Public institutions and oversight

An institution reports what it did. Its records show what it produced. When the two differ, the difference is the finding. DDA reads what public bodies publish as one system, across years and departments, and compares reported outcomes with observable behaviour.

01 The delivery chain

A public commitment moves through a chain before it changes anything. A minister or a board states a priority. A mandate letter or a strategy records it. A budget line funds it. A program or a unit carries it out. Staff apply rules to cases. Results appear in outputs, and outputs appear in annual reports.

Each link can hold the commitment or lose it. A priority with no budget line has no path. A program with a budget and no staff has no action. A staffed program with no decision rights has no effect.

02 The evidence problem

Public bodies publish a large amount: annual reports, service plans, budgets, policy manuals, tribunal decisions, enforcement records, and sector data. Each one is a summary written by the body it describes. The figures in them change definition, period, and denominator from year to year. A rate can fall because the count fell or because the base changed.

Statements and actions need comparison in the same period. A body can describe an issue as under control in a public statement while its own briefing shows the numbers rising. Both items are evidence.

An outcome often needs a decomposition. An enforcement rate can combine cases, files, and charges. A backlog can grow because intake rose or because completions fell. The stock-and-flow identity, change in backlog equals intake minus completions, decides which.

Absence is evidence. A report that is missing from a series, a measure that stops in a year, and a table with no data for a group all tell the analyst something. DDA records the absence and states what it does not prove.

Appeals and complaints show how the system works at its edge. A spike in appeals after a rule change points to a change in decisions or in access.

03 Where it usually breaks

The measure breaks first. An institution reports the quantity it can count. Activity, such as inspections or files opened, is easier to count than outcome, such as harm prevented. Reports then show growth in activity while the outcome stays the same.

The conversion breaks second. Many actions do not lead to a result: warnings with no penalty, files with no decision, funds allocated and never spent. The ratio from one stage to the next shows where the chain fails.

The incentive breaks third. A unit that reports on its own performance has a reason to define success in a way it can meet. Formal objectives do not always explain how a body spends its money.

The feedback breaks fourth. A complaints process that changes nothing gives the institution no signal. Delay between a signal and a response lets a small failure grow.

Resources and design need separation. A body with money and a broken process needs a redesign. A body with a sound process and no money needs funds. Public debate usually assumes the second, and the evidence can show which applies.

04 What the work can produce

The work can produce a claim-and-source register, a chronology of statements, decisions, and actions, a mandate delivery analysis that compares each commitment with budget and output, a conversion analysis by stage, a rules-in-use review, a briefing memo, a publication-ready report with a verification appendix, and a corrections log.

This is research and analysis. It does not give legal advice and does not find that an allegation is true.

This area uses Institutional and policy analysis, Evidence reconstruction, and Fiscal and scenario modelling. Read When an unsupported number carries the answer for a worked example. For plans and public hearings, see Housing and municipal planning.

06 Bring the question

Send the public claim, the mandate, the annual report, or the set of records nobody has assembled. DDA will build the register and show what the public record supports and what it leaves open.

Talk to DDA