Discretionary Enforcement and Strategic Compliance

We extend the canonical 2 × 2 inspection game to a 3 × 3 framework that allows strategic non-compliance and an inspection–quality choice, and test it in nine laboratory sessions (164 subjects) in India, Singapore, and Kazakhstan. Static evidence confirms the Harrington paradox: firms comply far more than risk-neutral theory predicts, yet Boards choose the cheap, low-detection audit in over 60% of rounds. Dynamic logit and multinomial specifications reveal four mechanisms. First, strong inertia: current compliance is the strongest predictor of future compliance. Second, inspection quality matters but only briefly. Thorough inspections raise compliance on impact and the effect dissipates within approximately three rounds. Third, cursory audits have no immediate effect; in longer-lag models their positive association with compliance is consistent with behavioral inertia rather than direct deterrence. Fourth, punishment works slowly: individual fines are insignificant on impact, while a history of fines has a modest cumulative influence, indicating graduated deterrence. Moral framing partially counteracts the compliance drop caused by strategic complexity, and country/demographic controls leave core results unchanged. Together, these findings reconcile the Harrington paradox: sticky firm behavior sustains excess compliance, whereas fading inspection salience and the prevalence of cheap cursory audits generate under-enforcement. Policy-wise, improving inspection quality, rather than increasing frequency or fine size, yields the highest marginal deterrence when regulators face capacity constraints and firms can game detection technologies.