A legislative impact assessment is not a verdict on whether a bill is good or bad. It is a structured explanation of what policymakers expect a proposal to change, who may be affected, what it may cost, and how reliable those estimates are. This guide shows you how to read an assessment in plain English, test its assumptions, compare alternatives, and create a repeatable record for checking whether later evidence supports the original case.
Overview
Governments use different names for similar documents, including regulatory impact analysis, regulatory impact statement, explanatory memorandum, fiscal note, business impact assessment, or policy appraisal. The format and legal requirements vary by jurisdiction, but most assessments address a common set of questions:
- What problem is the proposal intended to address?
- What would happen if no action were taken?
- What options were considered?
- What are the expected costs, benefits, risks, and implementation requirements?
- Which people, organisations, sectors, or regions would experience the effects?
- How will the policy be monitored after it takes effect?
Start by identifying the document’s status. An initial assessment may accompany a consultation or bill, while a final assessment may reflect amendments, consultation responses, or a revised preferred option. Do not treat the assessment as the same thing as the bill itself. The bill’s text determines the legal wording; the assessment explains the reasoning and expected consequences behind that wording.
For a reliable legislation summary, read the assessment alongside the latest bill version, explanatory materials, fiscal documents, and any relevant government notices. If the proposal has changed, an older assessment may no longer describe the current text. A useful legislative history can help you identify which estimates belong to which version.
How to estimate
Read the assessment as a chain of estimates rather than as one large conclusion. Follow these steps.
1. Define the baseline
The baseline, sometimes called the “do nothing” or “business as usual” scenario, describes what is expected to happen without the proposal. Check whether it assumes existing rules will remain unchanged, whether other announced measures are included, and what period is covered. A cost or benefit has meaning only when compared with a stated baseline.
2. Separate one-off and recurring effects
Implementation may involve one-off costs such as system changes, training, consultation, or registration. Recurring effects may include administration, compliance activity, staffing, reporting, or savings. Record these separately. Combining them can make a proposal appear cheaper or more expensive than it is over the period being assessed.
3. Identify the affected units
Look for the number of businesses, households, public bodies, workers, transactions, applications, or other units used in the calculation. A simple estimate can often be expressed as:
Total estimated effect = number of affected units × effect per unit × duration
This is not a substitute for the government’s model, but it is a useful reasonableness check. Ask whether the affected-unit count matches the scope of the bill and whether the per-unit estimate includes staff time, technology, legal review, delays, and indirect effects.
4. Check timing and discounting
Assessments may spread effects across several years. Check the start date, transition period, phase-in arrangements, and any assumed commencement date. Some analyses convert future costs and benefits into present values using a discounting method. You do not need to reproduce the entire calculation to evaluate it, but you should record the time horizon and ask whether important effects fall outside it.
5. Compare the preferred option with alternatives
A strong assessment explains why the chosen approach is preferred over alternatives, including non-legislative action, voluntary measures, guidance, enforcement changes, or a narrower rule. Compare options using the same questions: who pays, who benefits, how certain are the effects, and how difficult is implementation? A preferred option may not have the lowest direct cost if it is expected to achieve a different outcome, but that trade-off should be visible.
| Question | Option A | Option B | Preferred option |
|---|---|---|---|
| Problem addressed | What changes? | What changes? | What changes? |
| Direct cost | Who pays and when? | Who pays and when? | Who pays and when? |
| Expected benefit | What outcome is measured? | What outcome is measured? | What outcome is measured? |
| Distributional effect | Who gains or loses? | Who gains or loses? | Who gains or loses? |
| Implementation risk | What could delay delivery? | What could delay delivery? | What could delay delivery? |
Use the table to compare like with like. If one option has detailed monetised estimates and another is described only qualitatively, mark that difference as an evidence limitation rather than assuming the qualitative option has no cost.
