The Inspector General Is the Only Auditor

Public accountability is imagined as external. A legislature investigates, a court rules, a newspaper exposes, and the institution is corrected from outside.

Those mechanisms are real and they are rare. The overwhelming majority of examination happens inside, conducted by offices that sit within the bodies they audit.

The statutory framework sets up those offices with a peculiar dual reporting line: to the head of the agency, and to the legislature, simultaneously.

The duality is the whole design. An auditor reporting only upward is a management function; one reporting only outward would never get the access that makes the work possible.

The budget mechanism deserves attention because it is the quietest lever. An audit office whose funding sits inside the agency's own request is asking its subject to pay for its scrutiny.

Access is the asset. An internal auditor can read the file, interview the official and see the system, which no external investigator can do without a subpoena and a year.

Nobody needs to threaten anything for that to shape behaviour. An office that has to justify its headcount annually to the people it examines will find its priorities adjusting without any instruction.

The cost of that access is proximity. The auditor's staff, budget and office space come from the body being audited, and those are the levers that shape what gets examined.

The staffing consequence follows. Auditors are recruited from the agencies they examine, because that is where the necessary knowledge sits, and they frequently return to them afterwards.

The published reports are the output, and reading a run of them tells you as much by what is absent as by what is found.

That circulation is what makes the work competent and it is also what makes a career-ending finding an unlikely thing for any individual to write.

Findings cluster around procurement irregularities, improper payments and IT security, which are real problems and also the problems least likely to implicate a policy choice.

None of this describes dishonesty. It describes an arrangement in which the honest thing and the safe thing point the same way in almost every case, and diverge only rarely.

What appears far less often is an examination of whether a programme achieves anything, because that question cannot be answered without evaluating the decision to run it.

The distinction is between audit and evaluation. Audit asks whether the money went where it was supposed to. Evaluation asks whether sending it there accomplished the purpose.

The second question is harder, more contestable and far more useful, and the incentives of an internal office point steadily away from asking it.

The budget documents reinforce the pattern by presenting programmes in terms of obligation and outlay rather than result, so the available data supports the narrower question.

There is a further structural weakness in what happens after a finding. An auditor recommends; the agency responds; the recommendation is tracked as open or closed.

Closure is determined largely by the agency's own account of what it did, which means the last step of the accountability loop is self-reported by the party being held to account.

The award data would allow independent verification in many cases, and almost nobody performs it, because doing so is nobody's assigned job.

That is the recurring theme. Every individual piece of the system is defensible and the assembled whole leaves the most important question structurally unasked.

The obvious external alternative has its own limits. A legislature's investigative capacity is finite and is spent on whatever is politically live, which is not the same as whatever is most costly.

A press corps has less capacity still and is drawn to the same subjects, so the areas nobody is arguing about receive no scrutiny from any direction at all.

Which is where most of the money is. Large, uncontroversial, long-running programmes attract neither political attention nor journalism, and are examined only by an office inside the building.

The reform that would matter is not more independence for the auditor, which is the usual proposal and which addresses the wrong constraint.

It is a mandate to publish evaluations of outcome on a schedule, whether or not anyone asked, so that the absence of a finding stops being indistinguishable from the absence of a problem.