Present bias

Fruit for Friday

On Monday, an office manager takes snack orders for Friday’s meeting and picks an apple for himself. On Friday, the tray arrives with apples and brownies, and he takes a brownie.

Between Monday and Friday nothing changed about the apple or the brownie except that the brownie went from “later” to “now”. Choosing a brownie isn’t a mistake. The claimed bias is the reversal: a preference that held while both options were distant flipped once one of them was immediate. This snack pattern is one of the settings where studies have found it most clearly (see Evidence).

Starting tomorrow

Every evening for three months, someone plans to spend the next evening scanning a box of old family photos. Each next evening, it seems more reasonable to start tomorrow.

The task is unpleasant now and valuable later, so each day’s effort is weighed more heavily than the same effort a day off. The plan is sincere every time it is made; what defeats it is that tomorrow’s evening becomes today’s. Economists call this a person being naive about their present bias: each night they expect their future self to follow through. Studies of unpleasant work tasks have found both parts of the pattern, putting off effort and underestimating how much they would.

A dollar today, or later

Asked in a survey, a man prefers $100 today to $110 in a week. Asked about $100 in 52 weeks or $110 in 53 weeks, he picks the $110.

The trade is the same in both questions: one more week for ten more dollars. Only the first involves “now”, and the choice flips. This is the textbook demonstration, and it’s where the research is weakest: with money, carefully designed studies often find this reversal small or absent. Taking the $100 today can also be reasonable for reasons that have nothing to do with bias, which is why the look-alikes below matter.

• Procrastination: when the cost comes now and the reward later (chores, paperwork, exercise), present bias leads people to put things off. Illustration: the photo box above.
• “Preproperation”: when the reward comes now and the cost later (a treat, a purchase), it leads people to do things too soon. The term is O’Donoghue and Rabin’s.
• Naive and sophisticated: a naive person expects their future self to be patient; a sophisticated one expects the bias and plans around it, for example by removing temptations in advance. In the O’Donoghue and Rabin model, sophistication reduces procrastination but can make preproperation worse.

• When the later reward is less certain. A promised payment may not arrive, and a future benefit may not materialize. Preferring the sure thing now is a response to risk, not a bias about time.
• When you need the resources now. If money today pays a bill or avoids a fee, or can earn more than waiting would pay, taking it early is sensible.
• When something besides timing has changed. Reversing a plan because your circumstances, information or tastes really changed isn’t time inconsistency.
• When the impatience is consistent. Preferring sooner things at a steady rate, whether the delay starts today or in a year, is ordinary discounting. Present bias is the extra weight on now.

The test: if both options were pushed a month into the future, would I still choose the same one, and if not, what besides the timing is different?

The client who pays late

A freelance designer is offered $900 on delivery or $1,000 in 60 days. The client has paid her late twice before, once only after several reminders. She takes the $900.

She’s choosing less money sooner, which on the surface looks like the survey reversal above. But the $1,000 isn’t really $1,000 in 60 days; it’s a promise from someone with a poor record of keeping it. That’s the less certain condition. Pushed into the future with the same client, the choice between a payment on delivery and a larger, later one would carry the same risk.

Rent due Friday

A student is owed $400 from a summer job. He can take $350 this week or the full $400 next month. His rent is due Friday, and paying late costs a $75 fee. He takes the $350.

Waiting would gain him $50 and cost him $75, so the earlier, smaller payment is worth more to him. That’s the need the resources now condition. The research on money choices points to exactly this problem: people can borrow, save or face bills, so a choice between payment dates doesn’t directly show how they weigh the present.

Economists describe present bias with models of discounting, the rate at which future outcomes lose weight in a decision. In the standard model, the rate is constant, so preferences between two dates never reverse as they approach. Hyperbolic discounting refers to a discount rate that falls with delay, so a week’s wait matters more when it starts today than when it starts next year. The quasi-hyperbolic model used by David Laibson and others simplifies this into a single extra penalty on everything that isn’t immediate. Both are models of the pattern rather than other names for it, and the evidence that measured discount rates decline with delay has other possible explanations.

A common informal explanation is that immediate rewards and costs are vivid and felt, while future ones are abstract: the effort of a task today is real, and the effort of the same task next week is only imagined.

Present bias is closely related to misprediction of your future self. If you don’t expect to change your mind, you don’t guard against it, which is part of why plans slip. This overlaps with the Planning fallacy, where forecasts of how long a task will take are built from the plan; present bias adds a reason the plan isn’t followed once the work is due.

