Policy Boomerang

A hypothesis that a corrective policy is changing behavior in ways that later defeat the correction.

Six months ago your company required director approval for any expense over 100 dollars, to stop waste. Spending dipped, then came back, and now you notice something odd in the ledgers: dozens of 95-dollar purchases, split invoices, and personal cards awaiting reimbursement in bulk. The control might now be generating the very opacity it was built to eliminate.

Used as a diagnostic, Policy Boomerang is a hypothesis card: a corrective policy is changing behavior in ways that later defeat the correction. You run it by extracting predictions. If the boomerang is real, you should see adaptation clustered right at the policy's edges: purchases bunched just under thresholds, workarounds that trade visible waste for invisible waste, and the original problem stable or worse when measured a way the policy cannot see. Name the strongest rival: perhaps spending genuinely fell and the sub-threshold cluster is coincidence or seasonality. Then pick the distinguishing observation, such as auditing total real spend including reimbursements against the pre-policy baseline. If the hypothesis survives, the intervention is redesigning the rule around the adaptation, not enforcing it harder. What changes is that you debug the rule instead of blaming the rule-followers. Resist confirming it from a single anecdote; one gamed invoice does not make a boomerang.