Inputs and assumptions
The most important part of a regulatory impact analysis is often not the headline total but the assumptions underneath it. Create a record for each major estimate using five fields: input, source, time period, affected group, and uncertainty.
Inputs may include population size, expected take-up, compliance rates, processing time, wage or operating costs, enforcement activity, market responses, or the number of organisations covered. For every input, ask whether it is observed, surveyed, modelled, or simply assumed. Evidence based on current administrative records may be more directly relevant than a broad forecast, but even observed data can become outdated if the proposal changes behaviour.
Pay close attention to distributional effects. An average result can hide substantial differences between small and large organisations, renters and owners, urban and rural communities, consumers and producers, or different income groups. Record whether the assessment identifies affected groups separately and whether impacts are transferred from one group to another.
Use this evidence-quality checklist:
- Scope: Does the analysis cover the full legal proposal, including delegated legislation and transition rules?
- Baseline: Is the counterfactual clear and plausible?
- Inputs: Are key numbers defined, dated, and attributable to a source or method?
- Uncertainty: Are ranges, scenarios, sensitivity tests, or confidence limits provided where appropriate?
- Alternatives: Are options assessed consistently rather than compared through different standards?
- Distribution: Are effects broken down by relevant groups and locations?
- Implementation: Does the assessment explain capacity, systems, enforcement, guidance, and timing?
- Review: Is there a measurable plan for checking results after commencement?
Classify each important conclusion as high, medium, or low confidence, and write one sentence explaining why. This prevents a precise-looking number from receiving more weight simply because it is presented in a table.
Worked examples
Suppose a proposal would require a defined group of organisations to complete a new annual report. An assessment estimates a one-off setup cost and a recurring annual reporting cost. To review it, record the number of organisations, the expected hours per report, the cost assigned to an hour, the setup period, and the number of years included.
Your working record might look like this:
| Field | Record | Review question |
|---|---|---|
| Affected organisations | Figure used by the assessment | Does it match the bill’s scope? |
| One-off setup | Systems, training, and preparation | Is the cost repeated or genuinely one-time? |
| Annual activity | Hours, submissions, or cases | What evidence supports the volume? |
| Unit cost | Staff, technology, or professional time | Are overheads and internal time included? |
| Uncertainty | Low, medium, or high | What happens if take-up or workload differs? |
| Later outcome | Measure to review after implementation | Where will the result be published? |
For a benefits estimate, apply the same discipline. If the claimed benefit is fewer incidents, faster processing, or improved compliance, identify the baseline rate, the expected change, the affected population, and the method for measuring the result. Do not record a benefit merely as “improved outcomes”; write down the observable indicator that would demonstrate improvement.
Keep the estimate tied to the bill version you reviewed. If you later use a bill tracker to compare amendments, note whether a change affects the affected population, commencement date, reporting duty, enforcement power, or funding assumption. Even a small amendment can invalidate a cost calculation.
When to recalculate
Revisit your analysis whenever the underlying inputs change. The main triggers are an amended bill, a new fiscal note, a final impact assessment, a changed commencement date, a new implementation regulation, or a government response that accepts or rejects a key assumption. Also recalculate when pricing inputs, wage assumptions, benchmark rates, expected volumes, or compliance time estimates move materially.
Set a practical review routine:
- Save the assessment, bill version, and publication date together.
- Highlight every assumption that drives a major cost or benefit.
- Subscribe to relevant government notices, consultation updates, committee papers, and implementation guidance.
- Compare later documents against your original input table rather than starting again from the headline total.
- Record what changed, why it changed, and whether the confidence rating should be revised.
- After commencement, look for monitoring reports, statutory reviews, evaluation findings, or post-implementation assessments.
This approach turns a one-time reading exercise into a living legislative summary. It also helps distinguish a change in the law from a change in the evidence supporting it. For related research, see our guides to comparing bill versions, tracking implementation, and following a public consultation. An impact assessment is most useful when its assumptions remain visible, testable, and connected to the current text of the proposal.