Basically harmless

Cochrane said the administration's experiments so far seem basically harmless: "government spam, messaging people to tell them to do x or y or z." But he said he was troubled by the possibility of more forceful nudges devised by technocrats influenced by lobbyists or other political considerations.

"The case for the free market is not that each individual's choices are perfect," he wrote in a recent blog post. "The case for the free market is long and sorry experience that government bureaucracies are pretty awful at making choices for people."

Civil servants in Britain are being trained in behavioral science, and the governments of Australia, Denmark and South Africa are among those dabbling in nudgery. The British program, spun out of the government, was recently hired by Bloomberg Philanthropies to work with American cities. The World Bank is mulling the lessons for economic development.

Shankar's team asks federal agencies about their priorities, then mines the academic literature for ways to improve those policies. The team also has created a network of academic institutions to provide advice.

Experimentation key

The tweaked rebate forms, for example, were inspired by a 2012 study. Researchers similarly tweaked the form sent to customers of an auto insurance company asking them to report how many miles they had driven. Those asked to sign at the top reported driving about 2,400 more miles.

(The persistence of these effects is an open question. The first time a person signs, they answer the questions more honestly, but by the 10th time, the nudge might be powerless. Shankar said there had been "a little dip" in rebate payments since last summer, but the effect remained quite strong. Thaler noted that alarm clocks wake people up every morning.)

Perhaps even more than new ideas, the behavioral group is bringing a new approach to government. Experimentation is the key: different nudges are tried systematically, results are quantified and, even after the best approach is selected, the team goes back to see how things are working.

The team, for example, created eight versions of an email encouraging members of the military to enroll in a retirement savings program. About 80,000 service members received each version. The most effective, which combined step-by-step instructions with an example of how much a person could save by putting away a little money each month, roughly doubled enrolment rates.

These efforts are informed by common sense as much as academic insight. People are more likely to do things that are easier to do. Yet simplicity has rarely been a priority in the development of federal programs.

"They've been more worried about making sure it's legally perfect than making sure it's understandable to anybody," Thaler said. "So there's a lot of room to simplify things."