Commitment Device

A commitment device is a self-imposed arrangement whose purpose is to make a future intention easier to carry out. The standard reference is a 2010 review by Gharad Bryan, Dean Karlan and Scott Nelson in the Annual Review of Economics, whose definition is worth quoting because the popular version of it is looser: "Broadly, a commitment device is an arrangement entered into by an individual with the aim of helping fulfill a plan for future behavior that would otherwise be difficult owing to intrapersonal conflict stemming from, for example, a lack of self-control. We exclude actions that accrue significant current benefits or that are taken with a strategic motive." Both exclusions carry weight. Without them the category swallows every prudent arrangement anyone has ever made, and the concept stops predicting anything.

The review's own examples show what the exclusions do. Painting your nails with a foul-tasting anti-nail-biting treatment is a commitment device. Buying nail polish that looks good and happens to taste unpleasant is not, because the current benefit is the point. Setting up an automatic transfer out of checking may be a commitment device, and doing the same thing in order to limit a spouse's spending is not, because the motive is strategic rather than self-directed. Nothing about the mechanics differs between the pairs. The distinction is entirely in what the arrangement is for.

The useful split within the category is by what failure costs. The authors "refer to commitment devices that call for real economic penalties for failure, or rewards for success, hard commitments," and "any device that has primarily psychological consequences a soft commitment," while noting the line "is not perfectly binary." Their examples: a savings account that forfeits interest if a monthly deposit is missed is hard, and the depositor may also feel the psychological cost; an account labeled "send kids to college" is soft, and raiding it for a holiday party carries mostly disappointment plus the minor economic cost of a trip to the bank. A certificate of deposit with an early-withdrawal penalty is hard. A separate account with a name on it is soft. Both change behavior; only one can bill you.

The demand for commitment is what makes this more than a design idea, and the cleanest evidence comes from the SEED accounts studied by Nava Ashraf, Dean Karlan and Wesley Yin. A rural bank in the Philippines offered 700 people an account whose balance became available either at a chosen future date or once a chosen savings goal was reached. The account paid no extra interest for the lost liquidity, so on standard economic terms nobody should have wanted it, and 28 percent took it. Savings held at the bank rose about 80 percent against a control group after a year for those offered the account, and by an estimated 300 percent for the subgroup that actually opened one. The review attaches a caveat that is easy to skip and matters for anyone deciding what to build: this result is open to a soft-commitment reading, because the customers may have valued an account labeled for their goal rather than the lock itself. If the label did the work, the lock was not what they were paying for.

The failure mode has now been measured, and it is the reason this page does not end with a recommendation to lock money up. Anett John's field experiment, published in Management Science in 2020, randomly offered low-income individuals an installment-savings commitment account on which they chose their own savings plan and their own penalty for missing a deposit. Her finding is that "a majority appears to choose a harmful contract: while the average effect on bank savings is large, 55% of clients default and incur monetary losses." The explanation she offers as a possible one is that the penalties people picked were too low to overcome the self-control problem they were trying to solve. That is a specific and unusual shape of result. Someone who knows they need a commitment but misjudges how much force it takes selects one just weak enough to break, and then pays for breaking it. An average effect can be genuinely positive while most of the individuals in it end up worse off.

Two practical consequences follow, and neither requires taking a view on the underlying psychology. The first is to prefer inconvenience to penalties where both would do: the failure of a soft commitment costs a login and a wait, while the failure of a hard one costs money at the exact moment money is short. The second is to size the commitment to the circumstance rather than to the intention, since the evidence says people systematically get that judgment wrong in the direction of too weak. A commitment that binds harder than the plan deserves is its own hazard: circumstances change, and an arrangement designed to be difficult to escape is difficult to escape when escaping is the right answer.

Building one takes four decisions. Name the future behavior precisely enough that success and failure are distinguishable. Choose what failure costs, which is the hard-versus-soft decision. Put the arrangement in place before the moment it has to survive, because a commitment made in the moment is the thing it was meant to protect against. And decide in advance what happens when it fails, since on the evidence it sometimes will.

A hypothetical showing why the second decision is the one that matters. Two savers each intend to set aside $300 a month for a year. The first uses a separate account with a standing transfer that can be canceled in a minute; they cancel it after nine deposits and finish with $2,700 ($300 × 9). The second uses an account that charges $25 whenever a scheduled deposit is missed; they miss four, so they save $2,400 ($300 × 8) and pay $100 (4 × $25), finishing with $2,300. The penalty made the plan harder to abandon and made abandoning it more expensive, and in this version the second saver ends up behind. Whether that trade is worth taking depends on something neither saver can observe in advance, which is how much force their own plan actually needed.

Pros

• It converts a decision that would otherwise be made repeatedly, under worse conditions each time, into one decision made once in the calm.
• Demand for it is measured rather than assumed. People pay real liquidity costs for it, which is evidence that they know something about themselves.
• A soft commitment costs nothing to try, so the cheapest version of the idea carries no downside beyond the effort of setting it up.
• It targets a specific problem, a plan that reverses as its moment arrives, rather than a general shortage of discipline.

Cons

• Penalty-backed commitments harmed the majority of people who chose one in the field experiment that measured it, with 55 percent defaulting and losing money.
• People appear to misjudge how strong their own commitment needs to be, and the error runs toward too weak, which is the worst of both outcomes.
• A commitment sized for one set of circumstances becomes a trap when circumstances change, and money committed is money unavailable for an emergency.
• Part of the measured benefit may come from labeling rather than from locking, in which case paying for the lock is waste.
• None of it addresses a shortfall of income. A commitment device reallocates money across time; it does not create any.