397. How to Save $32 Million in One Hour

Episode Summary

Executive Summary: A live Freakonomics Radio episode from London examines the rise, limits, and future of behavioral science “nudges” in government, policing, firefighting, and healthcare. Speakers argue that small, evidence-tested changes can produce large gains, but most interventions fail and must be measured honestly. The episode emphasizes replication, defaults, and real-world experimentation as tools for better policy.

Main Topics: The rise of the UK Nudge Unit and evidence-based policymaking (Priority: 5/5): Stephen Dubner frames the Behavioural Insights Team as a quiet revolution inside government: using behavioral science and experiments rather than ideology or habit to shape policy. Successes and failures of nudges in public policy (Priority: 5/5): David Halpern explains that many interventions fail, but a few produce billion-dollar-scale effects, including pensions auto-enrollment, tax letters, and e-cigarette policy. Behavioral science in high-stakes public services (Priority: 4/5): Sabrina Cohen-Hatton shows how decision controls, simulation, and debriefing improve firefighter judgment under pressure and reduce human-error risks. Evidence-based policing and crime prevention (Priority: 4/5): Alex Murray describes randomized trials in policing, including target hardening for burglary prevention, redesigned speeding letters, and jail-cell graffiti experiments. Nudging in healthcare systems (Priority: 5/5): Mithesh Patel explains how a hospital-based nudge unit changed defaults and workflows to increase generic prescribing, opioid stewardship, cardiac rehab referrals, and reduce low-value imaging. The future: trust, cohesion, and bigger social problems (Priority: 3/5): Halpern argues nudging could be applied to social trust, labor market transparency, immigration enforcement, and even conflict prevention, though those challenges are more politically difficult.

Key Arguments: Behavioral policy works best when it is tested empirically; many popular ideas fail and should be abandoned rather than scaled. Small design changes can have outsized effects when they alter defaults, friction, or salience, such as generic prescribing and tax reminders. Most interventions do not work, so publishing failures is as important as celebrating successes to avoid repeating mistakes. In firefighting, structured decision controls improve situational awareness without slowing action, helping commanders make better choices under pressure. Policing should combine professional experience with randomized controlled trials to identify what actually reduces crime. Healthcare systems can save money and improve outcomes by embedding behavioral science into electronic health records and clinical workflows. Behavioral science may eventually be useful for broader structural issues like trust, mobility, and cohesion, not just narrow compliance problems.

Data Points: Behavioral Insights Team sunset clause: 2 years - Halpern says the unit was created with a two-year sunset clause if it failed. Estimated successful interventions: About half a dozen - Halpern says only a small number of BIT projects have billion-dollar-plus impact. E-cigarette quit impact: Circa 40,000 extra smokers a year quit - Halpern cites Public Health England estimates for the effect of e-cigarettes. Tax reminder effect: 9 out of 10 people pay their tax on time - Adding this line to letters improved payment rates. Fire service accidents caused by human error: 80% - Cohen-Hatton says most accidents across industries are due to human error. Decision latency: No increase - Using decision controls did not slow commanders’ decision making. Overall incidents attended by fire services: Down about 50% in the last 10 years - Firefighters now do more prevention and face fewer total incidents. Police cell graffiti experiment result: No effect whatsoever - Murray says growth-mindset graffiti in cells did not change outcomes. Generic prescribing improvement: 75% to 98% - Patel describes a default change that rapidly increased generic prescribing. Cost savings from generic prescribing change: $32 million over two years - The prescribing default change produced major savings. Cardiac rehab referral rate before intervention: 15% - Most eligible heart-attack patients were not being referred. Cardiac rehab referral rate after intervention: 85% - Automation and workflow redesign dramatically improved referrals. Cardiac rehab attendance rate: 33% to 55% - Referral and workflow changes increased participation. Opioid prescribing default: 30 pills to 10 pills - Changing the default reduced unnecessary opioid prescribing. Unnecessary opioid prescribing: Cut in half - After the default change, prescribing dropped substantially. End-of-life imaging rate: 70% - Many palliative cancer patients were getting daily imaging near the end of life. Crowdsourced clinician ideas: 225 ideas in two weeks - Patel says clinician engagement helped launch the hospital nudge unit. Burglary prevention study duration: 700 days - Police tested repeat-victimization effects over a long period.

Pivotal Quotes: "Most things don't work." — David Halpern: Halpern explains why the nudge field must be honest about null results and avoid overclaiming. "80% of accidents, across all industries, by the way, not just fire, are caused by human error." — Sabrina Cohen-Hatton: She argues behavioral science is highly relevant to firefighting and other safety-critical work. "We work with a lot of governments, a lot of departments across the world. Everybody wants to celebrate and talk about the examples that work." — David Halpern: Halpern emphasizes the importance of sharing failures as well as successes.

Implications: The episode suggests behaviorally informed design can improve public systems cheaply and at scale, but only if institutions embrace experiments, defaults, and transparency about what fails. It also points to a bigger future for nudging in trust, safety, and healthcare.