Does Precommitment Really Work
A popular self-help book backs precommitment with a smoking trial and a figure of 52 per cent. The paper, and two field experiments beside it, show that the device works for the people who take it up and that most people offered it decline.
Answer to the question
Why do we pay to limit our own choices?
People expect their later selves to give way, and a binding offer removes or prices the temptation before it arrives. Field trials in the Philippines and the United States found that people who take such offers do better, but most people offered them decline, so the average effect across everyone is smaller.
Key takeaways
- Halpern and colleagues (2015) randomised 2,538 participants: 90.0 per cent accepted a reward programme for smoking cessation and 13.7 per cent accepted a deposit programme.
- Across everyone assigned, six-month abstinence was 6.0 per cent on usual care, 15.7 per cent with rewards and 10.2 per cent with deposits. The 52 per cent in Eyal's book is the rate among the minority who accepted (52.3 per cent in a published report of the trial).
- Among people who would accept either kind of programme, deposits came out ahead of rewards by 13.2 percentage points (95 per cent confidence interval 3.1 to 22.8).
- In the Philippines, 202 of 710 clients offered a commitment savings account opened one (Ashraf and colleagues, 2006), and about 11 per cent of smokers signed a deposit contract that raised the six-month pass rate by three points (Giné and colleagues, 2010).
- Crockett and colleagues (2013) found in a scanner study that blocking a temptation beforehand made the larger, later reward easier to choose than willpower did, with the biggest gain for the most impulsive volunteers.
In the field trials that offered it, most people turned precommitment down. Precommitment means removing or pricing a future choice while the mind is clear, and it has genuine evidence behind it, from a bank, a smoking trial and a scanner.
Nir Eyal's Indistractable gives it the last part of his method and backs the section on money with a smoking trial in which, he reports, more than half of the people who staked their own cash quit. The trial is real and that figure is in it. Another figure sits beside it in the paper: only a small minority of the people offered the deposit programme accepted it. That denominator changes how the Philippine and American trials around it read as well.
Eyal's Pacts and One Trial
Indistractable builds its method in four parts, and the fourth, on pacts, comes last by design. Eyal describes three kinds. Effort pacts add friction, such as a kitchen safe with a timer lock or software that blocks sites. Price pacts put the reader's money at stake, and Eyal explains their pull through loss aversion, the tendency to weigh a loss more heavily than an equal gain. Identity pacts attach a behaviour to a self-image, so that a choice becomes a description of who the person is.
Of the effort kind, Eyal admits a determined person could smash the kitchen safe with a hammer, and that the extra effort is the whole mechanism. His favourite illustrations of the first kind are writers: Jonathan Franzen worked on a computer with its wireless card removed and an Ethernet cable glued into the port and sawn off. A close cousin of the effort pact, temptation bundling, gets its own test in [Temptation Bundling Needs a Lock](https://thewhyweproject.com/articles/temptation-bundling-needs-a-lock).
Eyal is emphatic about order. Pacts are a last resort, to be used only after the reader has dealt with internal triggers, planned their time and handled external triggers, and he says pacts made earlier are useless. He also concedes that researchers do not fully understand why precommitment works as well as it does (pp. 155 to 159). The order is his own claim.
For price pacts he reports a three-arm trial from the New England Journal of Medicine (pp. 167 to 173). Six per cent of smokers quit on education alone, 17 per cent when offered $800 for quitting, and 52 per cent when they staked $150 of their own money, returned with a $650 bonus on success. The book names the journal and not the authors. The paper is Halpern and colleagues (2015), a randomised trial of four financial-incentive programmes with 2,538 participants, drawn from the employees of one American company and their relatives and friends. His figures for the two incentive arms are counted among acceptors only, which is the departure from the trial's headline result.
Who Proposed Binding the Self?
The first formal treatment the vault has confirmed is Robert Strotz's 1955 paper on myopia and inconsistency in dynamic utility maximisation. It asks what a person should do when the self that makes a plan and the self that carries it out want different things. Strotz set out two answers: plan around what the later self will do, or bind the later self in advance. Ulysses having himself tied to the mast is the stock picture of the second.
