Friction is policy.

Across twelve countries and four policy domains, one finding recurs with unusual consistency: when you reduce the steps required to comply with a government process, compliance goes up — sometimes dramatically, often more than any communication or incentive treatment.

Indonesia's Directorate General of Taxes found that a single billing code added to a penalty letter raised timely settlement by 32%. That code let taxpayers pay in one step rather than navigate the full payment portal. A standard deterrence letter — threatening consequences — raised settlement by 27%. The simplification letter outperformed the threat.

The Chicago library fine waiver required only a phone call to complete. In the period it was available, waiver rates doubled. After the program ended, late fees returned to baseline.

In the UK, the government switched the default for workplace pension enrollment from opt-in to opt-out. Participation jumped from 55% to 83% within two years — not because the pension terms changed, not because workers received new information, but because the path of least resistance now led toward enrollment rather than away from it.

In each case, the intervention was not a nudge. It was the removal of an obstacle.

The uncomfortable question

If administrative friction reduces take-up of a program, who chose that friction?

In most cases, the answer is: no one chose it deliberately. It accumulated. Forms were designed by lawyers optimizing for legal completeness. Portals were built by contractors optimizing for functional specification. Deadlines were inherited from predecessor agencies that no longer exist. Every piece of friction had a moment of origin — a committee decision, a vendor's default setting, a form field that seemed harmless at the time.

Administrative burden is usually not a policy decision. It is a policy default — and like all defaults, it can be changed.

What this means for practitioners

The practical implication is that simplification experiments are unusually cost-effective to run, because the baseline to beat is often very low. The comparison is not "our intervention vs. a well-designed alternative." It is "our intervention vs. a process nobody ever tried to optimize."

Three heuristics from the evidence:

**Count the steps.** The strongest predictor of drop-off in any enrollment or compliance process is the number of required actions. Every additional step loses a percentage of the eligible population. You don't need to run an experiment to know that a 14-step enrollment form is leaving people out — you know it from first principles. Run the experiment to know how many.

**Target the exit points.** Simplification experiments that track where participants abandon a process find that drop-off is usually concentrated at one or two specific steps — often a step requiring information the participant needs to retrieve (an account number, a document, a date) rather than provide from memory. Fixing that step, and only that step, often recovers most of the abandonment.

**The optimal intervention is usually "fewer steps," not "better explanation of steps."** Communication interventions that explain a complex process more clearly tend to underperform interventions that shorten the process. Understanding the maze is not the same as removing walls.

A note on equity

Simplification experiments tend to show heterogeneous effects: the populations that benefit most from friction reduction are usually the populations with the fewest resources to absorb that friction. Older adults, people with limited English proficiency, households without reliable internet access, and people working multiple jobs are disproportionately harmed by complex processes. When a simplification intervention raises average take-up by 20%, the gain is rarely distributed evenly.

This makes simplification not just an efficiency intervention but an equity intervention — one that requires no additional targeting and no eligibility criteria.

Next issue: What do we know about the persistence of behavior change from default effects? When does opt-out enrollment produce durable enrollment, and when does it produce grudging enrollment that reverses?