• Commit in advance, if you know you tend to delay. In Augenblick, Niederle and Sprenger’s study, participants who showed more present bias in allocating work were more likely to choose a binding commitment that locked in their earlier plan. That’s evidence that people use commitment when they recognize the problem, not a trial showing commitment always helps. Paying up front to bind yourself is also the legitimate case the Sunk cost fallacy entry describes.
• Don’t assume your future self will be more patient. In one study, people understood no more than about a quarter of their own present bias. Plan for the self who will actually face the choice.
• Apply the test above. Ask whether you’d make the same choice if both options were in the future, and whether anything other than timing has changed.

Status: contested. The pattern clearly exists in some settings: studies of effort and consumption repeatedly find people choosing differently when an option is immediate. But its size and generality are seriously disputed. A meta-analysis finds little or no present bias on average in carefully designed studies of choices about money, the effect in effort studies is weakened by signs of selective reporting, and reviewers argue that common measurements mix time preference with trust, uncertainty and financial options. The review for this entry found no preregistered or multi-lab replication of the central claim.

Foundational work.

• Thaler (1981) asked people how much money they would need in one month, one year or ten years to be indifferent to receiving $15 now. The median answers ($20, $50 and $100) imply annual discount rates of about 345%, 120% and 19%: impatience falling as the delay grows, the pattern hyperbolic discounting describes.
• Preference reversals of the “$100 now over $110 tomorrow, but $110 in 31 days over $100 in 30 days” kind have been reported in many studies, reviewed by Frederick, Loewenstein and O’Donoghue (2002).
• Laibson (1997) analyzed a consumer with hyperbolic discounting and showed why such a person would want to restrict their own future choices, for example by holding savings in forms that are hard to spend. O’Donoghue and Rabin (1999) modeled procrastination and doing things too soon, and showed how much outcomes depend on whether people are naive or sophisticated about their bias.

Where the evidence is strongest: effort and consumption.

• Read and van Leeuwen (1998) had people choose snacks either a week in advance or at the time of eating. Averaged across conditions, unhealthy snacks were chosen 51% of the time for the following week and 83% of the time for immediate eating.
• Augenblick, Niederle and Sprenger (2015) ran a longitudinal study in which participants allocated unpleasant work tasks between dates, alongside a companion study of money. They found very limited time inconsistency with money but considerably more with effort (estimated present-bias parameters of about 0.97 and 0.90 respectively, where 1 means no present bias and lower means more).
• Augenblick and Rabin (2019) had 100 participants choose, over seven weeks, how many unpleasant transcription tasks to do now and on future dates. They chose 10–12% fewer tasks for the present than for any future date, and their predictions implied they understood no more than 24% of their present bias.
• DellaVigna and Malmendier (2006) found that gym members who paid over $70 a month attended about 4.3 times a month, paying more than $17 a visit when a pay-per-visit option cost $10. The authors’ leading explanation is overconfidence about future self-control or efficiency, which fits naive present bias but isn’t a direct measure of it.

Reasons for doubt.

• Money choices often show little present bias. Using a method that lets people divide money between two dates, Andreoni and Sprenger (2012) found reasonable discounting and, in their words, “surprisingly, dynamically consistent time preferences”.
• Meta-analysis. Imai, Rutter and Camerer (2021) pooled 220 estimates from 28 articles using that method. On average people were present-biased, but only slightly (estimated parameters of about 0.95 to 0.97, on the same scale). Studies using money averaged close to no present bias; studies using effort averaged about 0.88. They found evidence of selective reporting in the effort studies, and the difference between money and effort weakened after correcting for it. Depending on the correction method, the corrected averages ranged from modest present bias to almost none.
• Measurement. Cohen, Ericson, Laibson and White (2020) review why choices between amounts of money at different dates may not reveal time preferences at all: people can borrow and save, so money received today isn’t necessarily spent today, and choices also reflect trust that the later payment will arrive, confusion and other factors. Frederick and colleagues had earlier listed confounds including uncertainty, inflation and the chance to invest. In one study they describe, 82% of respondents took a smaller immediate sum over a larger one a month later when both were certain, but only 39% did when both were given a 50% chance.
• Other explanations for declining rates. Some of the evidence that impatience falls with delay may instead reflect how intervals are divided up, a pattern called subadditive discounting, which Frederick and colleagues review.

What remains uncertain. How large present bias is outside effort and consumption tasks; how much of the pattern with money reflects time preference rather than uncertainty, trust and financial options; how well lab estimates predict real behavior; and whether the effort findings hold up in preregistered, larger replications.