Economists later turned the question into field experiments on precommitment and its cousins. A 2010 review of commitment devices in the Annual Review of Economics, by Bryan and colleagues, covers laboratory and field studies and stresses that predictions about when people will demand commitment depend heavily on the modelling assumptions. Demand for precommitment, in other words, is something to be measured and not assumed.
The laboratory case rests on a scanner study. Healthy male volunteers chose between a smaller reward at once and a larger one after a delay, sometimes by resisting the temptation as it appeared and sometimes by blocking it beforehand. Blocking made the larger, later reward easier to choose, and the gain was bigger for the more impulsive volunteers (Crockett and colleagues, 2013, Neuron). Brain activity differed too: precommitment drew on the lateral frontopolar cortex, at the front of the brain, which the willpower trials did not.
The result says that precommitment moves the work of self-control earlier, to a moment when it costs less. It cannot say whether anyone will choose to do the moving. Whether willpower itself is a depleting resource is a separate dispute, taken up in [Willpower Is Not a Fuel Tank](https://thewhyweproject.com/articles/ego-depletion-debunked).
Three Trials, One Denominator
Three field experiments put a binding offer in front of real people and counted who took it.
| Trial | The offer | Took it up | Result across everyone offered |
|---|---|---|---|
| Ashraf, Karlan and Yin, 2006 | A savings account with restricted withdrawals, offered to 710 clients of a Philippine bank | 202 opened one | Balances after twelve months 81 percentage points higher than controls |
| Giné, Karlan and Zinman, 2010 | A six-month deposit, forfeited to charity if a nicotine test was failed | About 11 per cent | Three percentage points more passed the six-month test; the gap held in surprise tests at twelve months |
| Halpern and colleagues, 2015 | Reward programmes against deposit programmes, with a usual-care arm | Deposit 13.7 per cent, reward 90.0 per cent | Six-month abstinence: usual care 6.0 per cent, reward 15.7, deposit 10.2 |
Set against Eyal's three kinds, the savings account and the smoking contract are price pacts, since money is locked away or forfeited. The scanner task is closer to an effort pact, since volunteers made the temptation unavailable. None of the three field trials tests an identity pact, and the book's identity-pact evidence is a separate body of work.
In the savings trial, 202 of the 710 clients offered the account opened it, about 28 per cent. The estimate for the whole group offered, takers and decliners together, was still a large rise in balances after a year. The authors argue that it reflects a lasting change in saving and not a short reaction to a new product.
The American trial splits more cleanly. Reward programmes were accepted by 90.0 per cent of those offered them and deposit programmes by 13.7 per cent. Counted across everyone assigned, the reward arm won, and all four incentive programmes beat usual care. The trial's instrumental-variable analysis asks a different question, what happens among the estimated 13.7 per cent of participants who would accept either kind of programme, and there the deposit programmes came out ahead, by 13.2 percentage points (95 per cent confidence interval 3.1 to 22.8).
Eyal's figures belong with the second kind of comparison. The 52 per cent is the six-month rate among people who agreed to the deposit programme, a self-selected group, and his 17 per cent is the matching rate among those who agreed to rewards (52.3 and 17.1 per cent in a published report of the trial). It is a true rate for them and a misleading one for the programme, which, across everyone assigned to it, produced 10.2 per cent.
It is a true rate for them and a misleading one for the programme.
Why Most Decline the Lock
The smoking contract in the table shows the problem with precommitment most plainly. About 11 per cent of the smokers offered it signed up, and the offer raised the pass rate by three points across everyone. On the simplest assumption, that the offer changed nothing for those who declined, three points spread over 11 per cent of smokers is a gap of about 27 points among the takers. That figure is arithmetic on the published numbers and not a result the authors report, but it shows how a small average can hide a large effect.
