Problem diagnosis and decision boundaries · Lesson 2

The baseline and no-action scenario

5 min read

Lesson objectives

  • Construct a plausible no-action trajectory and distinguish it from a frozen snapshot and a proposal forecast.
Reading tools

To compare a regulatory proposal with current arrangements, estimate what is likely to happen without the proposed new action. This is the baseline or no-action scenario. It does not mean that nothing changes: demand may grow, a temporary scheme may expire, or a digital service already approved may come online. Crediting a new rule with an improvement that would have happened anyway exaggerates its benefit. Ignoring likely deterioration under present arrangements understates prevention. Define the decision, period, population and activities already committed within actual legal boundaries; then build a plausible trajectory with explicit reasons. The OECD recommendation calls for considering no action and non-regulatory alternatives in impact appraisal. The particular forecast and its evidence remain the analyst's responsibility.

Separate today's snapshot from a no-new-decision trajectory. A snapshot measures one point in time. A trajectory includes changes already approved or supported by evidence, not aspirations. For example, if applications may roughly double following an announced service expansion while staff remain constant, queues could grow without a rule change. Do not claim a precise increase unless capacity modelling or a historical series supports it. Label each assumption as a known commitment, a supported forecast or unresolved uncertainty. Where needed, use lower and upper demand scenarios rather than a single seemingly certain number. Never place the preferred proposal's benefits into the baseline; doing so erases the difference you are trying to estimate.

Test three sources of bias. Seasonality can make a peak month look worse than a quieter comparator. A definition change can start the 'complete application' clock at upload rather than at acceptance after screening, changing reported time without changing an applicant's journey. A parallel action such as new guidance or more staff can coincide with a rule change and blur attribution. Compare like seasons or explain the first effect; recalculate a stable measure for the second; separate interventions or acknowledge a combined estimate for the third. If data are insufficient, gathering evidence and running a bounded trial may be stronger advice than an invented net-benefit figure.

Create a baseline card with four columns: variable, starting value, no-action direction and supporting reason, and decision sensitivity if the assumption changes. In a permit example, monthly demand begins at 100; a hypothetical eligibility expansion might produce 110 to 130 applications after a year; a capacity constraint might then reduce the benefit of form improvements alone. These figures are training assumptions, not an official forecast. Add a comparability test: are all options assessed against the same baseline, over the same period, using the same outcome measure? If not, the supposedly fastest option may look best because it was compared with a worse counterfactual.

Lesson summary

Inputs: 200 applications; current delay rate 30%; no-amendment rate 20% if booking succeeds or 30% if it falters; post-proposal 15% as a hypothesis. Solution: 20%×200=40 delayed applications on the optimistic baseline, 30%×200=60 on the conservative