoid the payment, the consumer is entitled to have that payment refunded.

The same options, two architectures

Two account-opening pages offer exactly the same options; only the defaults, ordering and steps differ. Switch between the two designs to see what a hundred principals end up choosing.

Illustrative: assume seven in ten people accept the page's defaults and its first option, while the rest choose according to their own needs; when leaving takes three extra steps, only three in ten of those who want to leave actually complete it.

7. Keeping to the plan: Civilization Simulator, version 12

The simulator unlocks the last rule of Part One: "Defaults". Forty households must save for old age thirty years from now. You can turn automatic saving, automatic regular investment and a cooling-off period during declines on or off, and adjust how strongly people weight the present and how readily they sell in panic. Then see how much each household has saved after thirty years, and how far the return principals actually receive falls short of the return on the ship shares themselves.

Civilization Simulator · Chapter 12: keeping to the plan

Forty households save for old age thirty years from now, and each intends to save 12% of its income. Turn automatic saving, automatic regular investment and the cooling-off period on or off, and adjust how strongly people weight the present and how readily they sell in panic. See how much each household has saved after thirty years, and the return principals actually receive.

SharesMarketInformation in pricesDelegationRegulation & custodyDefaultsAgency costs fall · Part Two
Median savingsHouseholds from the 10th to the 90th percentileMedian with all three rules offYears of sharp declines

Illustrative model: each household earns 1 portion a year, and savings are measured in "years of income". Ship shares return 7% a year on average with 12% volatility, and about once every twelve years suffer an extra 25% decline. Without automatic saving, households must join on their own; the more they weight the present, the later they start and the smaller the share they save. With automatic saving, 12% is set aside from the first year. Without automatic regular investment, households save and buy more after a strong year and buy less after a fall. After a sharp decline, households prone to panic sell, hold cash for three years and then buy back; with a cooling-off period, only one in five actually sells. The return from "investing on schedule and never selling" is the internal rate of return of a household that buys 12% on schedule every year in the same run of years.

8. What this step leaves open

Starting from one person's harvest in one year, today's financial system has grown step by step. Looking back, each institution carries a cost:

  1. The cost of delegation. Research, monitoring, checking and fees all require someone's time and attention, so ordinary people can only hand their money to a small number of professional institutions.
  2. The cost of checking. Custody, disclosure and audit all exist to make the invisible visible, and seeing clearly itself takes labour.
  3. The cost of information. Gathering news and analysing fleets take time, so information advantages are worn away by competition.
  4. The cost of keeping to plans. Defaults and commitment devices help people overcome bias, but someone has to design and run them, and they can also be used against people.

A large part of these costs is the cost of paying someone to think and act on your behalf. If this cost fell sharply, which of these institutions would change, and which would not?

What this step leaves open

Every institution carries a cost. Select a cost to see the assumption behind it.

Part OneInstitutions derived in twelve steps

9. Implications for products

Regular investment plans, automatic-enrolment retirement plans and target-date funds all rest on the mechanisms in this chapter. Three implications follow. First, a product's value lies not only in the returns it can deliver but also in how far it helps users keep to their own plans; the return users actually receive is the test of a product. Second, the default is the most powerful design element in a product. It should be set to the option most users would choose after thinking it through, and changing it should be easy enough. Third, the architecture of choice can help people or harm them; any design that relies on inertia should be asked whose interest it serves.

Chapter summary: bias, inertia and defaults

Four conclusions, each tied to one mechanism in this chapter.

Weighting the presentPlans and actions diverge

In spring the plan is to save half; in autumn less gets saved. Locking in beforehand is what keeps the plan.

InertiaDoing nothing is also a choice

Whether people join and how much they save depend largely on what the default is.

Commitment and regular investmentMake the good arrangement the default

Commit now and increase out of future pay rises; buy on schedule rather than chasing last year's gains and losses.

The architecture of choiceWhom the default is set for

The same inertia can help people keep to their intentions, or become the seller's revenue.

References