Who says yes is informative. In the Philippine savings trial, women whose survey answers suggested they valued commitment were more likely to open the account. Signing a contract that forfeits money is itself a forecast that one's later self will need it. A reward programme asks for no such forecast, which may be why 90 per cent accepted it.
Our reading is that low take-up is part of how precommitment works, because it reaches the people who already expect to slip, and part of why its average effect in a population is modest. The reward programme asked participants to stake nothing, while in Eyal's account of the deposit programme they staked $150, so the two arms differed in what they demanded as well as in how they paid.
Opt-out enrolment is the usual answer to low take-up, and the evidence on it is in [Why You Never Change the Setting](https://thewhyweproject.com/articles/why-you-never-change-the-setting). Eyal's own list of pitfalls offers a different reason: he warns that price pacts punish people who are already struggling, so a person with a poor record may reasonably decline.
What Survives of the Rule
The part of Eyal's claim that survives is the general one. Precommitment, meaning removing or pricing an option in advance, changed behaviour in a scanner, in a Philippine bank and in an American incentive trial, for the people who adopted it. The 52 per cent does not survive as a statement about a programme, and the ordering rule, that pacts belong last, was not examined by any study cited here.
For anyone building a service with a lock in it, the figure to ask for is the share of users who switch it on. The surprise tests in the nicotine contract study, run a year after enrolment, bear most directly on whether binding alters behaviour or only delays it. The three-point gap between the group with the contract and the group without was still there when participants had not been told a test was coming, and that persistence is the strongest evidence in these trials that a lock leaves a lasting mark on what people do.
Common questions
Why do we pay to limit our own choices?
People expect their later selves to give way, and a binding offer removes or prices the temptation before it arrives. Field trials in the Philippines and the United States found that people who take such offers do better, but most people offered them decline, so the average effect across everyone is smaller than the takers' results suggest.
What is precommitment?
Precommitment is a decision made in advance that removes or prices a future choice, so that the later self has fewer options. Examples include a savings account with restricted withdrawals, a deposit forfeited on failure, or a computer with its network port sawn off.
Whom does Eyal's 52 per cent figure describe?
It is a rate among people who accepted a deposit programme, not among everyone assigned to it. Across everyone assigned in the Halpern trial, the deposit programme produced 10.2 per cent abstinence at six months, against 15.7 per cent for rewards and 6.0 per cent for usual care.
Sources
Eyal, N., with Li, J. (2019). Indistractable: How to Control Your Attention and Choose Your Life. BenBella Books. Ch. 22 to 24, pp. 155 to 173.
Strotz, R. H. (1955). 'Myopia and inconsistency in dynamic utility maximization.' Review of Economic Studies, 23(3), 165-180. doi:10.2307/2295722
Bryan, G., Karlan, D., & Nelson, S. (2010). 'Commitment devices.' Annual Review of Economics, 2, 671-698. doi:10.1146/annurev.economics.102308.124324
Crockett, M. J., Braams, B. R., Clark, L., Tobler, P. N., Robbins, T. W., & Kalenscher, T. (2013). 'Restricting temptations: neural mechanisms of precommitment.' Neuron, 79(2), 391-401. doi:10.1016/j.neuron.2013.05.028
Ashraf, N., Karlan, D., & Yin, W. (2006). 'Tying Odysseus to the mast: evidence from a commitment savings product in the Philippines.' Quarterly Journal of Economics, 121(2), 635-672. doi:10.1162/qjec.2006.121.2.635
Giné, X., Karlan, D., & Zinman, J. (2010). 'Put your money where your butt is: a commitment contract for smoking cessation.' American Economic Journal: Applied Economics, 2(4), 213-235. doi:10.1257/app.2.4.213
Halpern, S. D., French, B., Small, D. S., et al. (2015). 'Randomized trial of four financial-incentive programs for smoking cessation.' New England Journal of Medicine, 372(22), 2108-2117. doi:10.1056/NEJMoa1414293. The acceptor rates (52.3 and 17.1 per cent) are from a Pharmaceutical Journal report.