The Rational Actor
The foundational model of tax non-compliance was articulated by Michael Allingham and Agnar Sandmo in 1972. Their framework is elegant in its simplicity: a taxpayer deciding whether to declare income fully or partially performs a rational calculation. On one side of the equation sits the certain cost of compliance, the tax owed on declared income. On the other sits the uncertain cost of non-compliance, the probability of detection multiplied by the penalty imposed if caught. When the expected cost of non-compliance falls below the cost of compliance, the rational taxpayer chooses non-compliance.
This model has shaped tax policy for over fifty years. It implies two levers for improving compliance: increase the probability of detection through audits and information systems, or increase the severity of penalties for those caught. Most reform programmes in revenue administration have operated, explicitly or implicitly, within this framework.
It is also, on its own, insufficient to explain observed behaviour. If taxpayers were the rational calculators that Allingham and Sandmo described, compliance rates in most countries would be substantially lower than they actually are. The probability of audit in many jurisdictions is vanishingly small, often well below one percent. A purely rational analysis in many tax environments would strongly favour non-compliance. Yet millions of taxpayers in those same environments comply. The model captures something real, but it does not capture everything.
What Behavioural Science Reveals
The work of Daniel Kahneman, Amos Tversky, Richard Thaler, and their intellectual successors has fundamentally reshaped our understanding of how human beings make decisions under uncertainty. The implications for tax compliance are profound, though they have been incorporated into revenue administration practice far more slowly than into other domains of public policy.
Loss aversion, the empirical finding that people experience losses roughly twice as intensely as equivalent gains, means that the framing of compliance costs matters enormously. A taxpayer who perceives compliance as a loss will resist it more strongly than a taxpayer who perceives non-compliance as forgoing a gain.
Status quo bias, the tendency to maintain current behaviour unless compelled to change, creates a powerful inertia favouring whatever pattern a taxpayer has established. If a business has operated informally for years, the activation energy required to enter the formal system is far greater than the ongoing cost of staying informal.
Social proof, the tendency to align behaviour with perceived norms, means that compliance environments are self-reinforcing in both directions. In a community where most businesses comply, the social cost of non-compliance is high. In a community where most businesses are non-compliant, the social cost of compliance can feel prohibitive.
Temporal discounting, the tendency to weight present costs and benefits far more heavily than future ones, systematically undermines the deterrent effect of penalties that are imposed months or years after the act of non-compliance. A fine levied eighteen months after a filing deadline carries, in the taxpayer's psychological accounting, a fraction of its nominal weight.
The Friction Asymmetry
Perhaps the most consequential insight from behavioural economics for tax administration is the friction asymmetry between compliance and non-compliance.
In traditional tax administration systems, compliance is the high-friction path. The compliant taxpayer must navigate forms, often complex and ambiguous. They must queue, whether physically or digitally. They must calculate, or pay someone to calculate on their behalf. They must make payments, maintain records, and respond to queries. Each of these steps represents friction that the taxpayer must actively bear.
Non-compliance, by contrast, is frequently the low-friction path. The non-compliant taxpayer does nothing. They do not file. They do not queue. They do not calculate. They do not pay. In the absence of enforcement, the default state is non-compliance, and the default state requires zero effort.
Every improvement to the filing process represents an effort to reduce the friction of compliance. But they are fighting against a fundamental asymmetry: no matter how simple you make compliance, doing nothing will always be simpler. The question is not whether compliance can be made easier. The question is whether non-compliance can be made harder.
Architecture as Intervention
Enforcement architecture addresses the friction asymmetry directly. Rather than attempting solely to reduce the friction of compliance, it dramatically increases the friction of non-compliance. It does so not through episodic interventions but through continuous, structural, unavoidable consequences embedded in the taxpayer's daily interactions with the state.
When a business seeks to renew an operating licence and its compliance status determines the speed and cost of that renewal, the friction of non-compliance is no longer zero. When a property owner seeks to register a transaction and encounters a compliance checkpoint, the convenience of non-compliance evaporates. When a government contractor discovers that procurement eligibility requires current compliance, the cost of non-compliance becomes a direct commercial concern.
This approach works with behavioural tendencies rather than against them. Loss aversion is leveraged: the taxpayer faces tangible, immediate losses from non-compliance. Status quo bias is redirected: once compliance is established as the condition for normal service, maintaining compliance becomes the path of least resistance. Temporal discounting is neutralised: consequences are not deferred to a future audit but experienced at the next government interaction.
From Theory to Infrastructure
The behavioural economics literature offers a rich catalogue of individual interventions that have been shown to improve compliance in controlled settings. Reminder letters referencing social norms, simplified filing processes, pre-populated returns, and reframed penalty notices have all demonstrated measurable effects. These contributions are valuable and worth implementing wherever feasible.
But there is a categorical difference between episodic nudges and permanent structural architecture. A reminder letter improves compliance among recipients for a filing period. An enforcement architecture that connects compliance status to service delivery across every government interaction changes the structural incentives permanently. The former is a campaign. The latter is infrastructure.
The transition from behavioural insight to institutional infrastructure is the central challenge of modern revenue administration. The science is clear: human beings respond to immediate, visible, and certain consequences far more reliably than to deferred, invisible, and probabilistic ones. Compliance must become the path of least resistance, not through exhortation or education, but through architecture that makes any alternative path structurally more costly.
The behavioural evidence has been available for decades. What has been missing is the infrastructure to act on it at scale. Building that infrastructure is not a matter of better nudges. It is a matter of better architecture.