All podcasts / People I (Mostly) Admire / Summary

34. Maya Shankar Is Changing People's Behavior - and Her Own

2026-08-08 - 45 min - source - Read full transcript

Steve Levitt (host)Maya Shankar

Key insights

A well-targeted cold email can open doors institutions would never open through formal channels.

Shankar, a postdoc with no policy experience and no connections, sent a self-deprecating cold email to Cass Sunstein describing herself as not 'cool enough' to work with Obama. He forwarded her to the President's science advisor, and a week later she was pitching a new White House role. The lesson she and Levitt draw is that radical honesty about being underqualified didn't hurt her odds.

building-institutional-change

Institutional change lands only when it's framed as helping people hit goals they already have, not as a new mandate imposed from outside.

Shankar found that pitching a new outcome metric or goal got nowhere; success came from mapping her behavioral toolkit onto agencies' existing budget priorities and asking what they were already trying to accomplish. Levitt confirms the pattern from his own consulting work with Kahneman: top-down mandates get quietly outlasted by middle management.

building-institutional-change

Civil servants, not political appointees, are the highest-leverage allies for organizational change.

Shankar deliberately built relationships with career civil servants who understood the on-the-ground realities (veterans, student borrowers, low-income students) and would remain in government for decades, so their buy-in compounded into sustained multi-year collaborations. Levitt independently reached the same conclusion from FDA consulting work: the people closest to the data know the most and are asked the least.

building-institutional-change

Work needs to be institutionalized outside the seat of power to survive a change in administration.

Shankar's boss described watching eight years of Clinton-era work get wiped out overnight when Bush took office. Shankar responded by deliberately housing her team in the nonpartisan General Services Administration rather than the White House itself, which let it survive the Trump transition and keep working on G.I. benefits and the opioid epidemic even after she left.

building-institutional-change

Individual nudges usually produce small effect sizes, but government scale turns modest percentage gains into large absolute numbers.

Shankar concedes Levitt's skepticism about behavioral economics is well-founded outside of default changes - most interventions she ran moved outcomes by single-digit percentages. But a 3% lift applied across a population of millions, or an 80,000-veteran subgroup, is what made the work worth doing at government scale.

behavioral-economics-in-practice

Active-choice prompts are a practical bridge to full default changes when a default change is politically too costly to attempt immediately.

Rather than auto-enrolling 1.3 million service members into retirement savings on day one, Shankar's team inserted a forced yes/no decision into an existing enrollment workflow, producing a 4-point enrollment increase. Years later the government adopted full auto-enrollment for new recruits, suggesting the pilot built the case for the bigger change.

behavioral-economics-in-practice

Nudges remain ethical, in Shankar's framing, because there is no neutral, default-free way to design a choice, and because transparency about the nudge does not reduce its effectiveness.

She argues every policy or form already embeds a default (e.g., a burdensome 16-document application is itself a nudge away from access), so the question isn't whether to nudge but whether the existing nudge serves the stated goal. She cites an opioid-prescription default-quantity change (30 days to 14) that worked even though prescribers could see it was a deliberate design choice.

ethics-of-influence

Advice from successful people is subject to survivorship bias and may not generalize, but Levitt argues that alone doesn't make it worthless.

Answering a listener question, Levitt explains that studying only successful founders (e.g., 80% under-30 men) can invert once failed founders are counted (90% also under-30 men) - the real lesson may be about volume of attempts, not likelihood of success. He proposes soliciting advice from people who feel their potential was unrealized as a complementary data source.

ethics-of-influence

Attaching identity to a single pursuit is a fragile strategy; attaching it to transferable traits (love of visible progress, love of connection) survives loss better.

Shankar lost the ability to play violin professionally at 17 due to injury, which she says taught her early that identity is malleable rather than fixed to one activity. She now attaches her sense of self to traits like the joy of seeing measurable progress - a trait she later applied to learning Mandarin - rather than to any single domain.

identity-and-change

The psychological concept of 'identity foreclosure' - settling into a self-identity too early - helps explain why people struggle to quit things or change course.

Shankar connects this concept to Levitt's anecdote about over-identifying with his son's fencing career: when the singular pursuit ends (injury, in both their cases), the loss is compounded because the identity, not just the activity, collapses.

identity-and-change

Levitt's working theory of happiness is that needing control over outcomes predicts unhappiness, while accepting the limits of control predicts happiness.

He offers this as a loosely evidenced personal theory - based on casual observation and his own multi-decade shift from craving control to accepting its impossibility - and uses Shankar's story (forced loss of the violin, forced loss of the White House job) as a data point that letting go of a fixed identity can lead to durable adaptation.

identity-and-change

Books referenced

Nudge - Richard Thaler and Cass Sunstein - Levitt credits the book as the namesake and intellectual root of the U.K.'s Nudge Unit, and Shankar cold-emailed co-author Cass Sunstein to get her start in government.

Media referenced

A Slight Change of Plans - podcast - Shankar's own podcast about how people navigate life changes, launched in 2020; discussed at length including guests Tommy Caldwell and Michael Lewis.
The Happiness Lab - podcast - Laurie Santos's podcast; Santos was Shankar's undergraduate mentor who pointed her toward behavioral science work in government.

Companies

Obama White House / Social and Behavioral Sciences Team - Shankar built this team from scratch starting at age 26-27 with zero budget or mandate.
General Services Administration - Nonpartisan agency where Shankar deliberately relocated the behavioral science team so it would survive administration changes; it worked on G.I. benefits, wildfire assessments, and the opioid epidemic.
Google - Shankar later became global director of behavioral sciences there.
U.K. Behavioural Insights Team (the 'Nudge Unit') - Cited as the high-profile, well-resourced U.K. counterpart to Shankar's under-resourced U.S. effort; launched by PM David Cameron in 2010.
University of Chicago RISC - Levitt's own center, mentioned as an example of amplifying small behavioral effects by partnering with large firms.
Department of Veterans Affairs - Ran its first-ever A/B test after Shankar's team reframed a benefits email around the endowment effect, producing a 9% increase in access.
Department of Defense - Piloted an active-choice prompt during base transfers that raised retirement-plan enrollment 4 percentage points; the government later auto-enrolled all new recruits.

Techniques and frameworks

Power of defaults - Automatically enrolling low-income kids in free school lunch (rather than requiring opt-in) added 12.5 million kids to the program; framed as the most reliable behavioral lever discussed.
Active choice - Instead of changing a default outright, the DoD pilot inserted a forced yes/no decision about retirement savings into an existing enrollment process, a lower-friction path to eventual default change.
Endowment effect - Reframing a veterans' benefit as something 'earned' rather than something they were 'eligible for' raised uptake 9%.
Randomized controlled trials - Used as both the evaluation method for pilots and, per Shankar, a form of institution-building - getting an agency to run its first RCT ever mattered as much as the specific result.
Identity foreclosure - Psychology concept Shankar cites for why people struggle to quit or pivot: settling into a self-identity early forecloses seeing alternatives.
Survivorship bias - Levitt's own explainer segment: drawing conclusions from only the successful cases (e.g., young male founders) inverts once failures are included in the sample.

Summary

Steve Levitt talks with Maya Shankar, who at 26 talked her way into building the Obama White House's Social and Behavioral Sciences Team from nothing - no budget, no mandate, no policy experience - by sending a self-deprecating cold email to "Nudge" co-author Cass Sunstein. The first half of the conversation is a tactical postmortem on how she actually built institutional buy-in: work with career civil servants rather than political appointees, frame every pilot as solving a problem the agency already owns rather than imposing a new goal, and deliberately house the surviving work in a nonpartisan agency (the General Services Administration) so it doesn't evaporate the moment an administration changes. Levitt, a self-described behavioral economics skeptic, presses her on whether any of it actually works, and Shankar largely agrees with him: effect sizes are usually small, but multiplied across a population the size of the federal government or a major veteran or military cohort, single-digit percentage gains translate into tens of thousands of people gaining access to a benefit.

Concrete examples anchor the discussion: automatic default enrollment added 12.5 million kids to the school lunch program; an active-choice prompt inserted into a routine enrollment workflow raised military retirement-savings enrollment by 4 percentage points and eventually led to full auto-enrollment for new recruits; and reframing a veterans' benefits email around the endowment effect (earned vs. eligible) produced a 9% uptake increase, with the more durable win being that the VA ran its first randomized controlled trial ever. Shankar also defends the ethics of nudging: since no policy design is ever neutral, and since transparency about a nudge's existence doesn't blunt its effectiveness (illustrated by a default-quantity change in opioid prescriptions), she argues the real ethical question is whether an existing, unexamined default is serving or undermining the program's own goals.

Levitt breaks for a listener-mailbag segment on survivorship bias, using it to interrogate his own show's premise: is it useful to keep asking successful guests for advice when their paths might not generalize, or might even invert once failures are counted in the sample? He floats soliciting advice from people who feel their own potential went unrealized as a counterweight.

The second half turns personal. Shankar was a Juilliard-track violin prodigy who studied with Itzhak Perlman until a hand injury ended that path at 17; she connects that early, forced loss of identity to the concept of "identity foreclosure" and argues it left her more comfortable attaching her sense of self to transferable traits (loving visible progress, loving connection) rather than to any single pursuit - a frame she now explores as host of her own podcast, A Slight Change of Plans. Levitt draws a parallel to over-identifying with his son's fencing career before an injury ended it, and closes with his own loosely-evidenced theory that people who most need control over outcomes tend to be the least happy, while those who accept the limits of control - a shift Shankar's career and Levitt's own life both illustrate - tend to be happier.

Notable Quotes

"I was one person trying to get the entire federal government and Obama leadership bought into this mission... when you get 80 no's, you're not sure actually that the 81st time you're going to get a yes." - Maya Shankar

"There's this concept in psychology called identity foreclosure. It basically refers to the idea that we can settle into a self identity early on and close ourselves off to other alternatives." - Maya Shankar

"The single best predictor of who will be most unhappy in life versus those who will be happiest is that the unhappy people need to feel control over things. Whereas the happy people accept the fact that much of life is beyond their control." - Steve Levitt

"There is no default-less state in the world. So every program and policy has a default design that will influence people one way or the other." - Maya Shankar